Aerotermoquimica

aerothermochemistry 2009/4/2 14:20 page 1 #1
Aerothermochemistry
Gregorio Millan Barbany

Catedratico de Mecanica de Fluidos y Aerodinamica
de la Escuela de Ingenieros Aeronauticos
Miembro de la Real Academia de Ciencias
Reedicion conmemorativa del 50 aniversario de la edicion original
Escuela Tecnica Superior de Ingenieros Aeronauticos

Universidad Politecnica de Madrid
Asociacion de Ingenieros Aeronauticos de Espana
Madrid, 2009
Aerothermochemistry, Gregorio Millan Barbany

Edicion conmemorativa del 50o aniversario de la edicion original de 1958 publicada
por el INTA, abril 2009
Editores cientcos: Manuel Rodrguez Fernandez, Carlos Vazquez Esp

Diseno de la cubierta: Javier Leones Ranz
c 2009 Escuela Tecnica Superior de Ingenieros Aeronauticos

Plaza del Cardenal Cisneros 3, 28040 Madrid, Espana
ISBN 13:978-84-86402-08-2
D.L. GU-114-2009
PRESENTACION
Amable Linan
Es un gran placer para m presentar esta re-edicion literal de la monografa

Aerothermochemistry de Gregorio Millan (1919-2004). La re-edicion por la Escuela
de Ingenieros Aeronauticos se produce cuando acaban de cumplirse 50 anos de la
publicacion por el INTA, en ciclostilo, de los 800 ejemplares de la edicion original.
Cuando Gregorio Millan cursaba sus estudios de Ingeniera Aeronautica, que

inicio en 1941, se estaba produciendo un cambio revolucionario en esta ingeniera. Se
acaba de iniciar el desarrrollo de los aerorreactores, sin los cuales era impensable que
los aviones pudiesen alcanzar velocidades transonicas o supersonicas; simultaneamen-
te, para la viabilidad de los cohetes de sondeo o de los misiles balsticos, se impulso el
desarrollo de los motores cohete. La Mecanica de Fluidos, disciplina central de las
Ciencias Aeronauticas y determinante del diseno de estos motores, fue elegida por
Gregorio Millan como objeto de su actividad docente e investigadora.
En la formacion de Gregorio Millan haba tenido un papel crucial el inusual

ambiente docente de la Academia Militar de Ingenieros Aeronauticos, que se reejaba
en la preocupacion del profesorado por el papel de las ciencias basicas y aplicadas en
el desarrollo de la Ingeniera Aeronautica. Este ambiente docente era heredero del que
ya se daba en la antecesora de la Academia, la Escuela Superior Aerotecnica, desde
su creacion en 1928, bajo la direccion de Emilio Herrera. Para las ensenanzas basicas,
Herrera haba conseguido la colaboracion, que se mantuvo en la Academia Militar, de
los profesores universitarios espanoles mas eminentes (muchos de ellos, igual que el
propio Emilio Herrera, miembros de la Academia de Ciencias).
Uno de estos profesores era Esteban Terradas, que fue Presidente del Patronato
del INTA desde su creacion. Terradas se propuso impulsar el desarrollo en Espana
de las Ciencias Aeronauticas, invitando a los cientcos extranjeros mas prestigiosos
en estas ciencias a impartir ciclos de conferencias. Entre ellos Teodoro von Karman
que en 1948 vino a Espana, por primera vez, para hablar de Aerodinamica Transonica
y Supersonica y sobre Turbulencia. De entonces nacio la colaboracion fructfera de
Gregorio Millan con von Karman, quien oriento la actividad docente e investigadora
posterior de Millan y tambien las investigaciones del Grupo Espanol de Combustion.
Von Karman, que se considera con justicia el padre de las Ciencias Aeronauti-
cas americanas, haba sido el Director de los Guggenheim Aeronautical Laborato-
ries del Instituto Tecnologico de California (Caltech). Poco antes de la ultima Guerra
Mundial inicio su preocupacion por el desarrollo de los cohetes, y despues por los
aerorreactores, siendo el creador del Jet Propulsion Laboratory. Comprendiendo que

el analisis de los procesos de combustion era esencial para el diseno de estos motores y
que deba hacerse con el apoyo de la Dinamica de Fluidos, se embarco en el proyecto
de establecer el marco multidiciplinar apropiado. Para ello, consiguio la colaboracion
del Profesor Saul Pennner del Caltech y de Gregorio Millan, Ingeniero del INTA y
Profesor de la Academia (luego Escuela) de Ingenieros Aeronauticos.
La Aerothermochemistry se basa en el ciclo de conferencias que Teodoro von

Karman impartio en la Sorbona durante el curso 1951-1952, en cuya preparacion y
desarrollo conto con la ayuda de Gregorio Millan. Por el interes suscitado por estas
conferencias de von Karman, el Air Research and Development Command (ARDC)
de las Fuerzas Aereas de Estados Unidos ofrecio a Gregorio Millan, en 1954, un con-
trato para apoyar la redaccion y actualizacion, mediante un programa de investigacion,
de las conferencias de la Sorbona. Para este proyecto, Gregorio Millan conto con la
colaboracion de un grupo de ingenieros y profesores de la Escuela de Ingenieros Ae-
ronauticos, que formaron el Grupo Espanol de Combustion. Este grupo inclua a Se-
gismundo Sanz Aranguez, Jesus Salas Larrazabal, Carlos Sanchez Tarifa, Jose Manuel
Sendagorta e Ignacio Da Riva. Al grupo se sumaron pronto los profesores Francisco
Garca Moreno y Pedro Perez del Notario, y yo mismo que empece como becario en
1958. Mas tarde, crecio notablemente el numero de participantes (entre ellos, Enrique
Fraga, Antonio Crespo, Jose Luis Urrutia y Juan Ramon Sanmartn) en los proyectos
de investigacion del Grupo. Tambien se ampliaron las fuentes de subvencion, que in-
cluyeron el US Forest Service del Departamento de Agricultura americano, as como
la European Space Research Organization (ESRO).
El ARDC facilito la difusion internacional, a traves de laboratorios universi-

tarios y centros de investigacion, de los 800 ejemplares de la edicion original de la
Aerothermochemistry. A pesar de ello, hace ya bastante tiempo que no se encuentran
disponibles ni accesibles copias del original. El valor historico y la actualidad de la Ae-
rotermoqumica de Millan nos ha animado a hacer el esfuerzo de transcribir el original
en TEX y rehacer las guras para una re-edicion. Esta tarea no hubiese sido posible sin
la decision espontanea de Manuel Rodrguez Fernandez de iniciar esa transcripcion,
llegando a completar casi la tercera parte. Este impulso inicial animo a las autoridades
academicas de la Escuela a nalizar la tarea, encargando a Carlos Vazquez Esp la
coordinacion y supervision de la transcripcion y de las correcciones, labor en la que
colaboraron Eva Villacieros y los estudiantes de la ETSI Aeronauticos Alfredo Giral-
da, Ramon Lacruz y David Marchante. En la re-edicion se han eliminado erratas del
original, se han rehecho la mayora de las guras con los metodos actuales de calculo,
que no cambiaron los resultados, y se han incluido algunas anotaciones signicativas.

Por todo el apoyo recibido, tenemos que agradecer a la Universidad Politecnica de
Madrid y a la Asociacion de Ingenieros Aeronauticos la publicacion en forma de libro
de la monografa.
En la Aerothermochemistry aparece por primera vez el marco multidiscipli-

nar coherente, que es necesario para el analisis de los procesos de combustion. Este
analisis, como haba anticipado von Karman, no puede hacerse sin ampliar las leyes
de la Mecanica de Fluidos con las de la Teora de los Fenomenos de Transporte, la
Termodinamica de Mezclas y la Cinetica de las Reacciones Qumicas.1 Aunque la
monografa de Millan contribuyo decisivamente a establecer la necesidad del trata-
miento multidisciplinar en la investigacion de los Procesos de Combustion y de la Ae-
rodinamica Hipersonica, el nombre Aerothermochemistry fue desplazado nalmente
por el de Ciencias de la Combustion.
La importancia de la Aerothermochemistry para la Escuela Espanola de Mecani-

ca de Fluidos ha sido trascendental, porque en ella se abordan problemas de frontera de
la Mecanica de Fluidos con un lenguaje moderno y unas tecnicas avanzadas. Ha sido
ejemplar, tanto para la ensenaza de la Mecanica de Fluidos como para la investigacion
y el uso de la dinamica de los uidos en ambitos multidisciplinares.
Por el carino con el que von Karman acogio y valoro las aportaciones del Gru-
po Espanol de Combustion, von Karman eligio Madrid como sede del Primer Con-
greso Internacional de Ciencias Aeronauticas, que se celebro tambien en 1958. A este
Congreso acudieron las personalidades mas relevantes de las Ciencias Aeronauticas;
entre ellos, Geoffrey Taylor, Leonidas Sedov y James Lighthill. Lighthill, por ejemplo,
hablo de los efectos en ujos hipersonicos de las reacciones qumicas de disociacion
de las moleculas del aire. Von Karman fue tambien esencial en la organizacion en
1960 de un Curso sobre Ciencia y Tecnologa del Espacio, en el INTA, que precedio a
la creacion de la CONIE y a nuestra participacion en la ESRO. Por las aportaciones
de von Karman a las Ciencias Aeronauticas, poco antes de su muerte en Aachen en
1963, recibio de las manos del Presidente Kennedy la primera Medalla Nacional de
Ciencias americana y su imagen fue recogida en uno de sus sellos.
Gregorio Millan agradece en el prologo la valiosa colaboracion de Carlos San-

chez Tarifa, Jose Manuel Sendagorta e Ignacio Da Riva a la Aerothermochemistry.
1 Conviene advertir al lector de la Aerothermochemistry que, por ejemplo, fue en las conferencias de la
Sorbona donde aparecio por primera vez la forma general de la ecuacion de la energa para la dinamica
de uidos reactantes. En ella se incorporan explcitamente los intercambios de energa qumica, termica y
cinetica, el ujo de calor y el trabajo de las fuerzas de presion y de viscosidad; todos ellos involucrados en
el funcionamiento de los motores termicos.
Los tres fueron determinantes en el desarrollo de la Escuela Espanola de Mecanica de

Fluidos. Jose Manuel Sendagorta (responsable de despertar en m la vocacion por la
investigacion) sustituyo a Gregorio Millan en la ensenanza de la Mecanica de Fluidos,
que tuvo que dejar cuando la fundacion y el desarrollo de Sener le absorbieron todo
su tiempo; pero consiguio convertir pronto a Sener en una de las grandes empresas
espanolas de Ingeniera. Primero Sanchez Tarifa y mas tarde Millan, cuando dejo la
direccion de Babcock & Wilcox, fueron llamados por Sendagorta para contribuir al
desarrollo de Sener. Sanchez Tarifa fue el responsable de la ensenanza de Propulsion
en nuestra Escuela, y director, en el INI, del proyecto y ensayo del primer motor de
reaccion espanol; desde Sener dirigio nuestra participacion en el diseno del motor del
Euroghter. La labor investigadora experimental en Combustion de Sanchez Tarifa fue
seminal, inspirando la labor de algunos de los investigadores mas distinguidos proce-
dentes de nuestra Escuela. Ignacio Da Riva, pronto catedratico de Aerodinamica, solo
dejo su fructfera dedicacion, incansable y ejemplar, a la ensenanza e investigacion
cuando la muerte le sobrevino en clase. Fue mi maestro y companero en la investiga-
cion, y puntal esencial en la Escuela Espanola de Mecanica de Fluidos.
Las contribuciones cientcas de Millan, que estan recogidas en parte en este

libro, fueron muy importantes y contribuyeron a su eleccion en 1961 como Miembro
de la Real Academia de Ciencias. Su inuencia en la ensenanza es difcil de imaginar
para las nuevas generaciones, pero es inolvidable para m; que aprend Mecanica de
Fluidos de el y de Jose Manuel Sendagorta, y adopte las notas para clase que nos
dejaron. Yo aprend en la Aerothermochemistry la base teorica de la Combustion; y fui
iniciado en la investigacion sobre las llamas de difusion por Gregorio Millan en las
reuniones semanales que los componentes del Grupo de Combustion tenamos con el
en su etapa (1957-1961) de Director General de Ensenanzas Tecnicas. El lector puede
encontrar en la resena necrologica2 que escrib para la Revista de la Real Academia
de Ciencias, un breve resumen de las aportaciones de Millan a la poltica educativa y
tambien de su labor como gestor a nuestro desarrollo tecnologico.
Espero que esta re-edicion de la Aerothermochemistry sirva de modesto home-

naje a los meritos de las aportaciones de Gregorio Millan y sus colaboradores a las
Ciencias Aeronauticas Espanolas.
2 In Memoriam: D. Gregorio Millan Barbany, Rev. R. Acad. Cienc. Exact. Fis. Nat., Vol 99, No. 1, pp.
183-185, 2005.
INSTITUTO NACIONAL DE TECNICA AERONAUTICA

ESTEBAN TERRADAS
AEROTHERMOCHEMISTRY
(A report based on the course conducted by Prof. Theodore von Karman
at the University of Paris)
BY
GREGORIO MILLAN
PROFESSOR OF AERODYNAMICS
AT
ESCUELA TECNICA SUPERIOR DE INGENIEROS AERONAUTICOS
MADRID (SPAIN)
January, 1958
The research reported in this document has been sponsored by the AIR
RESEARCH AND DEVELOPMENT COMMAND, UNITED STATES AIR
FORCE, under contract No. 61 (514)-441, through the European Ofce, ARDC.
aerothermochemistry 2009/4/2 14:20 page i #9
PREFACE
by
Theodore von Karman
I have great pleasure in introducing the report of Prof. Gregorio Millan on

Aerothermochemistry. This word refers to problems whose solution necessitates
the application of fundamental principles of Thermodynamics and Chemistry espe-
cially chemical kinetics. In other words they are ow problems with exchange of heat
and production of heat by means of chemical reaction. The phenomena of high speed
ight made it necessary for the aeronautical engineer to be fully familiar in addition to
Fluid Mechanics, i.e. Aerodynamics proper, with the fundamentals and applications
of Thermodynamics. This need created the science of Aerothermodynamics. The de-
velopment of jet propulsion introduced problems which have direct bearing in modern
aircraft and missile design and necessitate the understanding of changes in the compo-
sition of the owing medium, as dissociation and chemical reactions. Thus Aerother-
mochemistry deals with the interaction between chemical changes and the merely
aerodynamic phenomena as pressure and velocity distribution, momentum transfer
and alike and between the phenomena of Aerothermodynamics, as shockwaves, heat
transfer, diffusion etc. This novel extension in the scope of the Aeronautical Sciences
necessitates a thorough study of branches of Science, which the aeronautical engineer
in general gave little attention during his professional education. This report is writ-
ten for aeronautical engineers, in that the author does not assume that the reader is
acquainted with the kinetic theory of gases and with chemistry beyond quite general,
ideas.
A few chapters of the report are based on lectures which I gave as Fullbright
guest professor at the Sorbonne in Paris in the scholastic year 1951/52. This course
had the same general aim to acquaint the aeronautical engineer including myself with
the problems introduced by the phenomena of combustion occurring in owing media.
The author however completed, and I believe very successfully, the presentation of the
subject by including a systematic treatment of the Thermodynamics of gas mixtures,
of the theory of chemical equilibrium, the elements of chemical kinetics, theory of
transfer phenomena in gases and gas mixtures. In addition the problems of ame sta-
bilization, combustion of liquid droplets and diffusion ames, similarity which were
only touched on in my lectures are treated in this report systematically.
I am sure that the report will be helpful to many aeronautics engineers in their
aerothermochemistry 2009/4/2 14:20 page ii #10
aim to become acquainted with Aerothermochemistry and it may also contribute some
suggestions to those engaged in pursing the subject in their further researches.
For this reason the report includes rather extended bibliographic material after
each chapter.
I would like to mention without commitment that if I can secure the valuable
collaboration of Prof. Millan, I am playing with the idea to publish a book on the
fundamentals of Aerothermochemistry.
aerothermochemistry 2009/4/2 14:20 page iii #11
FOREWORD
During the academic term 195152, Professor Theodore von Karman con-
ducted a course on Aerodynamics of Combustion at the University of Paris. This
course was mainly intended for aeronautical engineers with the objective of encour-
aging the study of the problems risen by the new propulsion systems developed by
Aeronautics, which necessitates the assistance of the theories and methods of Aero-
dynamics, Thermodynamics and Chemistry of Combustion. The new Science born
to deal with these problems needed a name and Prof. von Karman suggested it be
called AEROTHERMOCHEMISTRY for its analogy with other classical science of
Aeronautics, such as Aeroelasticity or Aerothermodynamics. In a few years this des-
ignation has been widely accepted.
The author was invited by Prof. von Karman to participate as his assistant in
developing this course. Due to the amount of bibliography that has to be reviewed in
addition to the numerous studies and exercises required in its preparation it was not
possible at the time to write the lectures.
Later, when the European Ofce of the ARDC was activated it Brussels and
in view of the general demand for the Lectures by Prof. von Karman, the said Of-
ce offered the author the opportunity of preparing a Report on Aerothermochemistry
based upon the course conducted by Prof. von Karman, through a research contract
sponsored by ARDC.
Work was started in 1953 but for different reasons it had to be interrumped
in several occasions. At the same time investigation was progressing very rapidly in
the different elds of Aerothermochemistry and it was necessary to incorporate them
into this report, in order to keep it current. For this reason, initial planning had to be
continuously extended and nally carried far beyond its original intention.
The author wishes to express his deep appreciation to Prof. von Karman who
was a constant source of inspiration and encouragement and whose advise in preparing
this report proved invaluable. At the same time he trusts that Prof. von Karman will
excuse this report if it fails to reect the excellencies of the course on which it is based.
Special acknowledgment is given to the contribution of the European Ofce

ARDC, and the Instituto Nacional de Tecnica Aeronautica Esteban Terradas for the
help received in preparing this report for publication.
Finally, I wish to acknowledge the valuable collaboration of Carlos Sanchez

aerothermochemistry 2009/4/2 14:20 page iv #12
Tarifa, Manuel de Sendagorta and Ignacio Da Riva, aeronautical engineers of the

INTA, in the research work that made possible the publication of this report, and to
Mrs. de los Casares for her technical translation.
Madrid, January 1958

aerothermochemistry 2009/4/2 14:20 page v #13
Contents
1 Thermochemistry 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Thermodynamic functions of an ideal gas. . . . . . . . . . . . . . . 6
1.3 Mixture of gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Calculation of the thermodynamic functions . . . . . . . . . . . . . 16
1.5 Chemical reactions in a mixture of gases . . . . . . . . . . . . . . . 21
1.6 Chemical equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.7 Case of a mixture of ideal gases . . . . . . . . . . . . . . . . . . . . 24
1.8 Chemical kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2 Transport phenomena in gas mixtures 37

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2 Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3 Viscosities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.4 Thermal conductivity . . . . . . . . . . . . . . . . . . . . . . . . . 54
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3 General equations 59
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.2 Equation of continuity . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.3 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . 64
v
aerothermochemistry 2009/4/2 14:20 page vi #14
3.4 Energy equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.5 General equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.6 Entropy variation . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.7 One-dimensional motions . . . . . . . . . . . . . . . . . . . . . . . 72
3.8 Stationary, one-dimensional motions . . . . . . . . . . . . . . . . . 75
3.9 The case of only two chemical species . . . . . . . . . . . . . . . . 77
3.10 Stationary, one-dimensional motion of ideal gases with heat addition 79
3.11 Appendix: Notation and repertoire of vectorial and tensorial formulae 84
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4 Combustion Waves 89
4.1 Detonation and deagration . . . . . . . . . . . . . . . . . . . . . . 89
4.2 Kinds of detonations and deagrations . . . . . . . . . . . . . . . . 91
4.3 Velocity of the burnt gases . . . . . . . . . . . . . . . . . . . . . . . 94
4.4 Propagation velocity . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.6 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.7 Indeterminacy of the solution . . . . . . . . . . . . . . . . . . . . . 103
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5 Structure of the combustion waves 105
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.2 Wave equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.3 Characteristic times . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.4 Limiting form of the wave equations . . . . . . . . . . . . . . . . . 111
5.5 Detonations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.6 Deagrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.7 Transition from deagration to detonation . . . . . . . . . . . . . . 126
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
aerothermochemistry 2009/4/2 14:20 page vii #15
6 Laminar ames 131
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.2 Equation for the combustion wave (two chemical species) . . . . . . 134
6.3 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.4 Modication of the conditions at the cold boundary . . . . . . . . 137
6.5 Propagation velocity of the ame . . . . . . . . . . . . . . . . . . . 140
6.6 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.7 Reaction velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.8 Flame equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.9 Solution of the ame equations . . . . . . . . . . . . . . . . . . . . 148
6.10 Structure of the combustion wave . . . . . . . . . . . . . . . . . . . 159
6.11 Ignition temperature . . . . . . . . . . . . . . . . . . . . . . . . . . 161
6.12 General equations for the combustion wave . . . . . . . . . . . . . . 162
6.13 Ozone decomposition ame . . . . . . . . . . . . . . . . . . . . . . 169
6.14 Hydrazine decomposition ame . . . . . . . . . . . . . . . . . . . . 176
6.15 Flame propagation in Hydrogen-Bromine mixtures . . . . . . . . . . 187
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
7 Turbulent ames 207
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
7.2 Turbulent combustion theories . . . . . . . . . . . . . . . . . . . . . 210
7.3 Turbulence generated by the ame . . . . . . . . . . . . . . . . . . 218
7.4 Comparison with experimental results . . . . . . . . . . . . . . . . 219
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
8 Ignition, ammability and quenching 221
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
8.2 Ignition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
8.3 Flammability limits . . . . . . . . . . . . . . . . . . . . . . . . . . 225

aerothermochemistry 2009/4/2 14:20 page viii #16
8.4 Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
9 Flows with combustion waves 231

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
9.2 Conditions that must be satised by the jump across a ame front. . . 231
9.3 Normal ame front . . . . . . . . . . . . . . . . . . . . . . . . . . 235
9.4 Inclined ame front . . . . . . . . . . . . . . . . . . . . . . . . . . 236
9.5 Entropy jump across the ame front . . . . . . . . . . . . . . . . . . 239
9.6 Vorticity across the ame . . . . . . . . . . . . . . . . . . . . . . . 242
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
10 Aerothermodynamic eld of a stabilized ame 245

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
10.2 Tsien method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
10.3 Method of Fabri-Siestrunck-Foure . . . . . . . . . . . . . . . . . . 255
10.4 Cylindrical chambers . . . . . . . . . . . . . . . . . . . . . . . . . 258
10.5 Chamber with slowly varying cross-section . . . . . . . . . . . . . . 259
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
11 Similarity in combustion. Applications 261

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
11.2 Dimensionless parameters of Aerothermochemistry . . . . . . . . . 262
11.3 Scaling of rockets . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
11.4 Scaling of rockets for non-steady conditions . . . . . . . . . . . . . 270
11.5 Flame stabilization . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
12 Diffusion ame 281

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
12.2 General equations for laminar diffusion ames . . . . . . . . . . . . 289
aerothermochemistry 2009/4/2 14:20 page ix #17
12.3 Boundary conditions on the ame . . . . . . . . . . . . . . . . . . . 291

12.4 Simplied equations . . . . . . . . . . . . . . . . . . . . . . . . . . 292
12.5 Solutions of the simplied system . . . . . . . . . . . . . . . . . . . 296
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
13 Combustion of liquid fuels 301

13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
13.2 Atomization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
13.3 Mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
13.4 Combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
13.5 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
13.6 Continuity equations . . . . . . . . . . . . . . . . . . . . . . . . . . 308
13.7 Energy equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
13.8 Diffusion equations . . . . . . . . . . . . . . . . . . . . . . . . . . 313
13.9 Combustion velocity of the droplet . . . . . . . . . . . . . . . . . . 314
13.10 Integration of the equations . . . . . . . . . . . . . . . . . . . . . . 316
13.11 Numerical application . . . . . . . . . . . . . . . . . . . . . . . . . 319
13.12 Comparison with experimental results and limitations of the theory . 322
13.13 Inuence of convection . . . . . . . . . . . . . . . . . . . . . . . . 324
13.14 Combustion of fuel sprays . . . . . . . . . . . . . . . . . . . . . . . 325
13.15 Droplet evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . 325
13.16 Appendix: Application of Proberts method . . . . . . . . . . . . . . 331
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
aerothermochemistry 2009/4/2 14:20 page x #18
Chapter 1
Thermochemistry
1.1 Introduction
Aerothermochemistry deals with mixtures of reactant gases of variable composition.

Such variation in composition is due to the reactions between the various chemi-
cal species forming the mixture and to their mutual diffusion. In general, the pres-
sure of these mixtures is not larger than several atmospheres1 and their temperatures
range from ambient temperature to 3 000 or 4 000 K. Under such conditions the mean
distance between molecules is large respect to their size, and their potential energy
of interaction is negligible compared to the kinetic energy of the molecular motion.
Thereby, the mixture can generally be treated as an ideal gas. In those cases where it
is necessary to take into consideration the deviations in the behavior of the gas with
respect to that of an ideal gas, this can be attained by including correcting terms in the
thermodynamic equations of mixture. For example, its state equation can be approxi-
mated by the virial equation proposed by Kamerlingh Onnes
p
= 1 + B (T ) + 2 C (T ) + . . . (1.1)
Rg T
where p, and T are, respectively, the pressure, density and absolute temperature of
the gas. Rg is its particular constant and B (T ), C (T ) , . . . are the second, third, etc.
coefcients of the virial, respectively. These can be calculated from the potential
of interaction of the molecules. For example, if both molecules belong to the same
species and if the potential depends only on the distance r between their centers of
gravity but not on their relative orientation, the second coefcient, which generally is
1 Some ballistic problems are excluded (See Ref. [1]) as well as detonations of condensed explosives
(See Ref. [2]) where pressures can be very large.
1
2 CHAPTER 1. THERMOCHEMISTRY
3.0
2.5
2.0
1.5
/
1.0
0.5
0.0
0.5
1.0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

r/
Figure 1.1: Dimensionless Interaction Potential of Lennard-Jones, /, versus dimension-
less distance between molecules, r/.
the only needed except for very high pressures, is expressed in the form2
( )
N (r)
B (T ) = 2 1 exp( ) r2 dr, (1.2)
M 0 kT
where N = 6.0288 1023 mol1 is the Avogadro number, M is the molar mass
of the gas and k = 1.38047 1016 erg/grad is the Boltzmann constant. It has
experimentally been checked that the interaction potential of Lennard-Jones
(( ) )
12 ( )6
(r) = 4 , (1.3)
r r
represents, suitably, the actual behavior of many gases. In Eq. (1.3), is the radius
of the molecules and is a constant which has energy dimensions. Fig. 1.1 represents
Eq. (1.3). There, it is seen that if the distance r between the molecules is larger than
several times the radius of the molecule, the molecules attract each other whilst if
r < 1.12 they reject with a force that increases very rapidly as the molecules get
closer. By substituting (1.3) into (1.2) one veries that B (T ) can be expressed in the
form
b0
B (T ) = B (T ) , (1.4)
M
where
2
b0 = N 3 (1.5)
3
2 See Ref. [3].
1.1. INTRODUCTION 3
GAS /k (K) (A) b0 (cm3 /mol)

Air 99.2 3.522 55.11
N2 95.05 3.698 63.78
O2 117.5 3.580 57.75
CO 100.2 3.763 67.22
CO2 189 4.468 113.90
NO 131 3.170 40.00
N2 O 189 4.590 122.00
CH4 148.2 3.817 70.16
CH-CH 185 4.221 94.86
CH2 =CH2 199.2 4.523 116.70
C2 H6 243 3.954 78.00
C3 H8 242 5.367 226.00
n-C4 H10 297 4.971 155.00
i-C4 H10 313 5.341 192.18
n-C5 H12 345 5.769 242.19
n-C6 H14 413 5.909 260.25
n-C7 H16 282 8.880 884.00
n-C8 H18 320 7.451 521.79
n-C9 H20 240 8.448 760.53
Cyclohexane 324 6.093 285.33
C6 H6 440 5.270 84.62
CH3 OH 507 3.585 58.12
C2 H5 OH 391 4.455 111.53
CH3 Cl 855 4.455 48.49
CH2 Cl2 406 4.759 135.96
CHCl3 327 5.430 201.95
C2 N2 339 4.380 105.99
COS 335 4.130 88.86
CS2 488 4.438 110.26
Table 1.1: Force constants for the Lennard-Jones potential.
and
kT
T = . (1.6)

Table 1.1, taken from Ref. [3], gives the values of b0 and /k for several gases.
These values have been determined from experimentation. Fig. 1.2 gives the values
for B (T ) taken from the same reference.3 To get an idea on the order of magnitude
of coefcients B (T ) and C (T ), and, consequently, on the deviations that are to be
expected in the actual behavior of gases respect to their ideal model, Figs. 1.3 and
1.4 give the values of these coefcients for some gases. As can be observed B (T ) is
rst negative and then becomes positive, tending slowly to zero for T . Such
behavior of B (T ) is common for all gases. It is seen, for example, that in nitrogen
for p = 25 kg/cm2 and T = 800 K, B (T ) = 1 cm3 /gr. Therefore, the error made
3 Ref. [3] also includes a table for the calculation of C (T ).
with the assumption that in such a state nitrogen behaves as an ideal gas is one per
cent. Therefore under the conditions of the present study it is justied to consider the
gases as ideal. Hereinafter, unless the contrary is made explicit, such an assumption
is adopted.
*
T = 10
0.5
*
T = 100
0.0
*
0.5 T=
B*(T*)
1.0
1.5
2.0
2.5
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Figure 1.2: B as a function of T . Note the two changes of scale.
8
H2
O2 C H
7 16
4
N2
CO2
B(T) (cm /gr)
0
3
CH4
4
12
0 400 800 1200 1600 2000 2400 2800 3200 3600
T(K)
Figure 1.3: Virial coefcient B (cm3 /gr) , as function of temperature for several gases.
1.1. INTRODUCTION 5
14
N2
12
10 CH4
C(T) (cm6/gr2)
O2 H2 (0.1 C(T))
4
CO2 C H (0.1 C(T))
7 16
2
CO
0
0 400 800 1200 1600
T(K)
Figure 1.4: Virial coefcient C (cm3 /gr)2 as a function of temperature for several gases.
In the following paragraphs we will rst briey summarize the thermodynamic

functions of ideal gases and of their mixtures. A more extensive study of the subject
can be found, for example in the works mentioned in Refs. [4] and [5], where, the ther-
modynamic functions for the case of air gas are also included. A complete repertoire
of thermodynamic functions of the gases and their mixtures including the inuence of
some virial coefcients, can be found in Ref. [6]. Further on, the form in which these
thermodynamic functions can be obtained is briey considered. Finally, reference is
made to the thermodynamic equilibrium and to the chemical kinetics of gas reactions.
Throughout the following paragraphs the mixture will be assumed to be in thermody-
namic equilibrium,4 at rest and that its state and composition are homogeneous. The
effects of motion and of lack of homogeneity will be studied in chapter 2.
It is generally advantageous in Aerothermochemistry to refer the thermody-

namic functions to the unit mass, which is done in the following instead of referring
then to a mole, as usually done in Thermodynamics. The thermodynamic functions
(internal energy, enthalpy, entropy and free energy) when referred to the unit mass
will be denoted with small letters. If referred to a mole the same letters in capital will
be used.
4 Obviously except in the paragraph devoted to Chemical Kinetics.

1.2 Thermodynamic functions of an ideal gas.
The equation of state
The state equation of an ideal gas is given by the expression

p
= Rg T, (1.7)

where
R
Rg = (1.8)
Mg
is the particular constant of the gas, where R = 8.3144 joule/mol/grad = 1.9872 cal/mol/grad
is the universal constant of the gases.
Internal Energy
The internal energy of the gas is

T
u = u0 + cv (T ) dT. (1.9)
T0
Here u0 is the internal energy of the gas at temperature T0 , cv is the specic heat at
constant volume, which depends on temperature as will be analyzed in 4, and T0 is a
reference temperature. In particular, if cv is constant from T0 to T , we have
u = u0 + cv (T T0 ) , (1.10)
otherwise, in many applications cv is substituted by a mean value cv . In such case

(1.9) takes the approximate form
u = u0 + cv (T T0 ) . (1.11)
Enthalpy
The gas enthalpy h is dened by

p
h=u+ . (1.12)

Making use of the state equation (1.8) and of equation (1.9), one obtains for (1.12)
T
h = h0 + cp (T ) dT, (1.13)
T0
1.2. THERMODYNAMIC FUNCTIONS OF AN IDEAL GAS. 7
where h0 is the enthalpy of the gas at the temperature T0 , and cp = cv +Rg its specic
heat at constant pressure. As before, if cp is constant or is substituted by a mean value
cp , we have
h = h0 + cp (T T0 ) . (1.14)
Since only differences in energy can be measured a convention becomes nec-

essary to determine u0 and h0 . This convention is as follows:
1) A standard state is adopted for each chemical substance, dened by a value p0

of the pressure and a value T0 of the temperature.
2) The formation enthalpies of the chemical elements in their stable phase in the
standard state are zero.
3) If any of these substances is a gas its formation enthalpy for T = T0 and p 0
were equal to that of the actual gas.
For moderate pressures, this enthalpy differs slightly from the enthalpy of the actual
gas at temperature T0 and at pressure p0 . At present the values generally adopted for
p0 and T0 are those proposed by G. M. Lewis [7], namely T0 = 298.16 K, that is
25 C, and p0 = 1 atm. Formerly values of temperature slightly smaller were used.
The reduction from one state to another can be easily performed. The previous con-
vention xes the values of the internal energy and of the enthalpy of any substance as
per the rst principle of Thermodynamics. The value of h0 can be obtained, for exam-
ple, by measuring the heat of reaction at pressure p0 and at temperature T0 , when the
substance forms from its elements or from other substances for which the formation
enthalpies are known. Therefore, in the preceding formulae u0 and h0 are the internal
energy and the formation enthalpy of gas respectively. Between both the following
relation exists
p0
h0 = u0 + . (1.15)
0
Tables 1.2, 1.3 and 1.4 give the formation enthalpies of some substances and
their stable phase in the state of reference. Such values have been taken from Ref. [4],
pp. 98 and following, were additional information can be found. Tables NBS-NACA
of thermal properties of the gases [8] are particularly interesting.
Entropy
The entropy s of the gas is

( ) T
p cp (T )
s = s0 Rg ln + dT, (1.16)
p0 T0 T
Inorganic Substances
Substance State h0 s0 G0
H2 gas 0.00 5.482 0
H gas 51 675.60 27.176 48 575
Br2 liquid 0.00 0.228 0
Br gas 3 342.25 0.543 19 690
HBr gas -107.01 0.586 -12 720
O2 gas 0.00 1.531 0
O gas 3 697.44 2.404 54 994
O3 gas 708.33 1.183 39 060
H2 O gas -3 208.15 2.504 -54 635
H2 O liquid -3 792.02 0.928 -56 690
N2 gas 0.00 1.634 0
N gas 6108.37 2.614 81 476
NH3 gas -648.19 2.701 -3 976
NO gas 719.81 1.678 20 719
NO2 gas 175.86 1.249 12 390
N2 O4 gas 25.09 0.790 23 491
N2 O gas 442.79 1.195 24 760
HNO3 liquid -657.04 0.591 -19 100
C(graphite) solid 0.113 0
C(diamond) solid 37.72 0.049 685
CO gas -943.09 1.689 -32 808
CO2 gas -2 137.06 1.16 -94 259
Table 1.2: Formation enthalpies (cal/gr), entropies (cal gr1 K1 ) and free energies
(cal/mol). Inorganic substances.
where s0 is the entropy of the gas at pressure p0 and at temperature T0 . The third law of
Thermodynamics assigns zero values to the entropies of all substances at the absolute
zero of temperature. This determines the values of s0 without the need of a convention
as for the enthalpy case.5 Tables 1.2-1.4 also include the formation entropies of the
various substances.
Let s0 be the entropy of the gas at temperature T and standard pressure p0 that
is T
cp (T )
s0 = s0 + dT. (1.17)
T0 T
From this equation and Eq. (1.16) one obtains for s
( )
p
s = s0 Rg ln , (1.18)
p0
where s0 depends only on T , once p0 is selected.
5 See, i.e., Ref. [9] for complementary information and for discussion of the anomalies that appear in the
application of the Third Law.

1.2. THERMODYNAMIC FUNCTIONS OF AN IDEAL GAS. 9
Alcohols
Substance Formula State h0 s0 G0
Methanol CH3 OH g -1501.15 1.773 38 700
Methanol CH3 OH l -1780.04 0.946 -39 750
Ethanol C2 H5 OH g -1220.80 1.463 -40 300
Ethanol C2 H5 OH l -1440.39 0.834 -41 770
1-Propanol C3 H7 OH l -1212.43 0.767 -40 900
2-Propanol C3 H7 OH l -1279.00 0.716 -44 000
1-Butanol C4 H9 OH l -1074.07 0.735 -40 400
2-methyl-2-Propanol (CH3 )3 COH l -1206.29 0.611 -47 500
1-Pentanol C5 H11 OH l -976.33 0.691 -39 100
2-methyl-2-Butanol C2 H5 COH(CH3 )2 l -1094.32 0.622 -47 700
Diphenyl Carbinol (C6 H5 )2 CHOH s -110.62 0.311 30 700
Triphenyl Carbinol (C6 H5 )3 COH s 15.21 0.302 69 700
Cydohexanol C6 H11 OH l -85.47 0.476 -34 300
Ethylene Glycol CH2 OHCH2 OH l -1749.37 0.643 -77 120
Glycerol CH2 OHCHOHCH2 OH l -1728.23 0.540 -113 600
Phenol C6 H5 OH s -407.72 0.362 -11 000
Thiophenol C6 H5 SH s 0.477
Pyrocatechol C6 H4 (OH)2 s -795.31 0.326 -51 400
Resorcinol C6 H4 (OH)2 s -795.31 0.321 -53 200
Hydroquinone C6 H4 (OH)2 s -795.31 0.304 -52 700
Aniline C6 H5 NH2 l 78.82 0.492 35 400
(cal/mol). Alcohols.
Free Energy
Gibbs free energy
g = h T s, (1.19)
which can be written in the form

( )
p
g = g 0 + Rg T ln , (1.20)
p0
where
g 0 = h T s0 (1.21)
is the free energy at standard pressure.
The free energy G = M g per mole is also called chemical potential of the
gas ( )
p
= G = G0 + RT ln , (1.22)
p0
where
G0 = H T S 0 = 0 , (1.23)
and H, S 0 and 0 are, respectively, the molar enthalpy and entropy of the gas and its
chemical potential at standard pressure.
Hydrocarbons
Substance Formula State h0 s0 G0
Methane CH4 g -1 115.14 2.774 -12 140
Ethane C2 H6 g -673.01 1.824 -7 860
Propane C3 H8 g -562.89 1.463 -56 14
n-Butane C4 H10 g -512.94 1.275 -3 754
n-Pentane C5 H12 g -485.13 1.154 -1 960
Ethylene C2 H4 g 445.49 1.87 1 6282
Propilene C3 H6 g 115.95 1.516 14 990
1-Butene C4 H8 g 4.99 1.31 17 217
cis-2-Butene - g -24.28 1.282 16 046
trans-2-Butene - g -42.87 1.263 15 315
Isobutene - g -59.59 1.251 14 582
1-Pentene C5 H10 g -71.30 1.185 18 787
Acetylene C2 H2 g 2081.5 1.843 50 000
Methylacetylene C3 H4 g 1106.26 1.48 46 313
Cyclopentane C5 H10 l -360.76 0.696 8 700
Methylcyclopentane C6 H12 l -392.96 0.704 7 530
Cyclohexane C6 H12 l -443.7 0.58 6 370
Methylcyclohexane C7 H14 l -462.92 0.604 4 860
Benzene C6 H6 g 253.75 0.824 30 989
Benzene C6 H6 l 150.02 0.529 29 756
Toluene C7 H8 g 129.7 0.829 29 228
Toluene C7 H8 l 31.12 0.57 27 282
o-Xylene C8 H10 g 42.77 0.794 29 177
o-Xylene C8 H10 l -55.02 0.555 26 370
m-Xylene - g 38.81 0.805 28 405
m-Xylene - g -57.22 0.568 25 730
p-Xylene - g 40.41 0.793 28 952
p-Xylene - l -54.99 0.557 26 310
Durene C6 H2 (CH3 )4 s -242.68 0.437 19 000
Cumene C6 H5 CH(CH3 )2 l -81.94 0.556 20 700
Mesitylene C6 H3 (CH3 )3 l -126.34 0.544 24 830
Diphenyl C6 H5 -C6 H5 s 135.34 0.319 57 400
Diphenylmethane C6 H5 -CH2 -C6 H5 s 117.22 0.340 63 600
Triphenylmethane (C6 H5 )3 s 170.92 0.305 101 400
Naphthalene C10 H8 s 124.53 0.311 45 200
Anthracene C14 H10 s 154.86 0.278 64 800
Phenanthrene C14 H10 s 129.62 0.284 60 000
(cal/mol). Hydrocarbons.
1.3. MIXTURE OF GASES 11
Helmholtz free energy
f = u T s. (1.24)
Like in (1.20), here f can be expressed in the form

( )
0 p
f = f + Rg T ln , (1.25)
p0
where
f 0 = u R g T s0 (1.26)
is the free energy at standard pressure, which depends only on the gas temperature.
Remark
When desired to refer the thermodynamic functions to a mol, the previous formulae
should be changed as follows:
1) cv and cp are the molar heats.

2) Rg must be substituted by the universal constant R.
3) In the state equation (1.7), must be substituted by the number c of moles of the
gas per unit volume.
1.3 Mixture of gases
Now let us see the form taken by the thermodynamic functions in a mixture of gases.
For the purpose we shall consider a mixture of gases formed by ` different chemical
species Ai , (i = 1, 2, . . . , `). Let us assume that its state is such that the mixture
behaves as an ideal gas. The problem deals with the expression of the thermodynamic
functions by means of the corresponding functions of the chemical species of the
mixture.
Concentrations, mole fractions, densities and mass fractions
Let c and ci , (i = 1, 2, . . . , `), be the number of moles of the mixture and of each
species per unit volume respectively, and let Xi be the mole fraction ci /c of species Ai .
Between these magnitudes the following relations exist

cj =c , (1.27)
j

Xj =1 . (1.28)
j
Let be the density of the mixture, i the partial density of species Ai , Yi its
mass fraction and Mi its molar mass. The following relations exist
i = Yi , (1.29)

j = , Yj = 1 , (1.30)
j j

ci = Yi , (1.31)
Mi
Yj
c= , (1.32)
j
Mj
Yi /Mi
Xi = , (1.33)
Yj /Mj
j
M i Xi
Yi = . (1.34)
Mj X j
j
The mean molar mass Mm of the mixture is dened by

Mm = M j Xj . (1.35)
j
From Eqs. (1.30) and (1.35) one deduces for Mm as a function of Yi the
following expression
1 Yj
= . (1.36)
Mm j
Mj
Equation of state of the mixture
Since the number of moles of the mixture per unit volume is c, one has

p = cRT = RT cj = pj , (1.37)
j j
where
pj = cj RT = Rgj T = Xj p (1.38)
is the partial pressure exerted by species Aj when only this species occupies the vol-
ume of the mixture at temperature T and Rgj is the gas constant of the said species.
From (1.32), Eq. (1.37) can be written in the form

p Yj
= RT , (1.39)
j
Mj
or else, in the form

p
= Rm T, (1.40)

where
Yj R
Rm = R = = Rgj Yj (1.41)
j
Mj Mm j
is the gas constant of the mixture. Opposite to Rg , Rm changes with the composition
of the mixture. If the molar masses of the species are very different then the variations
of Rm with the composition of the mixture can be very large. Otherwise, these
variations are small. For simplicity, in many aerothermochemical problems Rm is
assumed to be constant, taken a mean value.
When needed to take into account the deviations of the state equation respect
to Eq. (1.41) this can be attained by using an equation similar to (1.1) for the mixture.
Thus, the second virial coefcient Bm (T ) for the mixture is given by the expression6
Brs Yr Ys
Bm = Mm . (1.42)
r s
Mr Ms
Here, if r = s, Brr is the second virial coefcient of species Ar previously

dened. If r 6= s, Brs is the second virial coefcient for the interaction potential
between the molecules of species Ar and As . Since, in general, the constants relative
to those mixed potentials are not available, several formulae have been proposed for
Brs as a function of Brr and Bss . For example7
( )
1 Mr Ms
Brs = Brr + Bss . (1.43)
2 Ms Mr
The potential energy of an ideal gas due to interaction between molecules is

zero. Therefore, each species adds to the value of any thermodynamic function per
unit mass of the mixture (internal energy, enthalpy, entropy, free energy, etc.) with
a value which is independent from the presence of other species. For Ai , the said
contribution to the value i of the corresponding thermodynamic function per unit
6 See Ref. [1], p. 153.
7 See Ref. [6], p. 211.
mass of the species at temperature T of the mixture and at partial pressure pi of Ai

multiplied by its mass fraction Yi . That is

= i Yi . (1.44)
i
In the following, Eq. (1.44) is applied to the calculation of the thermodynamic

functions of the mixture.
Internal Energy
The internal energy u per unit mass of the mixture is

u= ui Yi . (1.45)
i
When taking into this expression the value

T
ui = u0i + cvi (T ) dT (1.46)
T0
obtained when (1.9) is particularized for species Ai , it results

T
u = u0 + cv (T ) dT, (1.47)
T0
where

u0 = u0i Yi (1.48)
i
is the energy of the mixture at temperature T0 and

cv = cvi Yi (1.49)
i
is the heat capacity at constant volume per unit mass of the mixture.
Enthalpy
Similarly
T
h = h0 + cp (T ) dT, (1.50)
T0
with

h0 = h0i Yi (1.51)
i
and

cp = cpi Yi . (1.52)
i
Entropy
Similarly
T ( )
cp (T ) pj
s = s0 + dT Rgj Yj ln (1.53)
T0 T j
p0
with

s0 = s0j Yj . (1.54)
j
Equation (1.53) can also be written in the following form

( )
p
s = s0 Rm ln Rgj Yj ln Xj , (1.55)
p0 j
where

s0 = s0j Yj . (1.56)
j
Once p0 is xed, this value depends only on T and on the composition of the mixture.
In (1.55), the two rst terms of the right hand side give the entropy that the gas would
have at pressure p and at temperature T if instead of being a mixture it were a single
chemical species. The remaining term is the entropy of mixing whose contribution is
always positive.
Free Energy
Gibbs free energy
g = h T s, (1.57)
where h and s are given by (1.50) and (1.55) respectively. g can also be expressed in
the form ( )
p
g = g 0 + Rm T ln +T Rgj Yj ln Xj , (1.58)
p0 j
where
g 0 = h T s0 . (1.59)
Helmholtz free energy
f = u T s, (1.60)
where u and s are given by (1.47) and (1.55) respectively.

As in the previous cases, f can be expressed in the form

( )
p
f = f 0 + Rm T ln +T Rgj Yj ln Xj , (1.61)
p0 j
where
f 0 = u T s0 . (1.62)
Chemical Potentials
The chemical potential i of species Ai in the mixture is Gi

( ) ( )
pi p
i = Gi = G0i + RT ln = G0i + RT ln + RT ln Xi . (1.63)
p0 p0
1.4 Calculation of the thermodynamic functions
The preceding formulas show that all thermodynamic functions of a gas can be deter-
mined provided that one of them and the standard values for the others are known. For
example, the heat capacity, either at constant pressure or at constant volume, and the
formation enthalpy and entropy of the gas determine all other thermodynamic func-
tions. The law of variation of heat capacity as function of T is generally complicated.
Fig. 1.5, for example, obtained from the tables in Ref. [10], shows the curves of cp
versus T for some gases. These curves are usually approximated with empirical for-
mulae. For example, the following Table 1.3, taken from Ref. [4] p. 28, gives some
of these parabolic formulae as well as their maximum deviations within the range
300 < T < 2 000K. The use of linear formulas is advisable for smaller ranges of T .
Gas cp (cal gr1 K1 ) Max. error (%)

H2 3.440 + 0.033 103 T + 0.1395 106 T 2 1
N2 0.225 + 0.065 103 T - 0.0123 106 T 2 2
O2 0.196 + 0.086 103 T - 0.0241 106 T 2 1
CO 0.223 + 0.075 103 T - 0.0164 106 T 2 2
NO 0.207 + 0.081 103 T - 0.0204 106 T 2 2
H2 O 0.383 + 0.182 103 T - 0.0191 106 T 2 2
CO2 0.156 + 0.194 103 T - 0.0562 106 T 2 3
NH3 0.348 + 0.527 103 T - 0.1040 106 T 2 1
C2 H2 0.318 + 0.404 103 T - 0.1020 106 T 2 2 (for T > 400 K)
CH4 0.211 + 1.119 103 T - 0.2620 106 T 2 2
Table 1.5: cp (T ) correlations for several gases.

1.4. CALCULATION OF THE THERMODYNAMIC FUNCTIONS 17
1.4
1.2
CH
4
1.0
c (cal/gr K)
0.8
C2H2
H2O
0.6
p
H2 (0.1 cp)
0.4 CO2
0.2
O2
CO, N2
0.0
0 400 800 1200 1600 2000 2400 2800
T(K)
Figure 1.5: Specic heat at constant pressure of several gases as function of temperature.
When working at normal or not too high temperatures the determination of

the thermodynamic functions can be done through calorimetric measurements. On
the other hand, when working at very high temperatures, as are those encountered
in combustion processes, such measurements are very difcult when not impossible.
In such case one has to resort to Statistical Mechanics and to the use of the results
provided by the spectroscopic analysis of the gas. This method gives excellent results
specially when dealing with simple molecules.
The method is based on the formation of the so-called partition function of the
gas. Once this function is known all thermodynamic functions are really computed.8
Quantum Mechanics shows that the energy of a particle conned within a do-
main can only take discrete set of values. These values are called energy levels. Let
j (j = 1, 2, . . .) be the energy levels of the molecules, when the gas occupies volume
V at temperature T . The following expression

Q= exp (j /kT ) , (1.64)
j
is called partition function of the gas molecules. The summation is extended to all
energy levels. If one of them is degenerate, it must be counted as many times as cor-
responds to its order of degeneracy. That is, the number of discernible states in which
this level can be attained. Fixing volume V and temperature T of the gas its partition
8 For a full study on this matter see Refs. [11] and [12].
function is determined. Then, the problem lies in expressing the thermodynamic func-
tions of the gas by means of its partition function. According to Statistical Mechanics
the solution is as follows.
a) Internal energy.
( )
ln Q
u = Rg T 2 . (1.65)
T V
b) Enthalpy.
[( ) ( ) ]
ln Q ln Q
h = Rg T + . (1.66)
ln T V ln V T
c) Entropy.
[ ( ) ]
ln Q
s = Rg ln Q + T . (1.67)
ln T V
d) Free energy.
[ ( ) ]
ln Q
d.1) Gibbs: g = Rg T ln Q . (1.68)
ln V T
d.2) Helmholtz: f = Rg T ln Q. (1.69)
e) Chemical Potential.
[ ( ) ]
ln Q
G = Mg g = RT ln Q . (1.70)
ln V T
Heat capacities at constant volume, cv , and at constant pressure, cp , are ex-

pressed by u and h respectively in the form
u h
cv = , cp = . (1.71)
T T
This enables the calculation of their values as a function of Q by means of (1.65) and
(1.66).
Thus, the problem reduces to the determination of the energy levels for the for-
mation of Q. For the purpose, one must analyze the ways in which a gas molecule can
store energy. Except when working at very high temperatures, where it is necessary to
consider the states of electronic excitation in the molecule, this can be considered as
a system of material points which are its atoms. Then, the energy of each molecule is
the summation of the kinetic energies of its atoms plus the potential energy of the eld
of forces that hold them together. Be n the number of atoms in each molecule. Then,
its number of degrees of freedom is 3n. Of which three are external degree of freedom
corresponding to the translational motion of the center of gravity of the molecule. The
1.4. CALCULATION OF THE THERMODYNAMIC FUNCTIONS 19
remaining 3n 3 are internal degrees of freedom and they correspond to its rotational
and vibrational motions. Energy j of the molecule can always be expressed as
j = tj + ij . (1.72)
Here tj is the kinetic energy of the molecule due to the motion of its center of gravity,
and ij is the internal energy. Moreover, the translational levels tj are independent
from the internal levels ij . Therefore, the combination of either one of the tj with
either one of the ij gives the energy level j . Then, it is easily veried that the sepa-
ration (1.72) for j corresponds to a factorization
Q = Qt Qi (1.73)
for Q, where
( )
Qt = exp tj /kT (1.74)
j
and
( )
Qi = exp ij /kT (1.75)
j
are the translational and internal partition functions of the molecule respectively. All
thermodynamic functions are linear and homogeneous in ln Q and in its derivatives.
Hence, when factorization (1.73) is substituted into them, these functions are separated
into contributions from Qt and Qi respectively. Let be one of these functions. Then,
= t + i , (1.76)
where t is the contribution to from Qt , and i is the contribution from Qi . This

property simplies the calculation of the thermodynamic functions.
Translational partition function Qt is the same for all gases

( )3
2mkT 2
Qt = V . (1.77)
h2
Here, m is the mass of the molecule and h = 6.6238 1027 ergs is the Plancks
constant. On the other hand Qi depends on the structure of the molecule. At least in
rst approximation, ij can also be separated into contributions from internal energy
of rotation rj and internal energy of vibration vj
ij = rj + vj . (1.78)
Here, as before, rj and vj are approximately independent from each other. Therefore,
separation (1.78) gives for Qi
Qi = Qr Qv . (1.79)
Two cases can arise in the rotational motion of a molecule:
1) If the molecule is linear, that is, if all its atoms lie on a straight line, the moment
of inertia with respect to the axis of the molecule is zero and it cannot store
energy in such a degree of freedom. Hence, the rotational degrees of freedom of
the molecule reduce to two.
2) Whereas if the molecule is non-linear the number of its rotational degrees of
freedom is 3.
If the rotational coordinates are selected accordingly,9 the energies correspond-

ing to the rotational degrees of freedom are also separable. Then, except at tempera-
tures under ambient, the distribution Qrj from the rotational degree of freedom j to
the rotational partition function Qr is

2kT Ij
Qrj = 2 , (1.80)
h
where Ij is the moment of inertia of the molecule for such degree. Hence, we have:
a) For linear molecules

8 2 kT I
Qr =
, (1.81)
h2
where I is the moment of inertia of the molecule respect to an axis passing
through its center of gravity and normal to the axis of the molecule.
b) For non-linear molecules
( ) 32
1 8 2 kT
Qr = Ix Iy Iz . (1.82)
h2
Here Ix , Iy and Iz are the three moments of inertia of the molecule with respect
to three orthogonal axis at its center of gravity.
In (1.81) and (1.82) is a symmetry factor, that accounts for undisguised stable
states due to symmetries of the molecule. For example, for linear molecules = 2.
Thus, Qr is determined when the moments of inertia of the molecule are known. Such
moments are obtained from spectroscopic analysis.
As for vibration energy a direct determination of its levels through spectro-

scopic analysis is best. Independent contributions of the vibrational degrees of free-
dom can, otherwise, be obtained by assuming that their corresponding energies are
separable. Which happens, for example, when vibration amplitudes are small enough
so that the potential energy of the molecule can be approximated by a quadratic func-
tion of the deviations of the atoms respect to their equilibrium positions. For this
9 For which the principal axis of inertia of the molecule must be taken.
1.5. CHEMICAL REACTIONS IN A MIXTURE OF GASES 21
case, a normal system of coordinates can be adopted where the vibrational energy is
separated into contributions from the natural modes of the molecule. Be s the fre-
quency of one of these modes. Under the assumption that the molecule behaves as an
harmonic oscillator the contribution Qvs of this mode to Qv is
( )( ( ))1
hs hs
Qvs = exp 1 exp . (1.83)
2kT kT
Frequency s is obtained from the vibration spectra of the molecule.
Finally, for very high temperatures the contribution of electronic excitation

must be included in the value of Q. The corresponding energy levels are obtained
from the spectra.
The preceding analysis assumes that the rotational and vibrational energies are
separable. Yet, this does not always happen, mainly for high temperatures. In fact, ro-
tation stretches the molecule and this introduces coupling terms between the rotational
and vibrational degrees of freedom, preventing a separate study of both contributions.
Besides, the assumption that vibration is harmonic not always is justied, which forces
the introduction of correcting terms to account for anharmonicity.10
1.5 Chemical reactions in a mixture of gases
Now, let us assume that the ` species Ai of the mixture can react according to a system
of r reaction equations of the form

0 00
ij Ai ij Ai , (j = 1, . . . , r), (1.84)
i i
0 00
where ij and ij are the stoichiometric coefcients of species Ai in the reaction j for
the forward and backward reactions respectively. Let
00 0
ij = ij ij . (1.85)
These coefcients must satisfy the following conditions

ij Mi = 0, (j = 1, . . . , r), (1.86)
i
which express that the mass of the mixture is not changed by chemical reactions.
Let us consider the unit mass of the mixture and be Cij the number of moles
of species Ai produced by the reaction j. Due to (1.84) the following relations exist
10 For further information on this problem see Refs. [11] and [12].
between Cij
C1j C2j C`j
= = ... = . (1.87)
1j 2j `j
Let j be the common value of these relations. j is called the degree of

advancement of reaction j (De Donder [13]). Thus dened, j has the dimensions
mole/gr. The number of moles of species Ai produced by reaction j is, therefore,
Cij = ij j , (1.88)
and the number Ci of moles of Ai produced by the r reactions is

Ci = Cij = ij j . (1.89)
j j
If Ci0 is the number of moles of species Ai existing when the reactions start,
the number Ci that will exists when the degrees of advancement of the reactions are
j , (j = 1, 2, . . . , r), is

Ci = Ci0 + ij j . (1.90)
j
Similarly the mass fraction Yi of species Ai produced by the r reaction is

Yi = Mi ij j , (1.91)
j
and if Yi0 is the initial mass fraction of this species, one has

Yi = Yi0 + Mi ij j . (1.92)
j
1.6 Chemical equilibrium
Conditions of equilibrium
Once the reactions are initiated they continue up to the point where the mixture reaches
its state of equilibrium. Such an equilibrium is determined by additional conditions
which x the state of the system. Let us assume, for example, that the process is
adiabatic and takes place at constant pressure. Such a case has a great practical interest
in the study of combustion processes. Then the First Law of Thermodynamics shows
that the enthalpy of the mixture must be constant. Therefore, the conditions that must
be satised by the mixture are as follows
h = h0 = const., p = p0 = const. (1.93)

1.6. CHEMICAL EQUILIBRIUM 23
Since the process is adiabatic the state of equilibrium is determined by the

condition that entropy s be maximum. The condition for this is that its rst variation
be zero. Moreover, from (1.93) the variations of h and p must also be zero. Therefore,
the state of equilibrium is determined by conditions
h = p = s = 0. (1.94)
Thermodynamics shows11 that the entropy variations of the mixture is given by

the expression
i
T s = h V p Yi . (1.95)
i
Mi
Here V is the volume of the unit mass of the mixture and i is the chemical potential
of species Ai in the mixture. This relation, by virtue of (1.94), reduces to
i
Yi = 0. (1.96)
i
Mi
Variations of Yi cannot be arbitrary since Yi must satisfy conditions (1.92). From

them one obtains

Yi = Mi ij j . (1.97)
j
This expression when taken into (1.96) gives

( )

i ij j = i ij j = 0, (1.98)
i j j i
and since the variations j are arbitrary this condition breaks into the following sys-
tem of equilibrium equations

i ij = 0, (j = 1, 2, . . . , r). (1.99)
i

The expression i ij is named afnity j of reaction j. Therefore, the equi-
i
librium is determined by the condition that the afnities of the reactions become zero
j = 0, (j = 1, 2, . . . , r). (1.100)
System (1.100), together with conditions (1.92) and (1.93), determine the val-
ues for the degrees of advancement of the various reactions and for the corresponding
mass fractions of the species as well as for the temperature T of the mixture.
11 See Ref. [4], p. 68.
Remarks
If the reactions take place at pressure and temperature (instead of enthalpy) both con-
stant the conditions of equilibrium are
p = T = g = 0. (1.101)
But
g = h sT T s. (1.102)
From Eqs. (1.95), (1.97), (1.101) and (1.102) conditions (1.99) are obtained again.
Thus conditions of equilibrium are the same for both cases, with the only difference
that here, temperature is given whilst in the previous case it must be calculated. The
computation is done by determining the conditions of equilibrium as a function of tem-
perature, and then looking for the value of T which satises the additional condition
h = h0 .
Likewise, for constant volume adiabatic reactions the conditions of equilibrium

are
u = V = s = 0, (1.103)
from which result again conditions (1.99).
1.7 Case of a mixture of ideal gases
For the case of a mixture of ideal gases the form taken by the conditions of equilibrium
is specially simple. In fact, as for (1.63) system (1.99) reduces to

Gi ij = 0, (j = 1, 2, . . . , r), (1.104)
i
or else
( )
pi
G0i ij + RT ij ln = 0, (j = 1, 2, . . . , r), (1.105)
i i
p0
which together with (1.38) gives

( )j ( )
p0 G0j
ij
Xi = exp , (j = 1, 2, . . . , r), (1.106)
i
p RT
where

j = ij (1.107)
i
1.7. CASE OF A MIXTURE OF IDEAL GASES 25
and
( )
G0j = Hi T Si0 ij (1.108)
i
is the increase in free energy due to reaction j at temperature T of the mixture and
at standard pressure p0 . Thus G0j depends only on T . Table 1.2 gives the values of
G0j for several species corresponding to their formation from the chemical elements
at the standard state. Therefore Eq. (1.106) can be written
( )j
p0
ij
Xi = Kj0 (T ) , (j = 1, 2, . . . , r), (1.109)
i
p
where Kj0 is called the equilibrium constant of reaction j depending only on temper-
ature T . The values of Kj0 versus T are given in Fig. 1.6, for several typical reactions
of interest in combustion problems.
30
1
2 H2 + OH H2 O
25
20
CO + 12 O2 CO2
15
log10(Kp)
H2 + 21 O2 H2 O
10 1
2 N2 + H2 O NO + H2 O
CO2 + H2 CO + H2 O H2 O H2 + O
5
1 1
2 N2 + 2 O2 NO
0 1
2 H2 H
5
1
2 N2 N
10
0 400 800 1200 1600 2000 2400 2800 3200 3600
T(K)
Figure 1.6: Equilibrium constant of reaction vs temperature, for several typical reactions of
interest in combustion problems.
The chemical equations (1.84) used for this calculation can be any ones within
the following limitations: a) they must be linearly independent; b) their number must
be the maximum that can exist between species. This number can be determined when
the phase rule is applied. Be `0 < ` the number of components of the mixture, then
the number of independent reactions will be ` `0 . In turn `0 can be determined as
follows: let Ej , (j = 1, 2, . . . , `), be the chemical elements forming the species and
aij be the number of atoms of element Ej in species Ai . The number of components
of the mixture is the rank of the matrix of coefcients aij . The coefcients of a
principal minor of order `0 of this matrix determine the species that can be adopted
as components of the mixture. A complete system of independent chemical reactions
can now be obtained by expressing each one of the remaining species as a linear
combination of the said components in the form
0

`
Aj = bji Ai , (j = `0 + 1, `0 + 2, . . . , `). (1.110)
i=1
Here, the rst `0 species are assumed to be the components of the mixture.
For example in the combustion of mixtures of hydrocarbons and some of their

compounds with oxygen and nitrogen, the combustion products are formed by the
species: CO2 , CO, H2 O, HO, H2 , H, O2 , O, NO, N2 and N. Therefore, one has:
` = 11; `0 = 4; and the number of independent chemical reactions is 7. The fol-
lowing systems of reactions can be adopted (see Ref. [13]) since they are linearly
independent:
C O2 + H2 C O + H2 O
1
N2 + H2 O N O + H2
2
2 H2 O 2 H2 + O2
H2 O O + H2 (1.111)
N2 2 N
H2 2 H
1
H2 O H2 + O H
2
The r equations (1.106), together with the `0 equations which express the con-
servation of components in the reactions, determine the equilibrium composition of
the mixture, once its temperature is known. Thus in the preceding example system
(1.111) must be completed with the four equations for the conservation of the number
of atoms of carbon, oxygen, hydrogen and nitrogen. It is not always necessary to con-
sider all possible reactions since depending on the state of the mixture the fractions of
some of the species may be neglected and the computation simplied.
For the case of adiabatic equilibrium the temperature of the products is un-
known. Hence, trial and error methods become necessary. In applying then, rst a
temperature is assumed and the equilibrium composition is determined for it. Once
this is known the corresponding temperature is obtainable through equation (1.10),
h = h0 , and if this temperature differs from the one previously assumed, the compu-
tation should be reviewed. Gayden and Wolfhard [14] recommend the application of
Damkohler and Edses method mentioned in Ref. [15]. The fundamentals and some
of the applications of this method are described in the book by Gayden and Wolfhard.
1.7. CASE OF A MIXTURE OF IDEAL GASES 27
0.30 3200
Tb
0.25 3000
H O
2
0.20 2800
CO (X /2)
CO CO
H
T(K)
2 2
Xi
0.15 2600
O (X /2)
2 O
2
H
0.10 2400
OH
0.05 2200
O
0.00 2000
0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20
YO / (YO )s
2 2
Figure 1.7: Compositions and temperature of the combustion products obtained when burn-
ing a mixture of ethylene and oxygen as a function of the ratio of the mass
fraction of oxygen to the corresponding stoichiometric mixture.
Other methods such as those in Refs. [16] and [17] include diagrams in order to re-
duce calculations as they are very cumbersome when the number of species is large.
Then, specially when many cases must be solved, it becomes necessary to systematize
the procedure. This is accomplished by giving computational schemes which enable
the use of electronic computers. Full description of these methods can be found, for
instance, in Refs. [18] and [19] where additional bibliography is included.
Fig. 1.7, taken from Ref. [14], shows the compositions and temperature of the
combustion products obtained when burning a mixture of ethylene and oxygen as a
function of the ratio of the mass fraction of oxygen to the corresponding stoichiometric
mixture.
Fig. 1.8, taken from Ref. [17], gives the theoretical combustion temperatures
for some hydrocarbons.
Table 1.6, taken from Ref. [14], gives the adiabatic combustion temperatures
for some typical mixtures, as well as these that would be obtained by neglecting dis-
sociations. The book by Lewis and von Elbe mentioned in Ref. [20] gives the experi-
mental adiabatic combustion temperatures for great number of mixtures. See Ref. [21]
for the experimental technique used for the determination of combustion temperatures.
2500
acetylene
2400
methylacetylene
2300
Tb (K)
ethene
2200
benzene
propylene
a:nhexane 1butene
b: npentane 1pentene
2100
c: nbutane
d: propane
cyclopentene
metane a,b,c,d
2000 etane
0.8 0.9 1.0 1.1 1.2 1.3 1.4
Y / (Y )
F F s
Figure 1.8: Theoretical combustion temperatures of several hydrocarbons in the air at at-
mospheric pressure.
Fuel CH4 CH4 C3 H8 C3 H8 C8 H18 C8 H18 C2 H2 H2

Mixture
8
< Fuel
> 0.2 0.1062 0.2245 0.1738 0.2218 0.1076 0.0918 0.0305
Yi O2 0.8 0.3865 0.7755 0.6005 0.7782 0.3777 0.2113 0.1360
>
:
N2 0 0.5073 0 0.2257 0 0.5147 0.6969 0.8335
Products
8
> H2 O
>
>
0.3719 0.2735 0.2988 0.2627 0.2562 0.1853 0.0745 0.1886
>
> O2 0.0691 0.0128 0.0665 0.0483 0.0880 0.0307 0.0015 0.00
>
>
>
>
>
> O 0.0326 0.0041 0.0375 0.0230 0.0462 0.0073 0.0008 0.00
>
>
>
>
>
> H2 0.0690 0.0343 0.0648 0.0495 0.0538 0.0175 0.0170 0.1505
>
>
< OH 0.1438 0.0402 0.1378 0.1010 0.1447 0.0453 0.0092 0.00
Xi
>
> CO 0.1512 0.0747 0.2064 0.1612 0.2167 0.0782 0.1159 0
>
>
>
> CO2 0.1166 0.0818 0.1372 0.1232 0.1450 0.1259 0.0810 0
>
>
>
>
>
> NO 0 0.0054 0 0.0100 0 0.0091 0.0019 0.00
>
>
>
>
>
> N2 0 0.4527 0 0.1880 0 0.2923 0.6934 0.5609
>
:
H 0.0458 0.0105 0.0510 0.0331 0.0493 0.0084 0.0048 0.00

Temperatures ( C)
Dissociation 2 737 2 410 2 776 2 689 2 809 2 447 2 300 1 358
No Dissociation 5 047 2 980 5 202 4 405 5 740 3 417 2 439 1 362
Table 1.6: Adiabatic combustion temperature for some typical mixtures.

1.8. CHEMICAL KINETICS 29
1.8 Chemical kinetics
In some of the Aerothermochemistry problems it is essential to know the reaction

rates of the species forming a mixture of gases.12 For which the following must be
determined:
1) The reactions that can take place between the species.

2) The governing laws of their reaction rates.
Such questions are the subject of Chemical Kinetics.
In Ref. [22] an introduction to Chemical Kinetics can be found mainly from

the viewpoint of jet propulsion. For a more complete study on this subject see Refs.
[23], [24] and [25]. The present paragraph gives a brief introduction to this matter.
First let us consider the case where the species react according to a single
irreversible reaction

i0 Ai i00 Ai , (1.112)
i i
and let be the degree of advancement of this reaction as was dened in paragraph 5.
The reaction rate of Eq. (1.112) is, by denition,
d
r= = . (1.113)
dt
From here, keeping in mind (1.88), the number dCi / dt of moles of species Ai
produced per unit mass of the mixture and per unit time is
dCi d
= (i00 i0 ) . (1.114)
dt dt
Likewise, the mass dmi / dt of these species produced per unit mass and time is
dmi dCi d
= Mi = Mi (i00 i0 ) . (1.115)
dt dt dt
Aerothermochemistry deals, normally, with mass wi of Ai produced per unit

volume and per unit time, which is
dmi d
wi = = Mi (i00 i0 ) . (1.116)
dt dt
The problem lies in determining the way in which wi depends on the state and com-
position of the mixture. Such determination is done through the so-called law of mass
action of Guldberg and Waage [26].
12 See, i.e., Chap. 6.
This law states that the number dci / dt of moles of Ai produced within unit
volume during unit time is
dci 0
= (i00 i0 ) k cj j . (1.117)
dt j
Here, k is independent from the composition and pressure of the mixture and depends
only on its temperature. It is called specic rate or rate coefcient of the reaction and
( )1 1
its dimensions are cm3 /mol s , where

= j0 (1.118)
j
is called order or molecularity of the reaction.
Since the molar mass of Ai is Mi , we have for wi , from (1.117)

dci 0
wi = Mi = Mi (i00 i0 ) k cj j . (1.119)
dt j
This formula can also be expressed as a function of the mass fractions of the reacting
species, by virtue of (1.31), in the form
0
wi = 0 kMi (i00 i0 ) Yj j . (1.120)
Mj j j
j
For example, in the case of a bimolecular reaction between species A1 and A2 which
produces A3 ,
A1 + A2 A3 , (1.121)
one has, due to (1.120),

w1 w2 w3 dc1 dc2 dc3 2
= = = = = = kY1 Y2 . (1.122)
M1 M2 M3 dt dt dt M1 M2
The way in which the specic rate k depends on temperature is given by Arrheniuss
law
k = AeE/RT (1.123)
where E is called activation energy of the reaction and A is a constant or the product
of a constant by a power of the absolute temperature T
A = BT . (1.124)
Here B is a constant called frequency factor and is an exponent within zero and one.
Arrhenius [27] deduced his law empirically as a generalization of vant Hoffs

law for the variation of equilibrium constants as functions of temperature. Statistical
Mechanics13 enables a theoretical derivation of this law throughout the so-called Col-
lisions Theory. For this derivation it is enough to assume that for a reaction to occur
the following is needed:
1) Coincidence of reacting molecules in a collision.

2) The collision energy in certain degrees of freedom must be larger than a given
value which depends on the activation energy of reaction.
When calculating the number of collisions per unit volume and per unit time that sat-
isfy the preceding conditions, one obtains expressions similar to the Arrhenius law for
the inuence of temperature and to the law of mass action for the inuence of concen-
trations. For example, in the simple case of a bimolecular reaction of the type (1.121)
the number n of collisions per unit volume and per unit time between molecules of
the species A1 and A2 is
( )
M1 + M2
n= N 2 12
2
c1 c2 8RT . (1.125)
M1 M2
Here 12 is the mean molecular diameter of both species. Let us assume that from the
n collisions (1.125) only are active, that is, only produce reaction, those that have a
relative kinetic energy of approximation larger than = E/N .14 The number na of
these active collisions is
na = n eE/RT . (1.126)
Hence, the velocity dc1 / dt of the reaction, in moles per unit volume and per unit
time, is
dc1 dc2 dc3 na
= = = . (1.127)
dt dt dt N
That is, due to (1.125),
( )
dc1 dc2 dc3 M1 + M2 E/RT
= = = N 12
2
c1 c2 8RT e . (1.128)
dt dt dt M1 M2
Due to (1.31), this formula can be written in the form
dc1 dc2 dc3

= = =
dt dt dt ( ) (1.129)
2 M1 + M2 E/RT
2
N 12 Y1 Y2 8RT e ,
M1 M2 M1 M2
13 See Ref. [23].
14 Itis not essential that the kinetic energy of approximation of the collision be precisely the one larger
than . The formulae holds as long as the collision energy in two separated degrees of freedom is larger
than .
which can be identied with Eq. (1.122) by making

( )
2 M1 + M2 E/RT
k = N 12 8RT e , (1.130)
M1 M2
that is, from Eq. (1.124),

1
= (1.131)
2
and ( )
2 M1 + M2
B= N 12 8R . (1.132)
M1 M2
When comparing Collisions Theory with experimental results, an exact numer-

ical agreement between the reaction rates predicted and observed is not to be expected,
but an agreement of the orders of magnitude. In fact, Collisions Theory does not in-
clude, among others, the inuence of the relative orientation of the molecules as they
collide which can decide the success of the collision. Such an effect can be of impor-
tance specially in molecules of complicated structure. It is taken into consideration
by including in Eq. (1.129) a numerical factor P of orientation equal or smaller than
unity
dc1 dc2 dc3

= = =
dt dt dt ( ) (1.133)
2 M1 + M2 E/RT
P 2
N 12 Y1 Y2 8RT e .
M1 M2 M1 M2
Experimental reaction rates agree, fairly, with those predicted by Collision Theory
only for a short number of reactions, yet in other cases experimental rates are as much
as 108 times larger or smaller than those predicted by theory. Such a discrepancy can
not be justied within the theory. These results that in the best cases Collision Theory
represents approximately the actual facts only for a short number of reactions. A more
correct formulation is provided by the so-called Theory of Absolute Reaction Rates.
In this theory it is assumed that the passing of the reacting species to the prod-
ucts takes place through the formation of an activated complex resulting from the
assembly of the reacting species. This complex is considered as located at the top of
an energy barrier which separates the species from the products. The reaction rate
is determined by the velocity at which the activated complex crosses the barrier. It
is, furthermore, assumed that such complex stands in equilibrium with the reacting
species and that its dissociation in products is due to the action of one of the vibra-
tional degree of freedom. Then, by applying the laws of Statistical Mechanics the
reaction rate can be computed when the structure of the activated complex is known.
This structure can be obtained from Quantum Mechanics or be postulated from the
structure of the molecules. The Absolute Reaction Rate Theory has been successfully
applied to the study of a great number of reactions well as to many other physical
and physicochemical phenomena. A full development of the theory can be found, for
instance, in the works mentioned in Refs. [24] and [25].
In the previous study it has been assumed that a single one directional reaction
takes place within the mixture. Generally, all reactions proceed simultaneously in both
senses at different rates. Thus, reaction (1.112) must be substituted by the following

i0 Ai i00 Ai , (1.134)
i i
and the rate wi of production of species Ai is given by the expressions

( )
0 00
00 0
wi = Mi (i i ) kf ci kb ci
i i
. (1.135)
i i
Here kf and kb are the specic rates corresponding to the forward and backward re-
actions respectively.
On the other hand to be able to form the products, the reacting molecules must
contact one another. This reduces the molecularity of a single elementary reaction to
1, 2 or 3, since the probability of a collision of more than three molecules is practi-
cally nil. Consequently, when the reaction rate depends in a complicated way on the
concentrations, it can be assured that it proceeds from the combination of several ele-
mentary reactions. The basic problem of Chemical Kinetics is then the establishment
of the set of elementary reactions that produce within the mixture. For example, the
stoichiometric equation for the decomposition of ozone is
2O3 3O2 . (1.136)
While the following is the actual system of reactions [28]:

kf i
O3 +G O+O2 +G, (1.137)
kb1
kf 2
O+O3 2O2 , (1.138)
kb2
kf 3
O2 +G 2O+G. (1.139)
kb3
Here G is one of the atoms or molecules of the mixture. Each one of these reactants
acts at its own reaction rate and the rate of consumption of the ozone results from
the combination of the rates corresponding to elementary reactions. Denoting species

O, O2 and O3 with subscripts 1, 2 and 3 respectively, one obtains for the rate of
consumption of O3
w3
= kf 1 cc3 kb1 cc1 c3 + kf 2 c1 c3 kb2 c22 . (1.140)
M3
This expression differs essentially from that obtainable from Eq. (1.136).
The complexity of the system of the elementary reactions that produce within
a mixture of reacting species originates an essential difculty for the study of these
problems which depend on the reaction rates of the species. One of such problems
is, for example, the calculation of the propagation velocity of a ame throughout a
combustible mixture.15 A great effort is being made by Chemical Kinetics towards the
enlightenment of the elementary reactions that produce in each case and of their corre-
sponding rates. Specially in the eld of Combustion, the Fifth International Congress
held in 1954 devoted the major part of its work to the study of Combustion Kinetics.16
In spite of the efforts the actual state of knowledge is still scant making impossible a
full study of the process except for a short number of particular cases.17
Even when assuming that the system of reactions is known, it is usually too
complicated to be fully included in the study of an Aerothermochemical problem. For
this case the system must be simplied, for example with the steady state assump-
tion,18 or else by adopting global reaction rates for the mixture such as
w = An (1 Y ) eE/RT ,
n
(1.141)
where Y is the mass fraction of products and n is the molecularity of the overall
reaction. In the following chapters frequent use will be made of similar expressions
for the approximate study of several problems.
References
[1] Corner, J.: Theory of Interior Ballistics of Guns. John Wiley and Sons, Inc.
New York, 1950.
[2] Taylor, J.: Detonation in Condensed Explosives. Oxford, at the Clarendon
Press, 1952.
15 See
chapter 6.
16 See
the Proceedings of the Congress, soon to be published.
17 See chapter 6.
18 See Ref. [28] where von Karman and Penner have shown that the introduction of such an assumption
for the oxygen atoms in the ozone ame is justied. Yet, for other cases it appears less justied. For
example, in the HydrogenBromine ame.
[3] Hirschfelder, J. O., Curtiss, C. F. and Bird, R. B.: Molecular Theory of Gases
and Liquids. John Wiley and Sons, Inc. New York, 1954.
[4] Prigogine, I. and Defray, R.: Chemical Thermodynamics. Longmans, Green
and Co., New York, 1954.
[5] Guggenheim, E. A.: Thermodynamics. North Holland Publishing Comp. Ams-
terdam, 1950.
[6] Taylor, H. S. and Glastone: A Treatise on Physical Chemistry, Vol. II, States of
Matter. D. Van Nostrand Comp. Inc., New York, 1951.
[7] Lewis, G. N. and Randall, M.: Thermodynamics and the Free Energy of Chem-
ical Substances. McGraw-Hill Book Co., New York.
[8] National Bureau of Standards: Selected Values of Chemical Thermodynamic
Properties. 1950.
[9] Rossini, F. D.: Chemical Thermodynamics. John Wiley and Sons, Inc., New
York, 1950.
[10] The NBS-NACA Tables of Thermal Properties of Gases.
[11] Mayer, J. E. and Mayer, M. G.: Statistical Mechanics. John Wiley and Sons,
Inc., 1940.
[12] Pitzer, K. S.: Quantum Chemistry. Prentice-Hall, Inc., New York, 1953.
[13] Hirschfelder, J. O., McClure, F. T., Curtiss, C. F. and Osborne, D. W.: Ther-
modynamic Properties of Propellant Gases. NDRC Rep. No. A-116, Nov. 23,
1942.
[14] Gayden, A. G. and Welfhard H. G.: Flames. Their Structure Radiation and
Temperature. Chapman and Hall Ltd., 1953.
[15] Damkohler, G. and Edse, R.: Zeitschrift fur Elektrochemi, Vol. 49, 1943, p. 178.
[16] Huff, V. W. and Caluart, C. S.: Charts for the Computation of Equilibrium
Composition of Chemical Reactions in the Carbon-Hydrogen-Oxygen-Nitrogen
System at Temperatures from 2 000 to 5 000 K. NACA Technical Note No. 1653,
July 1948.
[17] Vichnievsky, R., Sale, B. and Marcadet, J.: Combustion Temperatures and Gas
Composition. Jet Propulsion, Vol. 25, No. 3, March 1955, p. 105.
[18] Huff, V. N., Gordon, S. and Morrell, V.: General Method and Thermodynamic
Tables for Computation of Equilibrium Composition and Temperature of Chem-
ical Reactions. NACA Technical Report No. 1037, 1951.
[19] Brinialey, S. R.: Computational Methods in Combustion Calculations. Com-
bustion Processes, Sec. C, Vol. II of High Speed Aerodynamics and Jet Propul-
sion, Princeton University Press, 1956, p. 64.
[20] Lewis, B. and von Elbe, G.: Combustion, Flames and Explosions of Gases.
Academic Press., Inc. Publishers, New York, 1951.
[21] Bundy, F. P. and Streng, H. M.: Measurement of Flame Temperature, Pressure
and Velocity. Physical Measurement in Gas Dynamics and Combustion, Sec. I,
Vol. IX of High Speed Aerodynamics and Jet Propulsion, Princeton University
Press, 1954.
[22] Penner, S. S.: Introduction to Chemical Kinetics. AGARD AG 5/p2, December
1952.
[23] Hinshelwood, C. N.: The Kinetics of Chemical Change. Oxford University
Press, 1940.
[24] Glastone, S., Laidler, K. J. and Eyring, H.: The Theory of Rate Processes.
McGraw-Hill Book Company, Inc., New York, 1941.
[25] Laidler, K. J.: Chemical Kinetics. McGraw-Hill Book Company, Inc., New
York, 1950.
[26] Taylor, H. S.: Fundamentals of Chemical Kinetics. Combustion Processes, Sec.
D, Vol. II of High Speed Aerodynamics and Jet Propulsion, Princeton Univer-
sity Press, 1956, p. 101
[27] Arrhenius, S.: Zeitschrift fur Physikalische Chemic, 1889.
[28] von Karman, Th. and Penner, S. S.: Fundamental Approach to Laminar Flame
Propagation. Selected Combustion Problems, Vol. I, AGARD, Paris, 1954,
pp. 5-41.
Chapter 2
Transport phenomena in gas

mixtures
2.1 Introduction
When the state or composition of a mixture of gases are not uniform, a transport
of mass, momentum and energy takes place between different points of the mixture
tending to level the initial differences.
At the neighborhood of each point transport depends on the state and compo-
sition of the mixture and on the lack of uniformity from which it arises, according
to laws dealt with in the present chapter. Even though these phenomena arise from
molecular motion their description can be made through phenomenological variables
except when the gas is very rareed or when the variations in state and composition
within distances comparable with the molecular mean free path are large.1 If such
cases are excluded, gas can be assimilated to a continuous medium where at each of
its points density, pressure, temperature, mass fractions and velocities of the species
and mixture are denable. Under such conditions, let P , Fig. 2.1, be a point, xed or
in motion at a given velocity v, and d a surface element linked to it. Transport dF
of mass of one of the species, or of momentum or energy of the mixture, through d
during time dt is of the form
dF = f d dt, (2.1)
where f is the transport per unit surface and per unit time. f is a function of the ori-
entation of d, dened by unit vector n normal to d, of the variables that determine
the state and composition of the mixture at the point and of their derivatives. If P
1 See chapter 3 1.
37
38 CHAPTER 2. TRANSPORT PHENOMENA IN GAS MIXTURES
v
d
P n
Figure 2.1: Schematic diagram to dene the transport phenomena.
moves at velocity v of the mixture at the point2 and mass transport of the species is
considered, the phenomenon is called diffusion. By considering the transport of mo-
mentum one obtains pressure and viscous stresses. Finally when energy transport is
considered heat transfer is obtained.
In the following paragraphs these three phenomena will be studied in sequence

adopting as a starting point the expressions given for them by the Kinetic Theory of
Gases. This theory states the problem as follows: let fi (vi , x, t) be the velocity dis-
tribution function of species Ai of the mixture. By denition, the probable number
of molecules of species Ai whose velocities lie between vi and vi + dvi , and which
are contained in volume element dV around point P of spatial coordinate x, at instant
t, is fi (vi , x, t) dV dvi1 dvi2 dvi3 , where vi1 , vi2 and vi3 are the three components
of vi in a rectangular cartesian system of coordinates. Due to molecular collisions
a statistical equilibrium establishes which determines the values for fi . Such values
are determined by a system of Boltzmann integral equations. When the state of the
mixture is uniform, the solution to this system is Maxwells velocity distribution for
each species. Otherwise, the solution depends not only on the state at the point but
also on its rate of variation when passing from one point to another. This variation is
dened by the successive derivatives of the variables of state and composition at the
point. If such derivatives are small, that is, when relative variations of density, pres-
sure, temperature, velocities and mass fractions at the point are small within distances
of the order of the molecular mean free path,3 an approximate solution to Boltzmanns
system can be obtained by means of the method of the perturbations developed by En-
skog and Chapman. This method looks for solutions to Boltzmanns system which
differ slightly from Maxwells fundamental solution. Solutions show that transport of
mass, momentum and energy depend linearly on the rst order derivatives of the vari-
ables of state and composition at the point. Coefcients of such linear functions are
those of diffusion, viscosity and thermal conductivity of the mixture depending only
on the state and composition at the point and on dynamics of molecular collisions. For
a detailed study on the subject see Refs. [1] and [2].
2 See 2 of this chapter for denition of v.
3 Such is normally the case with exceptions as, for example, inside shock waves.
2.2. DIFFUSION 39
Hereinafter, notation in chapter 1 will be kept, adopting the vectorial and ten-
sorial notation dened in the Appendix to chapter 3.
2.2 Diffusion
Let vi be the velocity of species Ai at P .4 vi is generally different from velocity v of

the mixture. The difference vdi is called diffusion velocity of Ai
vdi = vi v. (2.2)
The following relations exist between v, vi and vdi

v = Yi vi , (2.3)
i

Yi vdi = 0. (2.4)
i
Flux dmi of Ai through d during time dt, when P moves at velocity v of the
mixture, is obviously
dmi = i vdi n d dt. (2.5)
The problem lies in calculating ux vector fi ,
fi = i vdi . (2.6)
Let us rst consider the case of a binary mixture formed by species A1 and A2 .
The problem has been solved by Enskog [3] and Chapman [4]. The ux of A1 is
( )
M1 M2 p T
f1 = Y1 vd1 = 2
D12 X1 (Y1 X1 ) DT 1 . (2.7)
Mm p T
Here D12 and DT 1 are, respectively, the coefcients of diffusion and thermal diffu-
sion5 of the mixture. Flux f2 of A2 is, from (2.4) and (2.6),
f2 = f1 . (2.8)
4 v
i is dened as follows: taking a volume element dV around P and forming the mean value of veloci-
ties vi , of the molecules Ai on it. Such value is vi
1 X j
vi = v ,
Ni j i
where Ni is the number of molecules of Ai on dV . If dV is large respect to the molecular scale and
small respect to the macroscopic scale, the value vi thus dened is independent from size and shape of dV .
Velocity v of the mixture is, by denition
X
v = Yi vi .
i
5 The coefcient of thermal diffusion must not be confused with the thermal diffusivity, Ed.
Formula (2.7) shows that there are three causes for diffusion corresponding to the three
terms in the right hand side of the equation. Namely, the differences in composition,
pressure and temperature.6 Elementary Theory of Diffusion only acknowledge the
rst of these three causes.7
In Aerothermochemical problems it is advantageous to eliminate mole fraction

X1 from Eq. (2.7). Furthermore, by doing so a simplied expression is obtained.
Elimination is immediately attained by keeping in mind the following relations8
Mm
X1 =
Y1 , (2.9)
M1
( )
1 Y1 Y2 1 1 1
= + = + Y1 , (2.10)
Mm M1 M2 M2 M1 M2
Thus obtaining
( )
M1 M2 p T
f1 = D12 Y1 + D12 Y1 (1 Y1 ) DT 1 . (2.11)
Mm p T
The rst term in the right hand side of this expression gives the diffusion ux cor-
responding to differences in composition. This term express Ficks law. The second
term gives the diffusion ux arising from differences in pressure. Since D12 is al-
ways positive, formula (2.11) shows that pressure diffusion tends to carry the heavier
molecules towards the higher pressures. The third term gives the diffusion ux arising
from differences in temperature. Thermal diffusion coefcient DT 1 can be positive,
negative or zero. Therefore, no general rules can be giving regarding the sens of this
ux.
Formula (2.11) shows that diffusions behaviour in a binary mixture of gases

is determined by the values of two coefcients: D12 and DT 1 . Frequently DT 1 is
substituted by
2
Mm DT 1
kT = . (2.12)
M1 M2 D12
Systematic measurements on diffusion coefcients of gas mixtures are not
available. Therefore theoretical computations become necessary. They can be per-
formed when knowing the interaction potential between molecules. If forces between
molecules are independent from their relative orientation, the interaction potential
takes the form ( )
r
= 12 f , (2.13)
12
6 Where selective elds of forces exists, that is, elds acting with different strength over the two species
of the mixture, they originate a new cause of diffusion. See Ref. [1], p. 143. Such occurs for example,
when there exist ionized molecules and the mixture is submitted to the action of an electromagnetic eld.
7 See Ref. [5], p. 198.
8 See chapter 1, pp. 12-13.
2.2. DIFFUSION 41
where r is the distance between centers of the molecules, 12 is an intensity constant

and 12 is a collision diameter. Consequently, once the form of f is known, the values
of 12 and 12 determine the potential. When forces between molecules depend on
their relative orientation, as occurs whit polar molecules, more complicated expres-
sions than (2.13) are needed for . Such expressions must include the inuence of
the relative orientation on the collision. An example of such potentials, which has
been used by Hirschfelder et al. for the computation of transport coefcients of polar
molecules [6], is Stockmayer potential
(( )
12 )12 ( 12 )6 2
= 412 12 f (1 , 2 , 2 1 ) . (2.14)
r r r3
The rst term in the right hand side is Lennard-Joness potential and it repre-
sents the part of interaction independent from the relative orientation of molecules.
The second term gives the attraction between two dipoles whose relative orientation
is dened in Fig. 2.2. So far very few results are available on transport coefcients in
mixtures of polar molecules.9
1 2
1 2
+ +
Figure 2.2: Schematic diagram of the angles dening the orientation of two polar
molecules.
The values for the coefcients of diffusion and thermal diffusion can be com-
puted through a method of successive approximations developed by Chapman and
Enskog.
For this, the form of the interaction potential between molecules as well as its
corresponding parameters must be known. Generally, a rst approximation is enough
for practical applications. For D12 rst approximation [D12 ]1 , gives
( )
3 kT M1 + M2 1
[D12 ]1 = 2 RT . (2.15)
p
8 2 12 M M
1 2
(1,1)
12 (T12 )

Here T12 is a dimensionless temperature dened by the expression
kT
T12 = , (2.16)
12
9 See Ref. [2], p. 597.
(1,1)
and 12 is a function of T12 whose form depends on the potential of molecular
(1,1)
interaction. For instance, if the molecules behave as rigid spheres 12 = 1 for all

values of T12 .
For numerical computations it is more convenient to express [D12 ]1 in the form

M1 + M2 3
T
2M1 M2
[D12 ]1 = 0.002628 , (2.17)
2
p12
(1,1) )
(T12
12
o
where p is measured in atm, T in K, 12 in A and [D12 ]1 in cm2 /s.
The above mentioned theory does not determine the interaction potential that
should be applied in each case. To the contrary a comparison between the theoret-
ical results corresponding to several potentials and experimental measurements, can
furnish information on such potentials. In Ref. [2], pp. 589 and following, some ex-
amples of such comparisons can be found showing the agreement that can be expected
between theoretical and experimental results. It is veried that for many cases the best
agreement is attained with a Sutherland interaction potential or with that of Lennard-
Jones. Sutherlands potential is the one that produces between rigid spheres attracting
one another with a force inversely proportional to a given power of its distance r.
Therefore, it has the form
r < 12 : (r) = ,
( ) (2.18)
12
r > 12 : (r) = 12 .
r
(1,1)
This potential gives for 12
(1,1) S12
12 =1+ , (2.19)
T
12
where S12 = g() is a constant whose value depends on the parameters of Eq. (2.18)
k
and generally is determined experimentally. Substituting Eq. (2.19) into (2.15) one
obtains
( ) 5
3 k M1 + M2 T2
[D12 ]1 = 2 R . (2.20)
p
8 2 12 M1 M2 S12 + T
If the interaction potential is of Lennard-Jones type, no simple expression can

(1,1)
be given for 12 , whose values must be calculated by numerical integration. Table
2.1 gives the values calculated by Hirschfelder et al. [7].
2.2. DIFFUSION 43
T (1,1) (2,2) T (1,1) (2,2) T (1,1) (2,2)

0.30 2.662 2.785 1.70 1.140 1.248 4.20 0.874 0.9600
0.35 2.476 2.628 1.75 1.128 1.234 4.30 0.8694 0.9553
0.40 2.318 2.492 1.80 1.116 1.221 4.40 0.8652 0.9507
0.45 2.184 2.368 1.85 1.105 1.209 4.50 0.8610 0.9464
0.50 2.066 2.257 1.90 1.094 1.197 4.60 0.8568 0.9422
0.55 1.966 2.156 1.95 1.084 1.186 4.70 0.8530 0.9382
0.60 1.877 2.065 2.00 1.075 1.175 4.80 0.8492 0.9343
0.65 1.798 1.982 2.10 1.057 1.156 4.90 0.8456 0.9305
0.70 1.729 1.908 2.20 1.041 1.138 5.00 0.8422 0.9269
0.75 1.667 1.841 2.30 1.026 1.122 6.00 0.8124 0.8963
0.80 1.612 1.780 2.40 1.012 1.107 7.00 0.7896 0.8727
0.85 1.562 1.725 2.50 0.9996 1.093 8.00 0.7712 0.8538
0.90 1.517 1.675 2.60 0.9878 1.081 9.00 0.7556 0.8379
0.95 1.476 1.629 2.70 0.9770 1.069 10.0 0.7424 0.8242
1.00 1.439 1.587 2.80 0.9672 1.058 20.0 0.664 0.7432
1.05 1.406 1.549 2.90 0.9576 1.048 30.0 0.6232 0.7005
1.10 1.375 1.514 3.00 0.9490 1.039 40.0 0.596 0.6718
1.15 1.346 1.482 3.10 0.9406 1.030 50.0 0.5756 0.6504
1.20 1.320 1.452 3.20 0.9328 1.022 60.0 0.5596 0.6335
1.25 1.296 1.424 3.30 0.9256 1.014 70.0 0.5464 0.6194
1.30 1.273 1.399 3.40 0.9186 1.007 80.0 0.5352 0.6076
1.35 1.253 1.375 3.50 0.9120 0.9999 90.0 0.5256 0.5973
1.40 1.233 1.353 3.60 0.9058 0.9932 100. 0.5170 0.5882
1.45 1.215 1.333 3.70 0.8998 0.9870 200. 0.4644 0.5320
1.50 1.198 1.314 3.80 0.8942 0.9811 300. 0.436 0.5016
1.55 1.182 1.296 3.90 0.8888 0.9755 400. 0.417 0.4811
1.60 1.167 1.279 4.00 0.8836 0.9700
1.65 1.153 1.264 4.10 0.8788 0.9649
Table 2.1: Values of (1,1) and (2,2) as a function of T for the Lennard-Jones potential.
When comparing theoretical values of the transport coefcients with those ex-
perimentally obtained, it is seen that in many cases Lennard-Jones potential gives the
best agreement for a more extensive range of temperatures.10
The preceding formulae require knowing the values for 12 and 12 , which
must be experimentally obtained. Generally, such values are known for potentials
between equal molecules but not for those of different kind. The approximate values of
the constants for the potentials between molecules of species A1 and A2 are obtainable
as follows.
10 See Ref. [2], p. 597.

Let (1 , 1 ) and (2 , 2 ) be the values of the constants for species A1 and A2 .

The values of the constants for mixed potentials between both species are11

12 = 1 2 ,
1 (2.21)
12 = (1 + 2 ) .
2
Table 1.1 in chapter 1 gives the values of and for certain number of gases. If such
values where not available any of the following formulas can be used for approximate
values [7]:
1
/k = 0.77Tc , = 0.710(Vc /N ) 3 ,
1
/k = 1.15Tb , = 0.984(Vb /N ) 3 ,
1
(2.22)
/k = 1.92Tm , = 1.031(Vm /N ) 3 ,
1
/k = 0.292TB , = 1.030(Vz /N ) 3 .
Here V is the molar volume. Subscripts c, m, b, B and z denote the critical, melting,
boiling, Boyles12 and absolute zero points.
Table 2.2, taken from Ref. [2], gives the diffusion coefcients for some bi-
nary mixtures. Theoretical values correspond to a Lennard-Jones interaction poten-
tial. Constants for the potential have been obtained through Eq. (2.21). It is seen that
agreement between theoretical and experimental values is, generally, excellent.
(1,1)
Except for rigid spheres where 12 is constant, this function decreases when
T is increased. Hence, according to formula (2.15) [D12 ]1 increases with T slightly
faster than T 3/2 . Fig. 2.3 represents as an example the coefcient [D12 ]1 versus T in
a mixture of N2 and CO2 for rigid spheres and for an interaction potential of Lennard-
Jones. In combustion problems, where temperature can change in a ratio of ten to one
from one point to another, the inuence of temperature on the values of the transport
coefcients is very important. Formula (2.15) shows, as well, that diffusion coefcient
is inversely proportional to the mixtures pressure and is independent from the mass
fractions of the species in rst approximation. In order to obtain the inuence of mass
fractions one must resort to higher approximations.
11 See Ref. [2], pp. 589 and following

12 Boyles temperature T is the temperature at which curve pV vs. p has a horizontal tangent for p = 0.
B
2.2. DIFFUSION 45
o
Gas Pair 12 (A) 12 /k (K) T (K) [D12 ]1 (cm2 s1 )
Calculated Experimental
273.2 0.656 0.674
N2 -H2 3.325 55.2 288.2 0.718 0.743
293.2 0.739 0.760
273.2 0.175 0.181
N2 -O2 3.557 102
293.2 0.199 0.220
N2 -CO 3.636 100 273.2 0.174 0.192
273.2 0.130 0.144
288.2 0.143 0.158
N2 -CO2 3.839 132
293.2 0.147 0.160
298.2 0.152 0.165
N2 -C2 H4 3.957 137 298.2 0.156 0.163
N2 -C2 H6 4.050 145 298.2 0.144 0.148
N2 -nC4 H10 4.339 194 298.2 0.0986 0.0960
N2 -iso-C4 H10 4.511 169 298.2 0.0970 0.0908
H2 -O2 3.201 61.4 273.2 0.689 0.697
H2 -CO 3.279 60.6 273.2 0.661 0.651
273.2 0.544 0.550
288.2 0.597 0.619
H2 -CO2 3.482 79.5
293.2 0.616 0.600
298.2 0.634 0.646
H2 -N2 O 3.424 85.6 273.2 0.552 0.535
H2 -SF6 3.922 89.1 298.2 0.473 0.420
273.2 0.607 0.625
H2 -CH4 3.425 67.4
298.2 0.705 0.726
H2 -C2 H4 3.600 82.6 298.2 0.595 0.602
H2 -C2 H6 3.693 87.5 298.2 0.556 0.537
CO-O2 3.512 112 273.2 0.175 0.185
273.2 0.128 0.139
CO2 -O2 3.715 147
293.2 0.146 0.160
CO2 -CO 3.793 145 273.2 0.128 0.137
CO2 -N2 O 3.938 204 273.2 0.092 0.096
CO2 -CH4 3.939 161 273.2 0.138 0.153
Table 2.2: Comparison of diffusion coefcients for some binary mixtures. Theoretical val-
ues correspond to Lennard-Jones interaction potential.
Approximation of order j can be expressed in the form13
j
[D12 ]j = [D12 ]1 fD 12
. (2.23)
2
For example, for the second approximation, fD 12
depends on the molar masses
of the species, on their mass fractions, viscosity coefcients and temperature of the
mixture. Its value is always close to one. For the Lennard-Jones potential, for instance,
it ranges from 1 to 1.03.
13 See Ref. [2], p. 539.
4.0
3.5
3.0
2.5
12 1
Rigid spheres
[D ]
2.0
1.5
LennardJones potential
1.0
0.5
0.0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
T(K)
Figure 2.3: Coefcient [D12 ]1 as a function of temperature in a mixture of N2 and CO2 , for
rigid spheres and for an interaction potential of Lennard-Jones.
Hence, the inuence of mass fractions on the value of D12 is always small and
can, generally, be neglected by adopting for it the value given by the rst approxi-
mation. Frequently, the value given by elementary diffusion theory is approximate
enough. According to it,
T 3/2
D12 . (2.24)
p
Or else, an empirical formula of the form
Tn
D12 , (2.25)
p
where n is an exponent ranging from 1.5 to 2 whose value is determined empirically.
With respect to thermal diffusion ratio kT , even in the rst approximation it is

a complicated function of the molar masses, mass fractions of the species and of the
temperature of the mixture. It is also very much inuenced by the laws of molecular
interaction. The expression of the rst approximation of kT can be obtained from
Ref. [2], p. 541, where tables for the calculation of their values are also included as
well as a comparison between theoretical and experimental values for many binary
mixtures. It appears that the agreement is not so good for kT as for the other transport
coefcients. When comparing the successive approximations to the value of kT , it is
also veried that in rst approximation the error is larger here than for other transport
2.2. DIFFUSION 47
coefcients.14 The values of kT are generally small. Normally they do not exceed 0.1.
Thereby thermal diffusion is of secondary importance in Aerothermochemistry and its
inuence is, usually, disregarded.
In Refs. [8], [9], [10] and [11] recent information on theoretical and experi-
mental values of transport coefcients can be found.
For mixtures containing more than two components, the problem has been
solved by Curtiss and Hirschfelder [2], [12]. In such case, the aforementioned three
causes of diffusion, namely, differences in composition, pressure and temperature, still
hold. Yet in this case diffusion ux of each species depends not only on the gradient
of its molar fraction but is a linear function of the gradients of all molar fractions.
Diffusion ux fi of species Ai (i = 1, 2, . . . , `) is
( )
Mi 0 p T
fi = Yi vdi = 2 Mj Dij Xj + (Xj Yj ) DT i . (2.26)
Mm p T
j6=i
0
Here Dij is the diffusion coefcient of species Ai and Aj in the mixture. For a binary
0
mixture, Dij is diffusion coefcient Dij previously dened. For the case of more than
two components it differs. DT i is the thermal diffusion coefcient of species Ai in
the mixture.
The ` equations (2.26) are not independent from each other. In fact, diffusion
uxes of ` species must satisfy the condition

fi = 0. (2.27)
i
0
Diffusion coefcients Dij in system (2.26) can be expressed as a function of
binary diffusion coefcients Dij of the species taken in couples. Ref. [2], pp. 541
and 543, and Ref. [12] contain expressions for these coefcients as well as for those
0
of thermal diffusion. Expressions for Dij and DT i are too complicated for practical
0
computations. Then it is advantageous to eliminate Dij by substituting system (2.26)
with the following that makes explicit the inuence of binary diffusion coefcients15
Xi Xj p
(vdj vdi ) =Xi + (Xi Yi )
Dij p
j6=i
(
Xi Xj DT j ) (2.28)
DT i T
.
Dij Yj Yi T
j6=i
As was done for the case of binary diffusion, it is also convenient to elimi-
nate molar fractions of the species by introducing their corresponding mass fractions.
14 A critical study on the subject will be found in Ref. [6], p. 492 and following, as well as bibliography.
15 See Ref. [2]. p. 517.
Making use of relations (1.33) and (1.36) in chapter 1, it is obtained
Yj [ vdi vdj Yi Yj Mj Mi p
+ +
Mj Dij Yi Yj Mm p
j6=i
( ) ] (2.29)
1 DT j DT i T
= 0, (i = 1, 2, . . . , `).
Yj Yi T
Diffusion of a species whose mass fraction is very small is particularly impor-

tant in certain problems. For example, diffusion of active particles which propagate
chemical reactions through a combustion wave. Then it becomes possible to give an
explicit approximate expression for the diffusion ux of such species. For diffusion
arising from differences in composition this expression is
( )
Yj Yj Yi

j6=i Mj Yj Yi
fi = Yi vdi = Yi Yj . (2.30)
j6=i Mj Dij
System (2.29) is, generally, too complicated to be used in the solution of many
of the Aerothermochemistry problems. Therefore, in some of its applications, diffu-
sion uxes of the species are substituted by approximate expressions under the as-
sumption that diffusion ux due to differences in composition fiY of each species is
proportional to its gradient of molar fraction. That is, by adopting for such ux an
expression of the form
Mi
fiY = Dim Xi . (2.31)
Mm
Here Dim is a binary diffusion coefcient of the species through the mixture formed
by the rest. This expression is exact only when binary diffusion coefcients and molar
masses of the species are all equal. If mean molar mass of the mixture is also constant,
one has
fiY = Dim Yi , (2.32)
by virtue of Eqs. (1.33) and (1.36) in chapter 1. This expressions will be used in
chapter 13. In aerothermochemical problems diffusion uxes arising from differences
in pressure and temperature are, in general, negligible. C. R. Wilke [14] has given an
explicit approximate expression, empirically deduced, similar to (2.31) for uxes due
to differences in composition.
2.3. VISCOSITIES 49
2.3 Viscosities
Mechanics of continuous media teaches that the state of stresses at each point of a
medium is determined by a symmetric tensor of second order

p11 p12 p13

e p21 p22 p23 , (2.33)
p31 p32 p33
of components pij = pji in a rectangular cartesian system of reference.16 The physical

meaning of this tensor is the following: pij is the component parallel to axis xj of the
stress acting upon a surface element normal to axis xi , see Fig. 2.4.
x3
p23
p13
p
21
p11 p22
x
p12 p33 1
p31
x2
p32
Figure 2.4: Schematic diagram showing the components of the stress tensor e .
Stress f acting upon a surface element d, see Fig. 2.5, whose orientation is
dened by the unit vector n, is given by the expression17
f = n e , (2.34)
whose components are

fi = nj pji , (i = 1, 2, 3), (2.35)
j
where nj are the components of n.
Kinetic Theory of gases shows18 that in a dilute gas the stress originates from
transfer of momentum of the gas through a surface element d moving at velocity v of
16 See, i.e., Ref. [15].
17 See appendix to chapter 3 for notation.
18 See Ref. [1], p. 31.
d
f f3
x3
n
P f1
x1
f2
x2
Figure 2.5: Schematic diagram showing the components of the stress f = n e .
the gas.19 Such transfer is produced by the thermal agitation of the molecules.20 If the
state of gas is uniform, the velocity distribution of thermal agitation is a Maxwells
distribution. In such case transfer of momentum is normal to d and independent
from the orientation n. The corresponding state of stresses is a pure compression,
dened by a scalar: the pressure p of the gas. Tensor of Eq. (2.33) reduces as follows
e = pU, (2.36)
where U is the unit tensor whose components are Kroneckers ij

1 0 0

U 0 1 0 . (2.37)
0 0 1
Stress f acting upon an element d in Fig. 2.5, is
f = pn. (2.38)
When the state of motion of the gas is not uniform other stresses produce, which add
to those of pressure, Eq. (2.38). Such stresses are given by viscous stress tensor

11 12 13

ev 21 22 23 . (2.39)
31 32 33
The Chapman-Enskog method described in the introduction to the present chapter
shows that components ij of the viscous stress tensor have the form21
( )
2 0
ij = 2ij ( v) ij . (2.40)
3
19 Within dense gases stresses are partially due to transfer of momentum and partially to the forces of
molecular interaction whose effect is not negligible.

20 If v is the gas velocity at a point and v the velocity of a molecule at the neighborhood of such point,
j
velocity va of thermal agitation is, by denition, va = vj v.
21 Actually Enskog and Chapmans theory was developed for monotonic dilute gases, where 0 = 0.
2.3. VISCOSITIES 51
Here and 0 are called viscosity coefcient and bulk viscosity coefcient respec-
tively, and
( )
3 1 vi vj
ij = + (2.41)
8 2 2 xj xi
are the components of deformation velocity tensor

11 12 13

vd 21 22 23 , (2.42)
31 32 33
which determines the deformation suffered by a gas element as it moves.22
The method of Chapman and Enskog allows calculation of the viscosity coef-
cient of a gas by means of successive approximations. Expressions of the form
j = []1 f j (2.43)
are obtained, where []1 is the rst approximation, j is the computed order of the ap-
proximation and f j is a function which differs slightly from unity. When comparing,
for example, the rst and third approximations for the Lennard-Jones potential it is
veried that error in the rst is smaller than 0.8%. Thereby, as for diffusion coef-
cients, the rst approximation is usually enough.
In the case of a pure gas this approximation is given by23

5 N 1 M RT
[]1 = 2 (2,2) . (2.44)
16 (T )
Here, as for diffusion, (2,2) is a function of reduced temperature T = kT /, whose

law of variation depends on the form of the molecular interaction potential. For ex-
ample, for rigid spheres (2,2) = 1.
For a Sutherland potential, Eq. (2.18), is
S
(2,2) = 1 + , (2.45)
T

where S = g () . In practice, this value is determined experimentally. When
k
Eq. (2.45) is substituted into Eq. (2.44) one obtains

5 N 1 M R T 2
3
[]1 = . (2.46)
16 2 S + T
22 See Ref. [15].
23 See Ref. [2], p. 527.
Values (2,2) as function of T for the Lennard-Jones potential have been

taken into Table 2.1. It is seen that for many gases this potential represents experi-
mental results better than others.24
For numerical computations it is more convenient to express Eq. (2.44) as

7 MT
10 []1 = 266.93 2 (2,2) , (2.47)
(T )
where all quantities are measured in the same units than in Eq. (2.17) and []1 is
expressed in gr cm1 s1 .
The preceding formulae show that the viscosity coefcient of a gas is indepen-

dent from pressure and increases with temperature slightly over T . Fig. 2.6 gives
the law of variation with temperature of viscosity coefcients for several gases. For
additional information, see Refs. [2], [8], [9], [11] and [16], where complementary
bibliography is listed.
7000
6000 O2
N2
5000
s )
1 1
4000
10 (g cm
CH4
3000
7
CO2
2000
H
2
1000
0
0 200 400 600 800 1000 1200 1400 1600
T(K)
Figure 2.6: Viscosity coefcient as function of temperature, for several gases.
Often, for practical applications an empirical formula of the form
Tn (2.48)
is used, where n is an exponent ranging from 0.5 to 1.
With respect to bulk viscosity coefcient 0 it is zero for dilute monatomic

gases. For polyatomic and dense gases 0 is generally different from zero, yet small.
24 See Ref. [2], pp. 589 and following.
2.3. VISCOSITIES 53
Its value depends on the easiness of energy transfer between external and internal
degrees of freedom of the molecules during collisions. Such facility is characterized
by a relaxation time for each internal degree of freedom. The value for 0 is expressed
as a function of these relaxation times. Generally, it is sufciently approximate to
assume 0 = 0. Additional information on this problem can be found in Ref. [2],
pp. 501 and 710.
Viscosity coefcient of a mixture of gases is a complicated function of molar

masses, mass fractions and viscosity coefcients of the species as well as of the in-
teraction potentials between different species.25 For binary mixture, for example, the
rst approximation []1 is given by
1 X + Y
= , (2.49)
[]1 1 + Z
where
X12 2X1 X2 X2
X = + + 2,
1 2
[ 12 ]
2
3 X12 M1 2X1 X2 (M1 + M2 ) 212 X22 M2
Y = A12 + + ,
5 1 M2 12 4M1 M2 1 2 2 M 1
[ (2.50)
3 X12 M1
Z = A12
5 M2
( ) ]
2 ( )
(M1 + M2 ) 12 12 X22 M2
+ 2X1 X2 + 1 + .
4M1 M2 1 2 M1
In these expressions, 1 and 2 are the rst approximations of the viscosity coef-
cients for both species. 12 is given by
( )
1 M1 M2
N 2RT
5 M1 + M2
12 = , (2.51)
16 2
12
(2,2) )
(T12
12
12 and 12 are the constants of the interaction potential between molecules of both
kT
species. A is a function of reduced temperature T12

= , whose value is given in
12
Table 2.3 for the Lennard-Jones potential.

T12 0.3 0.5 0.75 1 1.25 1.5 2
A 1.046 1.093 1.105 1.103 1.099 1.097 1.094

T12 3 4 5 10 50 100 400
A 1.095 1.098 1.101 1.11 1.13 1.138 1.154
Table 2.3: The coefcient A as a function of T12

.
25 See Ref. [2], p. 529.
The following Table 2.4 gives, as an example, the values of the theoretical and
experimental viscosity coefcients for a mixture of oxygen and carbon monoxide.
%O
T (K) 0 42.01 77.33
Experimental 1 776 1 900 1 998
300.06
Calculated 1 771 1 896 2 003
Experimental 2 548 2 741 2 908
500.06
Calculated 2 539 2 717 2 871
Table 2.4: Viscosity coefcient (107 gr cm1 s1 ) for a mixture of O2 -CO.
For mixtures of more than two components Buddemberg and Wilke give the
following semi-empirical expression which represents experimental results with good
approximation
X12
[]1 = . (2.52)
Xi2 RT
i + 1.385 Xi Xj
[i ]1 j6=i pMi [Dij ]1
This expression allows calculation of the viscosity coefcient of a mixture when those
corresponding to the species are known as well as the diffusion coefcients of the
same when taken by couples.
In the preceding formulae substitution of molar fractions by mass fractions is

straightforward by making use of relation
Mm
Xi = Yi . (2.53)
Mi
Fig. 2.7 gives the law of variation as a function of the composition of the viscosity
coefcients for some gas mixtures.
2.4 Thermal conductivity
Theory of heat conduction shows26 that heat transfer at a point of a material medium
is dened by a vector q of heat ux
q = q1 i1 + q2 i2 + q3 i3 . (2.54)
Physical meaning of q is the following: qi is heat ux per unit surface and per unit
time through a surface element normal to axis xi .
26 See, i.e., Ref. [16], pp. 3 and following.
2.4. THERMAL CONDUCTIVITY 55
2000
1800
H2 O2
H2 CO
107 (gr cm1 s1)
1600 H N
2 2
H CO
2 2
1400
H N O
2 2
1200
1000
800
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
XH
2
Figure 2.7: Viscosity for some gas mixture as function of composition.
Considering a surface element d, see Fig. 2.5, not parallel to any of the co-
ordinate planes and whose orientation is dened by unit vector n, it is easily demon-
strated that heat ux Q through d per unit surface and per unit time is

Q = q n = qi ni . (2.55)
i
Therefore, once the ux vector at a point is known, this determines heat transfer
through any surface element passing through it.
Kinetic Theory of Gases shows that within a dilute gas heat ux originates only
from molecular energy transfer when d moves at a velocity v of the gas at the point.
Such transfer originates from the motion of molecular agitation and from diffusion
between species forming the gas if it is a mixture. When such transfer is calculated
one obtain the expression27
Yj DT i
q = T + Yi hi vdi + Mm RT (vdi vdj ) , (2.56)
i i j
Mi Mj Dij
where is the thermal conductivity of the gas and hi is the total specic enthalpy
of species Ai . Since diffusion uxes of the species are produced by differences in
composition, pressure and temperature, it results according to (2.56) that this three
causes also produce heat transfer. In general, contributions from the third term in the
27 See Ref. [2], p. 498.
right hand side of (2.56) is small compared to the other two and can be neglected.
Then q reduces to

q = T + Yi hi vdi . (2.57)
i
In a pure gas q reduces to
q = T. (2.58)
If, moreover, the gas is monatomic, thermal conductivity coefcient can be

computed through successive approximations as done for diffusion and viscosity. The
rst approximation []1 is

T /M
107 [] = 1989.1 . (2.59)
2 (2,2) (T )
Here [] is measured in cal cm1 s1 K1 and the other magnitudes in the same units
as in Eq. (2.17).
When comparing (2.59) and (2.47) it is seen that for monatomic gases thermal
conductivity and viscosity coefcients are proportional. Similarly to what happens for
other transport coefcients the rst approximation is, generally, sufcient for practical
applications.
Expression (2.59) does not take into account for polyatomic molecules energy
transfer between internal and external degrees of freedom during collisions. Such
transfer is not important in the study of viscosity and diffusion but is fundamental in
thermal conductivity.28 In polyatomic molecules [] must be substituted by the follow-
ing approximate expressions, due to Eucken, which takes into account the inuence
of such energy transfer
( )
15 R 4 Cv 3
[] = []1 + , (2.60)
4 M 15 R 5
where Cv is molar heat of the gas at constant volume. Table 2.5 compares experimen-
tal values with those given by Euckens formula (2.60) for several gases.29
H2 O2 CO2 CH4 N2
Calculated 4 140 615 386 741 619
Experimental 4 227 635 398 819 619
Table 2.5: Values of the thermal conductivity coefcient (107 [] cal cm1 s1 K1 ) for
several gases at T=300 K .
It is seen that agreement is fair even if not as good as for other transport coef-
cients.
28 See Ref. [2], p. 489.
29 See Ref. [2], p. 574.
2.4. THERMAL CONDUCTIVITY 57
4000
3500
K )
3000
1
1 1
2500
s
H2 O2
10 (cal cm
2000 H N
2 2
H2 CO
1500
7
1000 H2 N2O
H2 CO2
500
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
X
H
2
Figure 2.8: Thermal conductivity for some gas mixture as function of composition.
In Ref. [2], pp. 526 and following, theoretical formulae are given for the cal-
culation of thermal conductivity coefcients in mixtures of monatomic gases as well
as empirical formulae for polyatomic gases. Here agreement between theoretical and
experimental results is also not so good as for other transport coefcients. Fig. 2.8
gives, as an example, the inuence of the composition on the value of for certain
binary mixtures at temperature 273.16 K.
Finally, it should be noted that Aerothermochemistry development is not yet

sufciently advanced so that transport coefcients are needed to be exact in most of its
applications. Frequently a rudimentary approximation is enough, usually empirical,
where the inuence of composition is neglected and that of temperature is approx-
imately taken into account. These approximations appear justied considering that
other features of the problems must be accounted for in a very rudimentary way, such
as frequently are chemical kinetics, and, in some cases, velocity eld and shape of the
ame, etc.
References
[1] Chapman, S. and Cowling, T. G.: The Mathematical Theory on Non-Uniform
Gases. Cambridge University Press, 1939.
[2] Hirschfelder, J. O., Curtiss, C. F., and Bird, R.B.: Molecular Theory of Gases
and Liquids. John Wiley and Sons, Inc., New York, 1954.
[3] Enskog, D.: Kinetische Theorie der Vorgange in massig verdunnten Gasen. Dis-
sertation, Upsala, 1917.
[4] Chapman, S.: On the Kinetic Theory of a Gas, Part II, A Composite Monatomic
Gas: Diffusion, Viscosity and Thermal Conduction. Philosophical Transactions
of the Royal Society of London, A-217, 1917, p. 115.
[5] Jeans, J.: An Introduction to the Kinetic Theory of Gases. Cambridge University
Press, 1948.
[6] Jost, W.: Diffusion in Solids, Liquids and Gases. Academic Press, Inc., Pub-
lishers, New York, 1952.
[7] Bird, R. B., Hirschfelder, J. O. and Curtiss, C. F.: The Theoretical Calculation
of the Equation of State and Transport Properties of Gases and Liquids. The
American Society of Mechanical Engineers, Heat Transfer Division, Annual
Meeting, New York. 29 Nov. to 4 Dec., 1953.
[8] Johnson, E. F.: Molecular Transport Properties of Fluids. Industrial and Engi-
neering Chemistry, Vol. 45, 1953, pp. 902-6.
[9] Baron, T. and Oppenheim, A. K.: Fluid Dynamics. Industrial and Engineering
Chemistry, Vol. 45, 1953, pp. 941-51.
[10] Pigford, R. L.: Mass Transfer. Industrial and Engineering Chemistry, Vol. 45,
1953, pp. 951-56.
[11] Hirschfelder, J. O., Bird, R. B. and Sportz, E. L.: Viscosity and Other Physical
Properties of Gases and Gas Mixtures. ASME, 1949, pp. 921-37.
[12] Hirschfelder, J. O., Bird, R. B. and Sportz, E. L.: The Transport Properties
for non Polar Gases. 1.- Journal of Chemical Physics, 1948, pp. 968-81.
2.- Chemical Review, 1949, pp. 205-31.
[13] Curtiss, C. F. and Hirschfelder, J. O.: Transport Properties of Multicomponent
Gas Mixtures. The Journal of Chemical Physics, Vol. 17, June 1949, pp. 550-
55.
[14] Fairbanks, D. F. and Wilke, C. R.: Diffusion Coefcients in Multicomponent
Gas Mixtures. Industrial and Engineering Chemistry, March 1950, pp. 471-75.
[15] Sommerfeld, A.: Mechanics of Deformable Bodies. Academic Press Inc., Pub-
lishers, 1950.
[16] Carslaw, H. S. and Jaeger, J. C.: Conduction of Heat in Solids. Oxford at the
Clarendon Press, 1950.
Chapter 3
General equations
3.1 Introduction
In the present chapter we shall establish the general equations of Aerothermochem-

istry, that is, the general equations of transformation of a mixture of gases that react
one to another and are in motion. It is assumed herein that the gas can assimilate to
a continuous medium as previously done in the preceding chapter when analyzing the
transport coefcients. This assumption is well justied except in the following cases:
a) When a characteristic length of the macroscopic scale of the process, for example
the size of an obstacle submerged in the gas, is of the order of magnitude of the
molecular mean free path.
b) In the case where the phenomenological variables suffer considerable variations
within a distance of the order of magnitude of the said mean free path.
The rst case occurs when dealing with a very rareed gas (Knudsen gas). The
second case occurs, for example, in the analysis of the structure of a shock wave. In
fact, the thickness of a sock wave is only a few times larger than the molecular free
path, except when the wave is very weak. In such cases, to speak of the macroscopic
functions of a point has no sense, and it is necessary to take into consideration the
molecular structure of the gas. Such cases are excluded from the present study.
The equations that govern the transformations of a mixture of reacting gases are
similar to the general Gas Dynamics equations. Their difference is due to the fact that
while Gas Dynamics deal with gases of homogeneous chemical composition invari-
able with time, in Aerothermochemistry the composition of the mixture varies. Such
variation is due to chemical reactions taking place between the various species that
form the mixture and to their mutual diffusion, and has the following consequences:
59
60 CHAPTER 3. GENERAL EQUATIONS
1) The need to establish a separate balance for each one of the species by means of
partial continuity equations for each different species. In each of these equations
the effects of the chemical reactions and of diffusion must be included.
2) The modications due to diffusion of some of the transport coefcients of the
mixture. As previously seen in chapter 2, diffusion does not change the values
of the viscosity coefcients but modies the heat ux, mainly due to the transport
of the enthalpy of the various species through diffusion.
3) The presence of a new form of energy: the chemical energy released or absorbed
in the chemical reactions.
n q
F
Figure 3.1: Schematic diagram of the actions upon a isolated gas element.
In Gas Dynamics the equations are established by applying the laws of the con-
servation of mass, momentum and energy to an ideally isolated gas element following
its motion.1 The material element thus isolated (Fig. 3.1) is bounded by a uid sur-
face, that is to say by a surface , in which the velocity at each point is the velocity
v of the gas particle at the same point. This element evolves in accordance with the
aforementioned laws of conservation and under the action of the surrounding mass,
which acts upon its boundary . This action result in the surface forces and in the
heat transport q throughout each of its surface elements. Furthermore, the action of
the possible force elds acting upon the element must be included; for example, the
action F of the gravity eld. In this manner2 we obtain:
a) The continuity equation, which expresses the laws of the conservation of mass.
b) The equation of motion which expresses that Newtons Second Law of Mechan-
ics is satised.
c) The energy equation which expresses that the First Law of Thermodynamics is
satised.
1 See,
i.e., Ref. [1], pp. 337 and following.
2 Thesame equations could be established by starting from the molecular structure of the gas and apply-
ing the laws of the Kinetic Theory of gases. See, i.e., Ref. [2], p. 25.
3.2. EQUATION OF CONTINUITY 61
The system of equations thus obtained must be completed with the state equa-
tion of the gas and with the law of variation of the viscosity coefcient and of the
remaining thermodynamic functions that deal with the problem (internal energy or
enthalpy, etc.) as a function of the gas state variables.
In order to establish the general equations of Aerothermochemistry, a simi-

lar procedure can be followed, taking into account the three observations previously
stated.3 In such a case it is to be understood that when speaking of the material ele-
ment we refer to an element bounded by a surface which moves with a velocity that at
each point concurs with the mixture velocity. This specication is due to the need of
distinguishing between the velocity of the mixture and the velocities of the different
species in cases where diffusion exists. These velocities are not equal (see chapter 2).
In the following, the equations thus obtained are given under the same notation used
in the preceding chapters.
Vectorial and tensorial notation is preferably used as it enables the concise

writing of the equations and a simple interpretation of their terms. A summary of the
symbols and formulae can be found in the Appendix.
3.2 Equation of continuity
Consider species Ai , for which the partial density at a point is i . The variations of
Di
i , following the motion of an element of the mixture, is , where the operator
Dt
D () ()
= + v () (3.1)
Dt t
is the substantial derivative of the motion of the mixture.
The variation of i is due:
a) To the expansion v of the volume element, which decreases i in i v.

b) To the ux by diffusion of species Ai , through the boundary of the element. If
vdi is the diffusion velocity of species Ai diffusion reduces i in (i vdi ).
c) To the mass of species Ai produced by chemical reactions. If wi is the reaction
rate of species Ai 4 under the conditions of state and composition that prevail at
the point, then the mass produced per unit volume and per unit time is wi .
3 The general equations for Aerothermochemistry may also be deduced from the molecular structure of
the mixture by applying the laws of Kinetic Theory.

4 See chapter 1.
Therefore, the continuity equation for species Ai is

Di
= i v (i vdi ) + wi . (3.2)
Dt
If there are ` different species Ai , there is an equation similar to Eq. (3.2) for
each of them.5
Adding the ` equations corresponding to the ` different species and making use
of the obvious relation

i = (3.3)
i
and of

wi = 0, (3.4)
i
which expresses the condition that the chemical reactions do not change the mass, and
of

i vdi = 0, (3.5)
i
which was deduced in chapter 2, the following expression is obtained

D
+ v = 0, (3.6)
Dt
which is the continuity equation for the mixture and is identical to the continuity
equation of Gas Dynamics.
Equation (3.6) allows simplication in the form of (3.2). For this purpose it is
convenient to express Eq. (3.2) as a function of the mass fraction Yi = i / of species
5 Thedetailed deduction of Eq. (3.2) is as follows.
The mass mi of species Ai contained in volume V is
ZZZ
mi = i dV.
V
Its time variation is (see Appendix)

ZZZ ZZZ ZZ ZZZ
dmi d i i
= i dV = dV + i (v n) d = + (i v) dV.
dt dt V V t V t
The mass diffusing from V through per unit time, is
Z ZZZ
i vdi n d = (i vdi ) dV.
V
The mass production in V by the chemical reaction, per unit time, is

ZZZ
wi dV.
V
Then when establishing the balance and expressing that this condition must be satised for all V , we obtain
Eq. (3.2).
In all the other formulae that follow the argumentation is identical, and for this reason the expansion of
the complete calculation is not considered necessary.
3.2. EQUATION OF CONTINUITY 63
Ai . By doing so and taking into account Eq. (3.6) the following simplied expression
is obtained, which replaces Eq. (3.2)
DYi
+ (Yi vdi ) = wi , (i = 1, 2, . . . , `). (3.7)
Dt
Aside from the density and velocity v of the mixture, the latter implicitly con-
tained in the substantial derivative, this system shows the mass fractions Yi of the
different species as well as their corresponding diffusion velocities vdi and reaction
rates wi .
As seen in chapter 2, diffusion velocities are linear functions of the gradients

of Yi , p and T . Therein it was shown that the diffusion velocities v0 di (i = 1, 2, . . . , `)
corresponding to the gradient of Yi and p are given by the system
Yi ( v0 di v0 dj Yi Yj Mj Mi p
)
+ + = 0,
Mj Dij Yi Yj Mm p
j6=i
(3.8)
Yj v0 dj = 0,
j
where
1 Yj
= (3.9)
Mm j
Mj
is the mean mass of one mole of the mixture (see chapter 1).
Similarly it was shown that the diffusion velocities v00 di corresponding to the
temperature gradient are given by expressions of the form
DT i T
v00 di = , (3.10)
Yi T
where DT i is the thermal diffusion coefcient of species Ai .
With respect to the reaction rate wi , it should be noted that the conclusions
in chapter 1 were established under the assumption that both the composition and
the state of the mixture are homogeneous throughout the mass. Therefore, when the
composition and state of the mixture are not homogeneous, the applicability of said
conclusions depends on the assumption that neither one varies appreciably within a
distance of the order of magnitude of a characteristic length of the chemical trans-
formations. This distance could be, for example, the average length of a chain when
chain reactions are produced. Within such a limitation the conclusions obtained in
chapter 1 can be applied here. In particular, if there are r different chemical reactions,
we have seen in the aforementioned chapter that wi is a linear combination of the r
reaction rates wij corresponding to said chemical reactions

wi = Mi ij rj . (3.11)
j
3.3 Equations of motion
The equations of motion are obtained by applying Newtons Second Law of Mechan-
ics, to the material element in Fig. 3.1.
Let V be the volume of said element. Then V will be the mass and V v the
momentum, for which the time variation when following the element is
D (V v) Dv
= V , (3.12)
Dt Dt
since V does not vary. Therefore, the variation of momentum of the mass contained
in the unit volume is (Dv/Dt). Such variation originates from:
1) The surface forces acting upon the unit volume.

2) The mass forces acting upon the unit volume.
Let us calculate both, separately.
Let f be the force that acts upon the surface element d at a point of , Fig. 3.1.
It has been shown in chapter 1 that f is determined by the stress tensor e of compo-
nents ij and that as a function of the tensor it can be expressed as follows
f = n e d, (3.13)
where n is the outward normal to the surface element d at the point. As it can easily
be proved, the resultant of all forces f per unit volume is e .6
Let F be the eld intensity of the mass forces, that is to say, the force per unit
7
mass. The force per unit volume is, evidently, F .
Now, by expressing Newtons Second Law of Mechanics, we obtain
Dv
= e + F . (3.14)
Dt
This vectorial equation is equivalent to three scalar equations corresponding to

the three components of the reference system. For example, in rectangular cartesian
coordinates, the component of this equation parallel to the coordinate axis xi (i =
1, 2, 3) is
vi vi ij
+ vj = + Fi , (3.15)
t xj xj
RR
6 Infact, the resultant
RR of the forces
RRRacting upon is n e d. Now, by applying Ostrogradskys
theorem we obtain n e d = V e dV. Therefore, the force per unit volume is e .
7 If the force per unit mass depends on the species and F is the intensity for species A , F must be
P P i i
substituted by i Fi = Yi Fi .
i i
3.4. ENERGY EQUATION 65
where the subscripts indicate components, and the repeated subscripts indicate, ac-
cording to Einsteins rule, summation respect to them, from one to three.
As shown in chapter 2, the components ij of the stress tensor are expressed as

a function of the gas pressure p at the point, and of the velocity v of the motion, in the
following form
( ) ( )
vi vj 0 2
ij = pij + + + ( v) ij , (3.16)
xj xi 3
where ij is the Kronecker symbol and and 0 are the coefcients of viscosity and
volumetric viscosity of the mixture, respectively. These coefcients depend on the
state and composition of the mixture as indicated in the above mentioned chapter. In
particular, for the diluted monatomic gases 0 = 0 and in general, a good approxima-
tion is obtained by assuming that it is also zero for all other cases.
In Eq. (3.14) the pressure and viscous stresses can be made explicit, by using
the expression
e = pU + ev (3.17)
previously given in chapter 2. By taking this expression of e into Eq. (3.14), we

obtain
Dv
= p + ev + F . (3.18)
Dt
This equation is identical to those obtained in Gas Dynamics for mixtures of uni-
form composition with no chemical reaction. The variations in the composition of the
mixture act only through their inuence on the values of the viscosity coefcients
and 0 .
3.4 Energy equation
This equation is obtained when expressing that the variations of the energy contained
in V , Fig.3.1, are due to the work done by the forces acting upon the element and the
heat received by the said element. Let us calculate, separately, the different terms in
this equation.
a) Energy.
Let ui be the internal energy per unit mass of species Ai . The internal energy u
per unit mass of the mixture is

u= Yi ui , (3.19)
i
and the internal energy of the mass contained in V is V u. Similarly, the kinetic
energy of the mass contained in V is V 12 v 2 . Therefore, the total energy8 of the
( )
mass contained in V is V u + 12 v 2 and its time variation when following the
motion of V is
( )
D 1 2 D 1
V (u + v ) = V (u + v 2 ). (3.20)
Dt 2 Dt 2
Consequently, the time variation of the energy of the mass contained in the unit
volume is
D 1
(u + v 2 ). (3.21)
Dt 2
b) Work.
The work done by f per unit time is
v f = v (n e ) = n (v e ) , (3.22)
and the work per unit time done by the surface forces that act upon the unit
volume is (v e ).9 The work per unit time done by the mass forces that act
upon the unit volume is F v. Consequently, the work per unit time done by all
the forces that act upon the unit volume is
(v e ) + F v. (3.23)
c) Heat.
The heat received by the element V through d per unit time is q n d, where
q is the vector of the heat ux dened in chapter 2. Therefore, the heat per unit
time received by the unit volume is
q. (3.24)
In the phenomena normally studied by the Aerothermochemistry the radiation

effects are of secondary importance. Furthermore, great difculty is generally
encountered for the study of the inuence of these effects. Nevertheless, if they
are to be taken into consideration, the vector qr of the radiation ux must be
added to the vector q of the heat ux. Information regarding the radiation ux
can be found in the work of Hirschfelder, Curtiss and Bird, mentioned in Ref.
[4], pp. 720 and following.
8 Other forms of energy different from the internal and the kinetic energies, as for example the radiation
energy, are not considered in the present study.

9 It can be proved in the same manner as for the calculation of the resultant of the surface forces.
The energy equation is now obtained by expressing that the energy variation,
given by Eq. (3.21), must be equal to the work done by the exterior forces, given by
Eq. (3.23), plus the heat received, given by Eq. (3.24), thus obtaining
D 1
(u + v 2 ) = (v e ) + F v q. (3.25)
Dt 2
Expanding the rst term of the right hand side of this expression, there results
(v e ) = v ( e ) + e : v , (3.26)
where v is the deformation velocity tensor, of components ij 10 dened by

( )
1 vi vj
ij = + . (3.27)
2 xj xi
In expression (3.26), the rst term of the right hand side measures the work
done by the surface forces that act upon the unit volume when the surface moves with
no deformation with velocity v. The second term measures the work done by the
surface forces in order to produce the deformation v .
Bringing forth the pressure and the viscous stresses in e by using expression
(3.17) for e , it can be veried that the deformation work (e : v ) consists of the
pressure work (p v) done during the expansion of the gas, and of the work
= ev : v (3.28)
dissipated by the viscosity to produce the deformation v . Function , which as it can

be easily proved is always positive, is the dissipation function of Lord Rayleigh. That
is
e : v = p v + . (3.29)
In order to bring forth the variation of the internal energy u, the energy equation
(3.25) can be simplied by making use of the equation of motion (3.14). In fact, by
subtracting from Eq. (3.25) the scalar product of Eq. (3.14) by v, taking into account
Eq. (3.26), there results
Du
= e : v q, (3.30)
Dt
or else, making use of Eq. (3.29),
Du
= p v + q. (3.31)
Dt
10 See chapter 2.
It has been shown in chapter 2 that q is given by the expression

Yj DT i
(vdi vdj )
i j Mi Mj Dij
q = T + Yi hi vdi + RT Yi . (3.32)
i
j Mj
In this expression the rst term of the right hand side gives the heat ux through con-
duction due to temperature gradient. This is the only existing term in Gas Dynamics.
The second term gives the enthalpy ux of the various species through diffusion. The
third term gives the heat ux due to the Dufour effect of reciprocal effect of the ther-
mal diffusion in Onsagers sens.11 Such an effect is zero when all the molecules have
the same mass since, then, DT i is zero. In general, the Dufour effect can be neglected
and thereby q reduces to the expression

q = T + Yi hi vdi . (3.33)
i
Taken this expression of q into Eq. (3.30), the following nal expression for
the variation of internal energy of the mixture is obtained
( )
Du
= p v + + (T ) Yi hi vdi . (3.34)
Dt i
This equation shows that the time variation of the internal energy originates
from:
a) The work p v done by pressure to compress the gas.

b) The work dissipated by the viscous forces to produce the deformation.
c) The heat (T ) received through conduction.

d) The enthalpy ux ( i Yi hi vdi ) of the species through diffusion.
Forms a), b) and c) are the same as those that exist in Gas Dynamics when there is no
change in composition.
In many cases it is convenient to work with Eq. (3.25) which includes kinetic
energy,12 bringing forth in this equation the enthalpy h of the mixture
p
h=u+ . (3.35)

For this purpose we shall start by making the pressure explicit in the rst term
of the right hand side of Eq. (3.25) by means of Eq. (3.17). Thus, the following is
11 See Ref. [5], p. 118.
12 See chapters 4 and 5.
3.5. GENERAL EQUATIONS 69
obtained for the said term
(v e ) = (pv) + (v ev ) . (3.36)
Now, if the rst term of the right hand side of this equation is expanded and
combined with the continuity equation (3.6), the following is obtained
( )
p D p
(v e ) = + (v ev ) . (3.37)
t Dt
Substituting this equation into Eq. (3.25), taking into account Eq. (3.35), it nally
gives
( )
D 1 p
h + v2 = + (v ev )
Dt 2 t
( ) (3.38)

+ F v + (T ) Yi hi vdi ,
i
where Eq. (3.33) has been used to eliminate q.
3.5 General equations
As a summary of preceding paragraphs we shall once more write the general system
of equations that govern the transformation of a mixture of reactant gases in motion,
that is to say the general equations of Aerothermochemistry.
Continuity Equation
For the mixture:

D
+ v = 0. (3.6)
Dt
For the species:
DYi
+ (Yi vdi ) = wi , (i = 1, 2, . . . , `). (3.7)
Dt
Equation of Motion
Dv
= p + ev + F . (3.18)
Dt
Energy Equation
( )
Du
= p v + + (T ) Yi hi vdi . (3.34)
Dt i
State Equation
p
= Rm T. (3.39)

Moreover the laws of variation of the thermodynamic functions, of the transport co-
efcients and of the reaction rates as functions of the state and composition of the
mixture must be known.
3.6 Entropy variation
It is interesting to bring forth the entropy variation and therewith the causes of the
irreversibility of the process. Moreover, the entropy variations have a great inuence
upon motion, for example, producing vortexes.13
Thermodynamics teaches14 that the variation dS of the entropy of a reactant

system corresponding to the variations dU , dV and dmi of its internal energy. vol-
ume and masses of the chemical species respectively, is given by the expression

T dS = dU + p dV i dmi , (3.40)
i
where i is the chemical potential of species Ai .
If one applies this expression to the mass contained in the volume element V
of Fig. 3.1 following the motion, and bearing in mind that for this volume
1 DV
S = V s, U = V u, mi = V Yi , = v, (3.41)
V Dt
where s is the entropy of the mixture per unit mass, the following expression is ob-
tained for the time variation of the entropy of the mass contained in the unit volume
( )
Ds 1 Du DYi
= + p v i . (3.42)
Dt T Dt i
Dt
13 See chapter 9.
14 See chapter 1.
3.6. ENTROPY VARIATION 71
Taking into this expression the variations of u and Yi given by Eqs. (3.34) and
(3.7) respectively, one obtains
Ds
T = + (T ) ( Yi hi vdi )
Dt i
(3.43)
i w i + i (Yi vdi ).
i i
As can easily be veried, this equation can be written in the form

[ ] [
Ds T 1 (T )2
= Yi (hi i )vdi + +
Dt T T i T T
] (3.44)
T
Yi (hi i )vdi Yi vdi i i wi
T i i i
Then we have15
i = hi T si , (3.45)
where hi and si are, respectively, the enthalpy and the entropy per unit mass of species
Ai at the partial pressure pi of the species and at the temperature T of the mixture.
When taking this expression into Eq. (3.44) one obtains
[ ] [
Ds T 1 (T )2
= Yi si vdi + +
Dt T i
T T
] (3.46)

T Yi si vdi Yi vdi i i wi .
i i i
This equation admits a simple interpretation. In fact the rst term of the right hand
side of this equation gives the reversible entropy ux across the boundary of the unit
volume. This ux originates from: a) heat transport through conduction, and b) diffu-
sion of the species. The second and third terms originate from the entropy generated
in the gas per unit volume and unit time, due to irreversibilities of the process. Such
irreversibilities, shown in Eq. (3.46), arise from: a) viscosity, b) heat conductivity,

c) diffusion, and d) chemical reactions. Making use of Eq. (3.11) the term i wi
i
corresponding to d) may be written in the form

i wi = rj j , (3.47)
i j
where j is the chemical afnity corresponding to reaction j.16 All contributions to

the variation of the entropy from the irreversibilities of the process must be positive.17
15 See chapter 1.
16 See chapter 1.
17 For further information, see Refs. [4] or [5].
3.7 One-dimensional motions
As an application we shall see the form taken by the general equations in the case of
one dimensional motions. These equations, in particular those relative to stationary
motions, will be widely used in the study of the detonations and ames.18
It will be assumed that:
a) No mass forces exist.

b) The coefcient of volumetric viscosity 0 is zero.
c) The effects of thermal diffusion and radiation can be neglected.
We shall adopt a cartesian rectangular system with the x axis parallel to the
direction of motion. Then, the only independent variables of the motion are coordinate
x and time t, and the only velocity component different from zero is that parallel to
the x axis, which will be designated as v.
The substantial derivative (3.1) reduces in this case to
D () () ()
= +v . (3.48)
Dt t x
Continuity equations
The continuity equations for the mixture reduces to
v
+v + = 0. (3.49)
t x x
Similarly, the continuity equations (3.7) for the various species, reduces to
Yi Yi
+ v + (Yi vdi ) = wi , (i = 1, 2, . . . , `). (3.50)
t x x
Equations of motion
In the vectorial Eq. (3.18) the only component not identically zero is that parallel to
x axis. Furthermore, the only component different from zero in the viscous stress
tensor ev is
4 v
,
xx = (3.51)
3 x
corresponding to viscous stress that acts upon the plane normal to the motion and
parallel to the x axis. This stress is subtracted from the pressure. Therefore, Eq. (3.18)
18 See chapters 5 and 6.
3.7. ONE-DIMENSIONAL MOTIONS 73
reduces to ( )
v v p 4 v
+ v = + . (3.52)
t x x 3 x x
Energy equation
Equation (3.34) reduces to

( )2
u u v 4 v
+ v =p +
t x x 3 x
( ) ( ) (3.53)
T
Yi hi vdi .
x x x i
Or if it is decided to use the total energy equation (3.38) it takes the form
( ) ( ) ( )
1 2 1 2 p 4 v
h + v + v h+ v = + v
t 2 x 2 t 3 x x
( ) ( ) (3.54)
T
+ Yi hi vdi .
x x x i
The system of equations (3.49), (3.50), (3.52) and (3.53), or (3.54), must be
completed with the following equations.
Total mass fraction equation

Yi = 1. (3.55)
i
Diffusion equations
The system of equations (3.8) which gives the diffusion velocities reduces to
Yj [ vdi vdj ( 1 Yi 1 Yj
)
+
j
Mj Dij Yi x Yj x
( )]
Mj Mi 1 p (3.56)
+ = 0, (i = 1, 2, . . . , `),
Mm p x

Yi vdi = 0.
i
State equation
The state equation is, by assumption, that of ideal gases

p
= Rm T. (3.39)

State functions
The internal energy ui of species Ai is19

T
ui = ui0 + cvi dT. (3.57)
T0
The internal energy of the mixture is

T
u= Yi ui = Yi ui0 + cv dT, (3.58)
i i T0
where

cv = Yi cvi (3.59)
i
is the mixture heat capacity at constant volume.
Similarly, for the enthalpy we have

T
hi = hi0 + cpi dT, (3.60)
T0
and for the enthalpy of the mixture

T
h= Yi hi = Yi hi0 + cp dT, (3.61)
i i T0
with

cp = Yi cpi . (3.62)
i
Transport coefcients
In system (3.56), Dij are the binary diffusion coefcients between species Ai and
Aj . For each pair of these species the coefcients depend on the state variables, as
indicated in chapter 2.
and are the coefcients of thermal conductivity and viscosity of the mix-
ture, respectively. These coefcients depend on the state variables and on the compo-
sition of the mixture as indicated in chapter 2.
Reaction rates
The reaction rate of species Ai is given in Eq. (3.11). In this equation the coefcients
ij and the velocities rj corresponding to the different elementary reactions of the
mixture are given by the formulae of chapter 1.
19 See chapter 1.
3.8. STATIONARY, ONE-DIMENSIONAL MOTIONS 75
3.8 Stationary, one-dimensional motions
If the motion is not only one-dimensional but stationary, the preceding equations are
considerably simplied. This type of motion, as previously stated, is of special interest
for the study of the structure of combustion waves.
The stationary motions are characterized by the condition

()
= 0. (3.63)
t
Therefore, the only independent variable is x.
Let us see how the equations in the preceding paragraph can be simplied when
condition (3.63) is introduced therein.
Continuity Equations
The continuity equation (3.49) reduces to

d dv d
v + (v) = 0. (3.64)
dx dx dx
This equation can be integrated, thus obtaining
v = m, (3.65)
where m is an integration constant which gives the mass ux per unit surface normal
to the direction of motion.
Similarly, equations (3.50) reduces to

d
(Yi v + Yi vdi ) = wi , (i = 1, 2, . . . , `) . (3.66)
dx
By introducing the ux of the different species, these equations reduce to a
very simple form. In fact, let mi = mi be the ux of species Ai through the unit
surface normal to the direction of motion, that is to say that i is the fraction of the
total ux corresponding to species Ai . Since the velocity vi of species Ai is
vi = v + vdi , (3.67)
evidently the said ux is

mi = Yi (v + vdi ) . (3.68)
Now, when this expression is compared with the left hand side of Eq. (3.66) it
is seen that this system can be written in the form
di
m = wi , (i = 1, 2, . . . , `). (3.69)
dx
The physical interpretation of this equations is obvious, m di is the difference

between the uxes of species Ai through two surfaces normal to the direction of mo-
tion, separated one from another by the distance dx. But this difference has its origin
in the amount of the said species produced per unit time by the chemical reactions that
take place within the space that separates both surfaces, which is evidently wi dx.
The ` equations of system (3.69) are not independent. In fact:
1) Formulae (3.11) shows that wi are linear combinations of the r reaction rates rj
corresponding to the different reactions that take place between the species of the
mixture. Consequently, the maximum number of independent wi is at most r.
2) The chemical reactions do not change the total number of atoms of the elements
that form the species. Let g be the number of different chemical elements of the
species and let aij (j = 1, 2, ...., g) be the number of atoms of the element j in
species i. The conservation of the element in the chemical reactions imposes the
following system of conditions between wi
wi
aij = 0, (j = 1, 2, . . . , g). (3.70)
i
M i
It might occur however that not all these conditions are independent. In fact
the number of independent equations in (3.70) is the rank of the matrix of coefcients
aij . This rank is the minimum number of components needed to form the ` species
Ai in the sense of the phase rule. Therefore, if g 0 g is the number of components of
the mixture, there exist g 0 independent linear relations between wi , due to Eq. (3.70),
and the number of independent wi is, at most, ` g 0 .
Thereby, the number `0 < ` of the independent wi is the smallest one of the
numbers r and `0 g 0 .20
System (3.69) shows that there are as many i independent from each other as
there are wi , that is to say, `0 < `, which makes it possible to reduce the number of
variables of the problem in ` `0 .
Equation of Motion

( )
dv dp 4 d dv
m = + , (3.71)
dx dx 3 dx dx
20 As it can easily be proved, the difference ` g 0 between the number of species and the number
of components is the maximum number of independent reactions equations which can exist among the `
species. This property simplies the interpretation of the conclusion concerning the number of independent
wi . In fact, if ` g 0 < r, the r chemical reactions are not linearly independent from each other.
3.9. THE CASE OF ONLY TWO CHEMICAL SPECIES 77
which by integration gives

4 dv
p + mv = i, (3.72)
3 dx
where i is an integration constant.
Energy Equation
In this case it is convenient to make use of the total energy equation (3.54) which by
integration gives
1 dT 4 dv
mh + Yi hi vdi + mv 2 v = e, (3.73)
i
2 dx 3 dx
where e is an integration constant.

The two rst terms of this equation give the enthalpy ux m i hi of the
i
mixture per unit surface normal to x, as can easily be proved by keeping in mind
(3.61) and (3.68). Therefore Eq. (3.73) may also be written in the form
( )
1 2 dT 4 dv
m v + i hi v = e, (3.74)
2 i
dx 3 dx
whose physical interpretation is obvious.
Diffusion Equations
Equations (3.56) remain valid

Yj [ vdi vdj ( 1 dYi 1 dYj
)
+
j
Mj Dij Yi dx Yj dx
( )]
Mj Mi 1 dp (3.75)
+ = 0, (i = 1, 2, . . . , `),
Mm p dx

Yi vdi = 0.
i
3.9 The case of only two chemical species
Let us now assume that in the preceding case the number of chemical species of the
mixture reduces to two, A1 and A2 . Then, a single chemical variable Y1 , denoted Y ,
is enough to dene the composition of the mixture, since Y2 = 1 Y . Similarly the
ux fraction i of A1 will be denoted and one has 2 = 1 .
System (3.69) reduces in this case to the single equation

d
m= = w. (3.76)
dx
Equation of motion (3.72) remains the same and the energy equation (3.74)
reduces to
( )
1 2 dT 4 dv
m h2 + (h1 h2 ) + v v = e. (3.77)
2 dx 3 dx
Finally, the system of diffusion equations (3.75) reduces to the single equation
( [ ])
Y vd1 dY (M2 M1 ) (M2 M1 ) Y + M1 Y (1 Y ) dp
+ + = 0, (3.78)
D12 dx M1 M2 p dx
which making use of Eq. (3.68) that in this case reduces to
m = Y (v + vd1 ) = mY + Y vd1 , (3.79)
can be written as
[ ]
dY (M2 M1 ) (M2 M1 ) Y + M1 Y (1 Y ) dp m
+ = (Y ). (3.80)
dx M1 M2 p dx D12
The unknown of the problem are in this case v, p, , T , Y and . The six equa-
tion which determine their values are the following: continuity (3.65) and (3.76), mo-
tion (3.72), energy (3.77), diffusion (3.80) and state (3.39). The system thus formed
will be studied in detail in chapters 5 and 6 in discussing the structure of the com-
bustion waves, specially under the simplied form studied in the following. Let us
assume, in particular, the following two conditions:
1) The heat capacities at constant pressure of both species are equal
cp1 = cp2 = cp . (3.81)
2) The molar masses of both species are equal, or else the gradient of pressure is
very small.
By virtue of the rst assumption, and taking into account (3.60), one has
h2 h1 = h20 h10 = q, (3.82)
where q is the heat of reaction per unit mass of the mixture.
Due to (3.82), equation (3.77) reduces to

( )
1 dT 4 dv
m h2 q + v 2 v = e. (3.83)
2 dx 3 dx
3.10. STATIONARY, ONE-DIMENSIONAL MOTION OF IDEAL GASES WITH HEAT ADDITION 79
By virtue of the second assumption, the diffusion equation (3.80) reduces to
dY m
= (Y ), (3.84)
dx D12
which is the expression of Ficks Law.
If, furthermore, the heat capacity cp is independent from temperature, then

equation (3.83), considering (3.60), takes the form
( )
1 2 dT 4 dv
m cp T q + v v = e, (3.85)
2 dx 3 dx
where e must now include the constant terms that come from (3.60).
3.10 Stationary, one-dimensional motion of ideal gases

with heat addition
Let us assume in the above case that the action of viscosity and thermal conductivity
can be neglected in the equation of motion (3.72) and energy (3.85). This is justied if
the gradients of velocity and temperature are not very large.21 Then (3.72) reduces
to
p + mv = i. (3.86)
Likewise (3.85) reduces to

( )
1
m cp T q + v 2 = e. (3.87)
2
These two equations together with (3.65)
v = m (3.88)
and the equation of state (3.39)

p
= Rm T, (3.89)

determine the values for p, , T and v corresponding to each value of , if the values of
such variables corresponding to a given value of are known, for example, to = 0.
This enables the determination of the constants of Eqs. (3.86), (3.87) and (3.88). In
such case, in Eq. (3.87) the term Q = q represents the heat added to the gas per unit
mass from the initial state = 0. The result is independent from the way in which the
heat is added; be it by chemical reactions or transmitted from external sources.
21 See chapter 5 for the exact meaning of this expression.
In the following, Rm is assumed to be constant as occurs if the molar masses

of both species are equal.
Taking in abscissas the heat added to the gas and in ordinates the values for
v and T 22 a diagram can be drawn which clearly shows the transformations taking
place within the gas. This has been done in Fig. 3.3 for the case where the relation
of heat capacities is 1.4. The dimensionless variables which will be dened in the
following are introduce in order to have an universal diagram.
1) Let
+ 1 mas0
b= , (3.90)
i

where as0 = Rm Ts0 is the velocity of sound at the stagnation condition Ts0
corresponding to the initial state Q = 0.
A simple calculation shows that b can be expressed as a function of the Mach
number M0 in the form:

( + 1) M0 1 2
b= 1+ M0 . (3.91)
1 + M02 2
Similarly, let M0 be ratio of v0 to the critical velocity vcr of the gas at the initial
state. b can also be expressed as a function of M0 in the form
M0
b= 2( + 1) . (3.92)
1 + M02
b and M0 are represented in Fig. 3.2 as functions of M0 for = 1.4.
2) Let Ts be the stagnation temperature of the gas when the added heat is Q, and
let n be the ratio of Ts to Ts0 ,
Ts
. n= (3.93)
Ts0
The relation between Ts and Ts0 is obtained when expressing e in Eq. (3.87) by
means of Ts and Ts0 . Thus resulting
Q
n=1+ . (3.94)
cp Ts0
The added heat is expressed by means of the variable x dened by
x = b2 n. (3.95)
3) The velocity is expressed in dimensionless form by

v
v = b . (3.96)
as0
22 See von Karman, Ref. [7].
1.2 6
1.0 5
b
0.8 4
0
0.6 3
b
M
0.4 2
M0
0.2 1
0 0
0.0 0.5 1.0 1.5 2.0 2.5
M*
0
Figure 3.2: M0 and b (Eq. (3.92)) as a function of M0 .
4) Likewise, the temperature is expressed in dimensionless form by
T
= b2 . (3.97)
Ts0
If these variables are taken into the previous system and this system is solved

for v and , one obtains
2x
(v 1) = 1
2
, (3.98)
+1
1 2
=x v . (3.99)
2
The values of v and are represented in the diagram Fig. 3.3 where the law
of variation of the Mach number M of the motion is also included. M is given by the
expression
v
M= . (3.100)

It can easily be veried that for v and to be real, x cannot exceed the value
+1
xmax = , (3.101)
2
to which corresponds a maximum value nmax of n given by the expression
+1
nmax = . (3.102)
2b2
2.5 1.25
M+
2.0 1.00
B v*
+
1.5 0.75
v*, M

1.0 + A 0.50
0.5 0.25
M
v*

O
0.0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
x
Figure 3.3: Mach number (M ), nondimensional velocity (v ) and temperature () as a func-
tion of the nondimensional heat added to the gas (x).
To this value corresponds a maximum value Qmax of Q given by the expression

( )
+1
Qmax = 1 cp Ts0 . (3.103)
2b2
This value is determined by the initial condition through b2 and cp Ts0 . Qmax
is the maximum heat that can be added to the ow consistent with the given initial
conditions.
The values of v1 , 1 and M1 of v , and M corresponding to x = xmax are
v1 = 1, 1 = 1, M1 = 1. (3.104)
Therefore the velocity of the gas at this point is the velocity of sound. In the
curve of velocities point A, corresponding to x = xmax , divides this curve into two
branches OA subsonic and AB supersonic. Fig. 3.4 shows that if the initial velocity
is subsonic the gas accelerates as the heat is added whilst if the initial velocity is
supersonic the gas decelerates. In both cases the maximum heat that can be added

is given by (3.102) or (3.103). Conversely, the maximum initial velocity M0,max for
each value of n is obtained when Eq. (3.102) is solved for M0 . Thus results

M0,max = n n 1. (3.105)

Sign plus corresponds to supersonic velocities since M0,max > 1. Sign minus
corresponds to subsonic velocities. This case has great importance in technical appli-
cations as it determines, for example, the maximum velocity permissible at the inlet
of a constant cross-section combustion chamber as a function of the heat released in
the same.23 The value of the corresponding Mach number M0,max is given by

2
2M0,max
M0,max = 2 . (3.106)
( + 1) ( 1) M0,max

The values of M0,max and M0,max as a function of n are represented in Fig. 3.4.
n (supersonic)
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1
14
subsonic
supersonic
12
M0,max, M0,max (supersonic)

(subsonic)
1.0 10
M0,max
0.8 8
0,max
0.6 6
,M
M*
0,max
0,max
0.4 4
M*
*
0.2 2
M0,max
0.0 0
1 3 5 7 9 11 13 15 17 19 21 23
n (subsonic)

Figure 3.4: Values of M0,max and M0,max as a function of n.
It is easily seen that this choking effect imposes an important limitation to the Mach
number at the inlet of the combustion chamber.
With respect to the law of variation of temperature as a function of Q it is

interesting to remark that the maximum temperature has the value
2
( + 1)
max = , (3.107)
4
and it is reached for a value of n slightly smaller that its maximum value as well as
for a subsonic velocity. This is due to the fact that for velocities slightly subsonic the
acceleration produced by the added heat is so large that the heat received is unable to
balance the cooling produced by the expansion of the accelerating gas. In chapter 5,
where these results are applied, it is seen that this effect produces in detonation.
23 See chapter 10.
3.11 Appendix: Notation and repertoire of vectorial

and tensorial formulae used in the present
chapter
Scalar: a
Vector: a (components a1 , a2 and a3 )
Scalar product: a b (scalar)
Vectorial product: a b (vector)
The Hamilton nabla:
() () ()
() = i1 + i2 + i3 (vectorial operator)
x1 x2 x3
i1 , i2 and i3 are unit vectors parallel to the coordinate axes.
Gradient of a scalar:
a a a
a = i1 + i2 + i3 (vector)
x1 x2 x3
Divergence of a vector:
a1 a2 a3
a = + + (scalar)
x1 x2 x3
Curl of a vector:
( ) ( ) ( )
a3 a2 a1 a3 a2 a1
a = i1 + i2 + i3 (vector)
x2 x3 x3 x1 x1 x2
Convective derivative:
() () ()
v () = v1 + v2 + v3 (scalar operator)
x1 x2 x3
( ) ( )
ab = b a + a b .
Tensor:
11 12 13

= 21 22 23
31 32 33
Transposed of tensor (permutation of rows by columns):

11 21 31

= 12 22 32
13 23 33
3.11. APPENDIX: NOTATION AND REPERTOIRE OF VECTORIAL AND TENSORIAL FORMULAE 85
Symmetric tensor: = .
If is symmetric:
ij = ji (i, j = 1, 2, 3)
Product of a vector by a tensor:
a = i1 (a1 11 + a2 21 + a3 31 )
+ i2 (a1 12 + a2 22 + a3 32 )
+ i3 (a1 13 + a2 23 + a3 33 ) (vector)
Divergence of a tensor:
( )
11 21 31
= i1 + +
x1 x2 x3
( )
12 22 32
+ i2 + +
x1 x2 x3
( )
13 23 33
+ i3 + + (vector)
x1 x2 x3
Gradient of a vector:
a a2 a3
1
x1 x1 x1

a1 a2 a3
a =
x
(tensor)

2 x2 x2
a a2 a3
1
x3 x3 x3
Double product:
( )
a b (scalar)
If is symmetric, then:
( ) ( )
a b = b a
Contraction of two tensors:

3
3
: 0 = 0
ij ji (scalar)
i=1 j=1
Therefore
: 0 = 0 :
and
(a ) = a ( ) + (a) : (scalar)
If is symmetric
1
(a ) = a ( ) + (a) : = a ( ) + (a + a) :
2
The Gauss formula:
In a domain V enclosed by surface S:

(n ) d = ( ) dV
S V
can be a scalar, vector or tensor. Symbol represents any of the several types of
products previously dened.
Derivative of an integral with changing boundary:

d (t, x)
(t, x) dV = dV + (n v) d
dt V (t) V (t) t (t)
where v is the velocity of at d and n is the outwards normal to at d.
References
[1] Kuethe, A. N. and Schetzer, J. D.: Foundations of Aerodynamics. John Wiley

and Sons, Inc. New York. 1950.
[2] Green, R. S.: The Molecular Theory of Fluids. Interscience Publishers, Inc,
New York, 1952.
[3] Richardson, J. M. and Brinkley, S. R.: Mechanics of Reacting Continua. Com-
bustion Processes, Sec. F, Vol. II of High Speed Aerodynamics and Jet Propul-
sion, Princeton University Press, 1956, p. 203.
[4] Hirschfelder, J. O., Curtiss, C. F. and Bird, R. B.: Molecular Theory of Liquid
and Gases. John Wiley and Sons, Inc., New York, 1954.
[5] De Groot, S. R.: Thermodynamics of Irreversible Processes. North Holland
Publishing Comp., Amsterdam, 1951.
[6] Prigogine, I. and Defay, R.: Chemical Thermodynamics. Longmans Green and
C., New York, 1954.
[7] von Karman, Th.: The Theory of Shock Waves and the Second Law of Thermo-
dynamics. LAerotecnica, Febr. 15, 1951, pp. 82-3.
[8] Chambre, P. and Lin, C. C.: On the Steady Flow of a Gas through a Tube with
Heat Exchange or Chemical Reaction. The Journal of Aeronautical Sciences,
Oct. 1946, pp. 537-42.
3.11. APPENDIX: NOTATION AND REPERTOIRE OF VECTORIAL AND TENSORIAL FORMULAE 87
[9] Shapiro, A. H. and Hawthorne, W. R.: The Mechanics and Thermodynamics

of Steady One-Dimensional Gas Flows. Journal of Applied Mechanics, Trans.
A.S.M.E., Vol. 14, No. 4, 1947, p. 317.
[10] Foa, J. V. and Rudinger, G.: On the Addition of Heat to a Gas Flowing in a
Pipe at Subsonic Speed. The Journal of Aeronautical Sciences, Febr. 1949,
pp. 84-94.

Chapter 4
Combustion Waves
4.1 Detonation and deagration
When an explosive mixture is ignited, a wave or reaction front originates which prop-
agates the combustion throughout the mixture. The wave thickness is normally very
small. By assuming this thickness to be zero, that is, assuming that gases burn in-
stantaneously as they cross the wave, the state of the mixture and the state of the burn
gases can be compared, in a way similar to that used when studying shock waves;1
that is, by applying the laws of the conservation mass, momentum and energy across
the front, without taking into consideration the intermediate transformations between
the initial and nal states Thus, the following system of three equations is obtained
1 v1 = 2 v2 , (4.1)
1 v12 + p1 = 2 v22 + p2 , (4.2)

1 2 1
v + h1 = v22 + h2 , (4.3)
2 1 2
where , p and h are the density, pressure and total enthalpy per unit mass and v the
normal velocity relative to the reaction front. Subscripts 1 and 2 refer to the unburnt
gases and to the combustion products respectively. In particular, v1 is the propagation
velocity of the wave, that is, the velocity at which the wave moves through the unburnt
gases, Fig. 4.1.
For a given composition, the specic enthalpy h1 of the unburnt gases is a

function of the pressure p1 and the density 1 of the mixture
h1 = h1 (p1 , 1 ) . (4.4)
1 Seei.e. Courant-Friedrichs: Supersonic Flow and Shock Waves. Intersciences Pub. Inc., New York,
1948, pp. 121 and following.
89
90 CHAPTER 4. COMBUSTION WAVES
Unburned gases Burned gases
V1 V2
p T1 p T2
1 1 2 2
Combustion wave
Figure 4.1: Schematic diagram of a combustion wave to obtain the relations between initial
and nal states.
Similarly, if the burnt gases are in thermodynamic equilibrium, h2 is only a

function of p2 and 2
h2 = h2 (p2 , 2 ) . (4.5)
When equations (4.4) and (4.5) are substituted into (4.3), this equation together
with Eq. (4.1) and (4.2) form a system of three equations with six unknowns: p1 , 1 ,
v1 , p2 , 2 and v2 . If three of these values are known, the system determines the values
for the other three. As said before, the state of the unburnt gases is dened by the
values p1 and 1 . Therefore, in order to determine the propagation, another of these
values must be known, for instance that of the propagation velocity v1 . The analysis
of the propagation velocity shows the existence of two types of essentially different
waves. In fact, the elimination of v2 between Eq. (4.1) and (4.2), gives for v1 ,
v
1u p2 p1
v1 = u . (4.6)
1 t 1 1

1 2
The study is simplied by introducing in the above system the specic volume
, which is related to the density by
1
= , (4.7)

thus obtaining for v1
p2 p1
v1 = 1 . (4.8)
1 2
Since the propagation velocity must be real, the following condition must be
satised
p2 p1
> 0. (4.9)
1 2
4.2. KINDS OF DETONATIONS AND DEFLAGRATIONS 91
Consequently, if p2 > p1 , then 1 > 2 , and if p2 < p1 , then 1 < 2 .

Therefore, when the pressure increases across the wave the gases contract, and if the
pressure decreases, the gases expand. Both types of waves are physically observed.
The rst type is called detonation and the second deagration. The propagation ve-
locity of a detonation wave in an explosive mixture is of the order of several thousand
meters per second. The propagation velocity of a deagration wave is normally of the
order of 1 m/s.
4.2 Kinds of detonations and deagrations
The study can be simplied by representing the state of unburnt and burnt gases in the
diagram (p, ), Fig. 4.2, proposed by Grussard [1].
III
I
H
p2
A
p1
P B
IV
II D
0 0 1 2 3
Figure 4.2: Sketch of the Hugoniot curve for the nal states.
In this diagram, let P (p1 , 1 ) be the representative point of the initial state.
Condition (4.9) excludes regions I and II from this diagram, since in these regions
(p2 p1 ) / (1 2 ) is smaller than zero. The representative points of the nal states
must be, consequently, within regions III and IV. The points in region III correspond to
detonations, since, in this region p2 > p1 and 2 < 1 . Points in region IV correspond
to deagrations.
The possible nal states corresponding to the initial state (p1 , 1 ) and compati-
ble with conditions (4.1), (4.2) and (4.3), lie on a curve H, called the Hugoniot curve.
This curve is obtained from the elimination of v1 and v2 between Eqs. (4.1), (4.2) and
(4.3). There results for H, the following equation
1
h2 h1 (1 + 2 ) (p2 p1 ) = 0. (4.10)
2
This curve has a detonation branch and a deagration branch.
Let p2 be the nal pressure corresponding to an adiabatic combustion at con-

stant volume 1 . Since the reaction is exothermic p2 > p1 and the representative point
of the state (p2 , 1 ), will be, for instance, point A. The detonation branch starts at this
point. The propagation velocity corresponding to this detonation at constant volume
is, according to Eq. (4.8), innite. Hence, at this point the combustion propagates
instantaneously throughout the mass.
Similarly, let 2 be the specic volume corresponding to an adiabatic combus-

tion at constant pressure p1 . Here 2 > 1 , and point B, corresponding to the state p1 ,
2 , is the starting point of the deagration branch. The propagation velocity of this
limit constant pressure deagration is, according to (4.8), zero.
Hereinafter, it is assumed that the Hugoniot curve satises the following con-
ditions ( ) ( )
p 2p
<0, > 0, (4.11)
H 2 H
where subscript H indicates differentiation along the Hugoniot curve. These con-
ditions mean that H is monotonically decreasing and turns its concavity to the axis
p > 0. Both conditions correspond to the curves generally observed in practice. In
general, curve H has an asymptote, parallel to the pressure axis, for = 0 > 0, and
cuts the specic volume axis at point 3 . Therefore, the general form of H is the one
shown in Fig. 4.2.
Let us consider a straight line that starts from point P , corresponding to the
initial state, and enters in region III of detonations. Let , Fig. 4.3(a), be the angle
between this line and the negative direction of the axis. Depending on the values of ,
the three following cases are possible:
1) If is smaller than a certain limit value min which corresponds to the tangent
from P to H, the straight line corresponding to cannot cut the Hugoniot curve.
2) If = min , the corresponding line is, as aforesaid, tangent to H at point J.
3) If > min , the corresponding line cuts the Hugoniot curve at two points E
and E 0 at different sides of point J.
4.2. KINDS OF DETONATIONS AND DEFLAGRATIONS 93
p C
E
J E
A
min
p1 P
0 1
(a) Detonation branch
p
P
p1 B
J
D
1 3
(b) Deflagration branch
Figure 4.3: Chapman-Jouguet points in the detonation and deagration branches of the
Hugoniot curve.
These properties can immediately be translated into properties for the propa-
gation velocity of the detonation. In fact, we have2
p2 p1
tan = . (4.12)
1 2
Taking this value into Eq. (4.8), the following expression for the propagation velocity
is obtained

v1 = 1 tan . (4.13)
Therefore, it results that the propagation velocity of a detonation is minimum

at point J, where the straight line P J is tangent to H. This property was observed
by Chapman in 1899 [2]. Point J is called the Chapman-Jouguet point and the
corresponding detonation the Chapman-Jouguet detonation. The Chapman-Jouguet
detonation is important since it is the one usually observed. Hence, one concludes
that of all the possible detonations, compatible with the given initial conditions, the
Chapman-Jouguet detonation is the one which propagates at a minimum velocity.
2 To account for dimensions a constant should be included. Hereinafter it is omitted for simplicity.
The detonations corresponding to the upper part JC of H, for which the pres-
sure of the burnt gases is greater than the pressure of the Chapman-Jouguet detonation,
have been called by Courant and Friedrichs strong detonations and those correspond-
ing to the lower part AJ, weak detonations.
In the same manner it can be proven, Fig. 4.3(b), that in the deagration branch
there exists a point J 0 in which H and P J 0 are tangent and the corresponding dea-
gration, also called Chapman-Jouguet, propagates with a maximum velocity. Point J 0
divides the deagration branch in two parts, one J 0 D of strong deagrations and the
other BJ 0 of weak deagrations.
4.3 Velocity of the burnt gases
It is interesting to determine the subsonic or supersonic character of the velocity v2

of the burnt gases with respect to the wave front, since the stability of the detonation
wave depends on the character of this velocity, that is, the possibility for this wave to
propagate unchanged through the gas. In order to determine the character of v2 it is
necessary to compare this velocity with the sound velocity a2 in the burnt gases. Let
us see how this can be attained.
Gas Dynamics shows3 that the sound velocity a in a gas is given by the expres-
sion ( )
p
a= , (4.14)
S
where subscripts S means differentiation along the isentropic that passes through the
representative point of the gas state.
Let S , Fig. 4.4, be the angle between the tangent to the isentropic that passes
through the point (p2 , 2 ) of the detonation branch, and the negative direction of x
axis. From Eq. (4.14) we have

a2 = 2 tan S . (4.15)
On the other hand, the velocity v2 of the burnt gases with respect to the ame front,
2
in virtue of Eq. (4.1), is v2 = v1 . Then, because of Eq. (4.13), this velocity can be
1
expressed as

v2 = 2 tan , (4.16)
where is the previously dened angle. If is larger, equal to or smaller than s ,
then v2 will be larger, equal to or smaller than a2 ; and the ow of the burnt gases
3 See Courant-Friedrichs: Supersonic Flow and Shock Waves. Intersciences Pub. Inc., New York, 1948.
4.3. VELOCITY OF THE BURNT GASES 95
> S : supersonic = S : sonic < S : subsonic
S

S
S
S ( 2 , p2 ) ( 2 , p2 ) ( 2 , p2 )
P( 1 , p1 ) P( 1 , p1 ) P( 1 , p1 )
Figure 4.4: Schematic diagram showing the line connecting the initial and nal states and
the corresponding isentropic curve.
relative to the wave will be supersonic, sonic or subsonic. Hence, the problem reduces
to a comparison between the slope of the isentropic that passes through each point of
H, and that of the radius vector that joins this point with point P of the initial state.
The three possible cases appear schematically in Fig. 4.4.
To perform this comparison let us start by studying the variation of the entropy
s, along the Hugoniot curve. The elemental entropy variation ds corresponding to the
variations dh and dp of enthalpy h and pressure p, is
T ds = dh dp. (4.17)
Now, by differentiating (4.10), keeping p1 and 1 constant and taking the result into
(4.17), the following expression for the entropy variation along H, is obtained
( ) [ ( ) ]
s2 1 2 p2 p1 p2
T2 = +
2 H 2 1 2 2 H
(4.18)
(1 2 ) (tan tan H )
= ,
2
where H is the angle between the tangent to H at point p2 , 2 , and the negative
direction of axis.
Fig. 4.3 shows that in the strong detonation branch H > . Therefore, in
virtue of (4.18), in this branch ( )
s2
< 0. (4.19)
2 H
On the contrary, in the weak detonation branch H < , that is
( )
s2
> 0. (4.20)
2 H
Finally, at the Chapman-Jouguet point H = , that is

( )
s2
= 0. (4.21)
2 H
From these three inequalities it results that the entropy of the burnt gases is
minimum at the Chapman-Jouguet point and increases from there on, both in strong
and weak detonation branches.
Since at point J condition (4.21) is satised, the isentropic passing through this
point is tangent to the Hugoniot curve. Therefore, at point J the following conditions
are satised
S = H = = min . (4.22)
These values, when taken into Eqs. (4.15) and (4.16), give at point J
a2 = v2 , (4.23)
thus resulting the following important property: in a Chapman-Jouguet detonation

the velocity of the burnt gases with respect to the wave is sonic. This property is
the one that best characterizes a Chapman-Jouguet detonation. As will presently be
seen, owing to this property, the detonations physically observed are of this same type.
Jouguet stated this property in 1905 [3].
For the purpose of determining the character of v2 in strong and weak det-
onations, it is necessary, as previously seen in Fig. 4.3, to compare and s . This
means that it is necessary to determine the relative position of the isentropic that passes
through each point of H, with respect to the straight line that joins P to that point. For
the determination of this relative position it is not sufcient to know the entropy vari-
ation along H, but it is also necessary to know the entropy variation along another
direction, for instance, that of the radius vector with its origin at point P . This can
easily be attained by considering the function
1
F (p1 , 1 ; p2 , 2 ) h2 h1 (1 + 2 ) (p2 p1 ) . (4.24)
2
On the Hugoniot curve H the value of this function is zero, as results from (4.10).
This curve divides the plane in two regions: the lower region, that contains point P ,
representative of the initial state, and the upper region. F takes opposite signs in both
regions. As can be easily be veried, in the lower region F < 0 and in the upper
region F > 0. For this purpose, it is enough to study, for instance, the sign of F in P .
In B we have (Fig. 4.2)
F (p1 , 1 ; p1 , 2 ) = h2 (p1 , 2 ) h1 (p1 , 1 ) = 0. (4.25)

4.3. VELOCITY OF THE BURNT GASES 97
Similarly, in P
F (p1 , 1 ; p1 , 1 ) = h2 (p1 , 1 ) h1 (p1 , 1 ) . (4.26)
Now subtracting (4.25) from (4.26), results
F (p1 , 1 ; p1 , 1 ) = h2 (p1 , 1 ) h2 (p1 , 2 ) . (4.27)
But, 1 is smaller than 2 , and, since to reduce the volume of the burnt gases from 2
to 1 without varying their pressure p1 it is necessary to cool the burnt gases, that is to
reduce their enthalpy, then
h2 (p1 , 1 ) < h1 (p1 , 2 ) , (4.28)
that is
F (p1 , 1 ; p1 , 1 ) < 0. (4.29)
Therefore F is negative in the lower region and positive in the upper region.
By means of the function F , the entropy variation can be expressed in a simple

manner by differentiating Eq. (4.24), keeping p1 and 1 constant and then combining
the result with Eq. (4.17), thus, obtaining
1 1
T2 ds2 = dF + (p2 p1 ) d2 + (1 2 ) dp2 . (4.30)
2 2
But the variations dp2 and d2 corresponding to the straight line that joins P with
(p2 , 2 ) must satisfy condition
dp2 p2 p1
= , (4.31)
d2 2 1
this expression, when taken into Eq. (4.30), gives for entropy variation along a radius
vector
T2 ds2 = dF. (4.32)
Therefore, along a radius vector, F and s2 increase and decrease in the same direction.
Now, at point E (Fig. 4.5), F increases from E to E 0 . Therefore, s2 also

increases from E to E 0 . In the same manner, at point E 0 , F and s2 increase in the
direction E 0 E.
Furthermore, s2 decrease along H in the directions E 0 J and EJ. Thereby it is

deduced that the isentropic curves that pass by E and E 0 must, necessarily, have the
positions shown in Fig. 4.5. There results that:
a) In E, S < and the velocity of the burnt gases is supersonic.

b) In E 0 , S > and the velocity of the burnt gases is subsonic.
p C
E
S2
J E
A
p1 S2
P
0 1
Figure 4.5: Position of the isentropic curves in the detonation branch of the Hugoniot curve.
Hence the velocity of the burnt gases is supersonic in the weak detonations and sub-
sonic in the strong detonations.
Similarly, it can be demonstrated that the entropy of the burn gases in the de-
agration branch is maximum at J 0 . It can also be demonstrated that the velocity of
the unburnt gases is subsonic in the weak deagrations, sonic in the Chapman-Jouguet
deagration and supersonic in the strong deagrations.
4.4 Propagation velocity
In order to complete the study, we shall analyze the character of the propagation ve-
locity v1 , by comparing its values with the value of the velocity a1 of the sound prop-
agation in the state (p1 , 1 ). The study can be made in an identical manner to that
previously used for v2 , that is by considering all the states (p1 , 1 ) of the unburnt gases
that are compatible with a given state (p2 , 2 ) of the burnt gases. All these states lie
on a Hugoniot curve H 0 which is obtained by xing in Eq. (4.10) the values of p2 and
2 and varying p1 and 1 . This curve, like curve H, has two branches (Fig. 4.6): a
lower detonation branch and an upper deagration branch. It can easily be proved that
in the former, the entropy of the unburnt gases decreases starting from A, and in the
latter increases starting from B. Furthermore by studying the variation of function F ,
which is dened by (4.24) xing the values of p2 and 2 and varying p1 and 1 , we
obtain that H 0 divides the plane (p, ) in two regions. The upper region containing
4.4. PROPAGATION VELOCITY 99
p H
S2
B ( 2, p2 )
p2
H
A
S2
2
Figure 4.6: Position of the isentropic curves in the Hugoniot curve of the admissible initial
states.
point (p2 , 2 ), in which F < 0, and the lower region, in which F > 0. Finally, it
is also veried that along the radius vectors that pass through the point (p2 , 2 ) the
variations of F and s1 are opposite. Now, by combining these properties as previ-
ously done for v2 , we can determine the relative position of the isentropic that passes
through each point H 0 with respect to the straight line that joins this point to the point
(p2 , 2 ). It results that this position is the one indicated in Fig. 4.6. Therefore, in the
detonation branch < S and in the deagration branch > S . Therefore, one
concludes that the propagation velocity of a detonation is always supersonic, and the
propagation velocity of a deagration is always subsonic.
As a result of the preceding study, the Table 4.1 can be constructed. This
table contains the results of Jouguets studies and denes the characteristics of the
combustion waves.
Detonations
Weak Chapman-Jouguet Strong
Velocity before Supersonic
Velocity after Supersonic Sonic Subsonic
Deagrations
Weak Chapman-Jouguet Strong
Velocity before Subsonic
Velocity after Subsonic Sonic Supersonic
Table 4.1: Characteristics of the combustion waves.

4.5 Applications
As an application of the previous study we shall consider the propagation of a com-

bustion wave, assuming that both the unburnt and burnt gases are perfect gases and
that the chemical composition of the burnt gases is independent from their state. In
this case, the enthalpy of the burnt gases can be expressed in the form
T2
h2 = h21 + cp2 dT, (4.33)
T1
where h21 is their enthalpy at the temperature T1 . Moreover, it is assumed that the
heat capacity cp2 is independent from the temperature in the interval (T1 , T2 ). Then
Eq. (4.33) reduces to
h2 = h21 + cp2 (T2 T1 ) . (4.34)
Now, by taking this expression into the Hugoniot equation (4.10), we have
1
cp2 T2 = (h11 h21 ) + cp2 T1 + (1 + 2 ) (p2 p1 ) , (4.35)
2
where h11 h21 is the difference between the formation enthalpies of the burnt and
unburnt gases at the temperature T1 of the unburnt gases. Let Q be this value. Then
Eq. (4.35) can be written in the form
1
cp2 T2 = Q + cp2 T1 + (1 + 2 ) (p2 p1 ) . (4.36)
2
Let
p2 2 = R2 T2 (4.37)
be the state equation of the burn gases, and 2 = cp2 /cv2 the ratio of their capacities.
Furthermore, let x = 2 /1 and y = p2 /p1 the specic volume and pressure of the
burnt gases, referred to the corresponding values for the unburnt gases, and
( )( )
2 1 Q 22 2 1
q=2 + , (4.38)
2 + 1 p1 1 2 + 1 2 + 1
where = 10 /1 and 10 is the specic volume that would correspond to the burnt
gases at the pressure p1 and temperature T1 of the unburnt gases, that is,
R2 Mu
= = , (4.39)
R1 Mb
where Mu and Mb are the average mole masses for the unburnt gases and for the
combustion products respectively.
4.5. APPLICATIONS 101
By substituting these values and that of the temperature given by Eq. (4.37)
into Eq. (4.36), the following expression, for the Hugoniot curve, is obtained
2 1
xy = q + (y x) . (4.40)
2 + 1
Therefore, the Hugoniot curve is an equilateral hyperbola, with an asymptote parallel
to the axis y for x0 = (2 1) / (2 + 1). This curve cuts axis x at the point x1 =
q (2 + 1) / (2 1).
Let us see how the Chapman-Jouguet points are determined. The equation of
an isentropic for the burnt gases is
yx2 = const. (4.41)
Therefore, the Chapman-Jouguet points are obtained by solving the system formed
by Eq. (4.40) and by the equations that express the tangency condition of the curves
Eq. (4.40) and Eq. (4.41), namely,
(2 + 1) y + (2 1) y
= 2 . (4.42)
(2 + 1) x (2 1) x
The solution of Eq. (4.40) and Eq. (4.42) gives for x and y the following equations
( )
y 2 (2 + 1)q (2 1) y + q = 0, (4.43)
1( )
x2 (2 + 1)q + (2 1) x + q = 0. (4.44)
2
When solving these equations the root x1 < 1 must be assign to the root y1 > 1 and
root x2 > 1 to the root y2 < 1.
In virtue of Eq. (4.8), the propagation velocity is

v1 1 y1
= , (4.45)
a1 1 1 x
where a1 is the sound velocity of the unburnt gases, and 1 the ratio of their heat
capacities.
Similarly, the velocity of the burnt gases with respect to the wave front is

v2 x y1
= . (4.46)
a1 1 1 x
The corresponding Mach number M2 is

( )
1 x y1
M2 = . (4.47)
2 y 1 x
By subtracting from the left hand side of Eq. (4.42) the right hand side multi-
plied by two, we obtain
y y1
2 = , (4.48)
x 1x
which, together with Eq. (4.47), show that at the Chapman-Jouguet points, the velocity
of the burnt gases with respect to the combustion point is equal to the sound velocity.
For example, by taking the typical values p1 = 1 atm, 1 = 1000 cm3 /gr,
T1 = 300 K, 1 = 2 = 1.4, Q = 1000 cal gr1 , cp2 = 0.46 cal gr1 K1 and = 1
the following is obtained for the Hugoniot curve
1
xy = 15.23 + (y x) . (4.49)
6
This curve has been represented in Fig. 4.7. The branch of detonations (Fig. 4.7(a))
corresponds to the values of x in the interval
1
x 1. (4.50)
6
The branch of deagrations (Fig. 4.7(b)) corresponds to the interval
13.2 x 91.4 . (4.51)
The Chapman-Jouguet points for detonations and deagrations are, respectively,
Detonations: x1 = 0.58, y1 = 35.7 .

Deagrations: x2 = 24.82, y2 = 0.45 .
The corresponding propagation velocities are
v1
Chapman-Jouguet detonation: = 7.68 .
a1
v1
Chapman-Jouguet deagration: = 0.126 .
a1
4.6 Remarks
If the variation of the chemical composition along the Hugoniot curve is taken into
account, that is the inuence of the dissociation of the combustion products on the
characteristics of the wave, the problem is far more complicated. To attain a solution
one must resort to the equations of thermodynamic equilibrium, which must be com-
bined with the state equations and the invariants across the wave. Then, the Hugoniot
curve must be drawn point by point. For additional information see Ref. [4].
4.7. INDETERMINACY OF THE SOLUTION 103
y
x0 =1/6
125
(a) Detonation branch
100
75
50
J
25
A
P
0
0.0 0.2 0.4 0.6 0.8 1.0 x
y
P B
1.0
(b) Deflagration branch
0.8
0.6
J
0.4
0.2
0.0
0 20 40 60 80 100 x
Figure 4.7: Hugoniot curve (x = /1 , y = p/p1 ) corresponding to p1 = 1 atm, 1 =

1000 cm3 /gr, T1 = 300 K, 1 = 2 = 1.4, Q = 1000 cal gr1 , cp2 = 0.46
cal gr1 K1 , = 1.
4.7 Indeterminacy of the solution
The study performed in this chapter allows the establishment of the different types of
combustion waves that are compatible with the principles of the conservation of mass,
momentum and energy. This has been attained by comparison between the initial and
nal state of the gases. In order to obtain this result it has not been necessary to take
into account the intermediate states undergone by the gas, within the wave. Thus, the
problem is essentially simplied. However, if the inuence of the intermediate states
is neglected the result obtained is incomplete, since the propagation velocity of the
combustion wave remains undetermined, whilst experience shows that it takes a well
dened value for each different case. In order to eliminate the said indeterminacy and
select among the combustion waves obtained herein those that will actually occur, it
is necessary to analyze the internal structure of the wave and see the mechanism that
propagates the reaction to the unburnt gases. This study will be the subject of the
following chapter. Therein, it will be seen that:
1) The weak detonations and strong deagrations are impossible.

2) The detonations physically observed must be of the Chapman-Jouguet type.
3) Deagrations propagate with a well dened velocity.
References
[1] Crussard, L.: Ondes de choc et onde explosive. Bulletin de la Societe de

lIndustrie Minerale, Vol. VI, 1907.
[2] Chapman, D. L.: On the rate of explosion in gases. Phylosophical Magazine,
Jan. 1899.
[3] Jouguet, E.: Sur la propagation des reactions chimiques dans les gaz. Journal
de Mathematiques Pures et Appliquees, 1905-6.
[4] Taylor, J.: Detonation in Condensed Explosives. Oxford University Press, 1952.
Chapter 5
Structure of the combustion

waves
5.1 Introduction
The different types of combustion waves compatible with the principles of the con-
servation of mass, momentum and energy throughout the wave have been previously
established in chapter 4. Table 4.1 of that chapter summarizes the results of the said
analysis. These results where attained simply by comparing the initial and nal state
of the gases. The conclusions are correct provided that the wave thickness is small
enough, so that its geometrical shape has no inuence upon the transformations that
occur within the wave.
Owing to its nature, the aforementioned analysis is necessary incomplete, due

to the fact that the possible inuence of the intermediate states is ignored. Therefore,
it is not to be expected that all the types of combustion waves compatible with the said
principles of conservation will actually occur. On the contrary, when analyzing the
internal mechanism of the propagation of the process, the existence of new limitations
is to be expected. It should also be expected that such limitations will enable us to
eliminate the indeterminacy, mentioned at the end of the preceding chapter, and to
select from all the waves compatible with the laws of conservation the one which will
actually occur in each case.
To make this point clear, it is necessary to take into consideration the internal
structure of the wave. This is done in the present chapter, following a similar pro-
cedure to the one used in Gas Dynamics [1] when studying the internal structure of
shock waves.
105
106 CHAPTER 5. STRUCTURE OF THE COMBUSTION WAVES
For simplicity, the study herein will be restricted to the case of a perfect gas
with heat capacity and number of moles constant throughout the wave. The limit-
ing solutions, obtained by assuming that a characteristic time of the thermodynamic
transformations is very small compared to a characteristic time of the chemical trans-
formations (as explained in 3), will be carefully analyzed. This plausible assumption
enables an essential simplication of the equations, favouring the study of the proper-
ties of the corresponding solutions. This analysis will follow, mainly, the line of von
Karmans reasoning [2]. More detailed studies, for example those of Friedrichs [3] and
Hirschfelder [4], which take into account the inuence of the terms neglected herein,
show that the discrepancy between both cases is very small. Therefore, the method
followed in this study is well justied. Furthermore, these studies allow one to derive
conclusions for the cases in which, due to the existence of abnormal reaction rates,
the simplied treatment developed herein is not applicable.
In the following analysis special attention is given to the study of detonations

(see 5), since deagrations are the subject of a detailed analysis in the following
chapter.
5.2 Wave equations
Let us consider an indenite plane wave which propagates in undisturbed uniform

combustible mixture. We shall adopt a reference system xed to the wave were the x
axis is parallel to the propagation direction and positive towards the burnt gases. With
respect to this reference system, the process is stationary and x is the only independent
variable. The values of x vary from x = for the unburnt gases to x = + for
the burnt gases.
For simplication it will be assumed that the following considerations are sat-
ised throughout the wave:
1) The mixture behaves as a perfect gas.

2) The heat capacity at constant pressure cp is independent from the temperature
and composition of the mixture.
3) The ratio of the heat capacities is independent from the composition of the
mixture.
4) The chemical composition of the mixture at each point of the wave is determined
by only one chemical parameter. We shall adopt as chemical parameter the de-
gree of advancement of the combustion, measured by the mass fraction (x) of
the gas that has burnt when the point x of the wave is reached. Due to the action
5.2. WAVE EQUATIONS 107
of diffusion, this degree of advancement of the combustion differs from the mass
fraction Y (x) of the burnt gases that denes the mixture composition at point x,
that is, the fraction Y (x) of burnt gases that would be obtained by analyzing the
composition of a gas sample taken from the said point.
The simplifying assumptions previously stated do not limit the extent of the
study that follows, which is of a qualitative nature. On the other hand, these assump-
tions simplify calculations.
The wave equations are obtained by particularizing the general equations of

continuity, momentum and energy1 to the case of a one-dimensional stationary motion
(/y = /z = /t = 0). When this is done, and in addition to the preceding
assumptions are taken into account, the following system of equations is obtained:
a) Continuity equation.
d (v)
= 0, (5.1)
dx
that is
v = m, (5.1.a)
where m is the mass ow normal to the wave, per unit surface.

b) Continuity equation for the burnt gases. Since the mass ow per unit surface is
v and the burnt fraction is , the mass ow of the burnt gases at section x is
v. Its variation with x is due to chemical reactions. The equation that gives
this variation is
d (v)
= w, (5.2)
dx
where w is the reaction rate. In virtue of (5.1.a), equation (5.2) can also be
written in the form
d
m = w. (5.2.a)
dx
c) Momentum equation.
( )
dv dp 4 d dv
v = + . (5.3)
dx dx 3 dx dx
This equation can be immediately integrated by taking into account (5.1.a), giv-
ing
4 dv
p + mv = i, (5.3.a)
3 dx
where i is an integration constant that must be determined by the boundary con-
ditions.2
1 Seechapter 3.
2 Inequations (5.3) and (5.4) it has been assumed that the second ` viscosity
coefcient of the mixture
(see chapter 2) is zero. When not, it is sufcient to substitute for + 43 0 in (5.3) and (5.4).
d) Energy equation.
( ) ( ) ( )
d 1 2 d dT 4 d dv
v v + cp T q = + v , (5.4)
dx 2 dx dx 3 dx dx
which can be immediately integrated, giving

( )
1 2 dT 4 dv
m v + cp T q v = me, (5.4.a)
2 dx 3 dx
where e is an integration constant which, like i, must be determined by the

boundary conditions.
e) Diffusion equations.
( ) ( )
d dY dY d
D = v , (5.5)
dx dx dx dx
which can be integrated and gives
D dY
Y + = f, (5.5.a)
m dx
where f is an integration constant.
The previous system must be completed with the state equation. It can easily be
proven that assumptions 2) and 3) imply that the average mole masses of the unburnt
and burnt gases are equal. Therefore, the particular constant Rm = R/Mm of the gas
is independent from its composition, and the state equation valid throughout the wave
is
p
= Rm T, (5.6)

where Rm is independent from Y .
Equations (5.1.a), (5.2.a), (5.3.a), (5.4.a), (5.5.a) and (5.6) form a system of
six equations for the six unknowns p, , T , v, and Y . With the adequate boundary
conditions, this system determines the possible solutions of the problem.
Let p0 , 0 , T0 and v0 , be the values for p, , T and v in the unburnt gases, and
pf , f , Tf and vf the corresponding values for the burnt gases. Furthermore, since in
the unburnt gases there are no combustion products, in them Y = = 0. Similarly,
in the burn gases, Y = = 1. Therefore, the looked-for solutions must satisfy the
following boundary conditions:
1) Unburnt gases,
x : p p0 , 0 , T T0 , v v0 , Y 0, 0. (5.7)
5.2. WAVE EQUATIONS 109
2) Burnt gases,
x + : p pf , f , T Tf , v vf , Y 1, 1. (5.8)
Such conditions imply, naturally, the following ones,

dp d dT dY d
x : 0, 0, 0, 0, 0. (5.9)
dx dx dx dx dx
Herein, the possibility of satisfying the previous boundary conditions, will not
be discussed since it is subjected to a careful study in the chapter dedicated to de-
agrations. Therein, it will be seen that the boundary conditions relative to the cold
boundary of the ame give rise to a problem which, so far, has not been satisfactorily
solved.
In the present study it will be assumed that the boundary conditions can be
satised and, therefore, that the problem is completely determined.
By taking the boundary conditions (5.7), (5.8) and (5.9) into the system of
equations (5.1.a), (5.4.a)) and (5.4.a), the following laws of conservation are obtained,
which agree with the invariants deduced in the preceding chapter
0 v0 = f vf , (5.10)
p0 + 0 v02 = pf + f vf2 , (5.11)
1 2 1
v + cp T0 + q = vf2 + cp Tf . (5.12)
2 0 2
The boundary conditions relative, for instance, to the burnt gases, make possi-
ble the elimination of constants i, e and f from equations (5.3.a), (5.4.a) and (5.5.a),
thus obtaining
4 dv
p + mv = pf + mvf , (5.13)
3 dx
(1 ) dT 4 dv (1 )
m v 2 + cp T q v = m vf2 + cp Tf q , (5.14)
2 dx 3 dx 2
D dY
Y + = 0. (5.15)
m dx
The discussion and analysis of the possible solutions of the previous system are
rather complicated. An example of a discussion for a similar system (but in which the
diffusion neglected) can be found in Friedrichss work [3]. As a result of his analysis,
Friedrichs concludes that except in the case where the reaction rate w is exceptionally
high (and takes a well dened value) weak detonations are impossible. Consequently,
the only possible detonations with a normal reaction rate are strong detonations and
Chapman-Jouguet detonations. As will presently be seen, the type of detonation that

occurs in each particular case depends on the boundary conditions on the burnt side of
the detonation wave. As for the deagrations, Friedrichs obtained the conclusion that
strong deagrations are impossible and that weak deagrations are possible only for a
well dened value of the propagation velocity. This value depends on the state of the
unburnt gases, on the reaction rate and on the transport coefcients of the mixture.
The discussion of the differential system of combustion waves is considerably

simplied by analyzing, as done by von Karman [2], the limiting system deduced from
the previous one by assuming that a characteristic time of the thermodynamic transfor-
mations, for example the average time between two molecular collisions, is very small
compared to a characteristic time of the chemical transformations, for example the av-
erage time between two collisions of molecules of different species under the required
circumstances for these molecules to react to one another. This assumption appears
justied if one considers that, in a mixture of reactants, of all the molecular collisions
only a few are accompanied by chemical transformations. In fact, for this to happen
a number of favourable circumstances must concur: for example, the molecules must
be of the adequate species, the orientation of the collision and the energy in certain
degrees of freedom must be the proper ones, etc.3
In the following, the solutions of the aforementioned differential system will

be discussed mainly from this stand-point, following von Karmans reasoning.
5.3 Characteristic times
As a characteristic t of the thermodynamic transformations the following ratio is

adopted

t = . (5.16)
p
This ratio is proportional to the average time between molecular collisions, which is
given by ratio `/v of the mean free path ` of the molecules to the average velocity v
of the molecular motion. In fact, the Kinetic Theory of Gases shows that is of the
form
v`, (5.17)
and that v is of the form

p
v . (5.18)

3 See chapter 1.
5.4. LIMITING FORM OF THE WAVE EQUATIONS 111
Now, by substituting the expressions of v and ` deduced from (5.17) and (5.18) into
the ratio `/v, it results that this ratio is proportional to t as dened by (5.16).
As a characteristic time c of the chemical transformations, the following ratio

is adopted

c = , (5.19)
w
that measures the average time between those molecular collisions successful in pro-
ducing chemical reaction.4
Let
t w
= = (5.20)
c p
be the ratio of the thermodynamic time t to the chemical time c . The assumption
that chemical transformations are very slow when compared to the thermodynamic
transformations is expressed by the condition that parameter is much smaller than
unity
1. (5.21)
The limiting form of the wave equations, obtained when taking assumption (5.21) into
the system deduced in the preceding paragraph, is given in the paragraph that follows.
5.4 Limiting form of the wave equations
In order to discuss the relative values of the various terms of the differential system
of the wave, we shall start by showing the inuence of and the ow Mach number

M = v/a = v/ p/.
For this purpose, let us rst eliminate x from (5.3.a), (5.4.a) and (5.5.a), making
use of equation (5.2.a), then
4 w dv
p + mv = i, (5.22)
3 m d
1 2 w dT 4 w dv
v + cp T q 2 v = e, (5.23)
2 m d 3 m2 d
Dw dY
Y + = 0. (5.24)
m2 d
4 See chapter 1.
Now, the system of equations (5.22), (5.23) and (5.24) can be written in the
following way, which evidences the inuence of and M ,
( )
4 1 dv
p 1 + M 2
= i, (5.25)
3 v d
( )
1 2 q 1 1 dT 4 1 1 dv
cp T 1 + M 2
= e, (5.26)
2 cp T P r M T d 3 v d
1 dY
Y + = 0, (5.27)
Sc M 2 d
where P r = cp / and Sc = / (D) are the Prandtl and Schmidt numbers, respec-
tively, for the mixture.
Since 1, the last term of the left hand side of equation (5.25) can be
neglected when compared to unity, thus obtaining, instead of equation (5.25)
( )
p 1 + M 2 = i. (5.28)
Similarly, in equation (5.26) one can neglect the last term of the left hand side. Since
the Prandtl P r and the Schmidt Sc numbers are of order unity,5 the order of magnitude
1 1 dT 1 dY
of terms of equation (5.26), and of equation (5.27) depends
P r M 2 T d Sc M 2 d
on the values of the ratio /M 2 . Here, there are two possible cases:
a) If the Mach number M of the ow is of the order of magnitude of unity or larger,

then /M 2 1, and the said terms can also be neglected.
b) Whereas if /M 2 are of the order one, then said terms must be preserved. In
this case, however, M 2 1, and M 2 can be neglected in equation (5.28)
1 2
and M can be neglected in equation (5.26). Therefore the two following
2
cases are obtained:
1) 1, M 2 1.
Will be designated case A. For this case, Eq. (5.28) is valid and will be
written in the form
p + v 2 = i. (5.28.a)
In Eq. (5.26) the fourth and fth terms of the left hand side disappear,
obtaining the following simplied equation
( )
1 2 q
cp T 1 + M = e, (5.29)
2 cp T
that can be written
1
cp T + v 2 q = e. (5.29.a)
2
5 See Hirschfelder, Curtiss & Bird: Molecular Theory of Gases and Liquids, John Wiley & Sons Inc.,
New York, 1954, p. 16.

5.4. LIMITING FORM OF THE WAVE EQUATIONS 113

Y = . (5.30)
Therefore in this case, the effects of viscosity, thermal conductivity and

diffusion can be neglected, and the mixture can be considered as an ideal
gas whose composition varies due to chemical reactions.
2) 1, /M 2 1.
Will be designated case B. As aforesaid, in this case M 2 1, and Eq
(5.28) is reduced to
p = i, (5.31)
that is to say, the combustion occurs in rst approximation at constant pres-
sure. In Eq. (5.26) the second and last terms of the left hand side can be
neglected. Then, the following simplied equation is obtained
( )
q 1 1 dT
cp T 1 = e, (5.32)
cp T Pr M 2 T d
which can be written in the following form, returning to the old variable x
by means of equation (5.2.a)
dT
cp T q = e. (5.32.a)
m dx
All the terms of equation (5.27) must be preserved. By returning to variable
x, equation (5.15) is obtained.
Of the two limiting cases determined herein, the former is represented by the
system of equations (5.1.a), (5.2.a), (5.28.a) and (5.30), that is, by
Case A: 1, M 2 1 .
v = m, (5.1.a)
d
m = w, (5.2.a)
dx
p + v 2 = i, (5.28.a)
1
cp T + v 2 q = e, (5.29.a)
2
Y = . (5.30)
As aforesaid, in this system, the inuence of viscosity, thermal conductivity

and diffusion has disappeared. As will presently be seen, the solutions corresponding
to this system represent detonation waves.
The second case is represented by the system of equations (5.1a), (5.2a), (5.31)
and (5.32a), that is, by
Case B: 1, /M 2 1 .
v = m, (5.1.a)
d
m = w, (5.2.a)
dx
D dY
Y + = 0, (5.15)
m dx
p = i, (5.31)
dT
cp T q = e, (5.32.a)
m dx
in which the inuence of the kinetic energy and viscosity has disappeared. This system
represents a combustion at constant pressure, as limiting case of the weak deagrations
that are physically observed.
In the following paragraphs both cases will be studied separately.
5.5 Detonations
Let us rst discuss the solutions represented by system A).
The three equations (5.1.a), (5.28.a) and (5.29.a), together with the state equa-
tion (5.6), allows one to express the variation laws of p, T and v as function of the
degree of advancement of the combustion, or of the mass fraction Y = of the
burnt gases in the mixture. This is the problem of heat addition in the ideal gas in
one-dimensional and stationary motion.6 In particular, the following equation for v is
obtained
i ( 1)
v2 2 v+2 (e + q) = 0, (5.33)
( + 1) m +1
which shows that to each value of satisfying the condition
( ( )2 )
1 2 i
0 e , (5.34)
q 2( 2 1) m
corresponds two different real values of v, given by the expression

( )
i ( 2 1) ( m )2
v= 1 12 (e + q) . (5.35)
( + 1) m 2 i
6 See chapter 3, paragraph 9.
5.5. DETONATIONS 115
Of these two velocities, one is subsonic and the other supersonic. In fact, the
critical velocity vcr , corresponding to the point dened by the value of the degree of
advancement of the combustion, is given, in virtue of (5.29a), by
2 2 ( 1)
vcr = (e + q) . (5.36)
+1
This value, however, is exactly the product of roots v1 and v2 of equation (5.33).
Consequently
2
v1 v2 = vcr , (5.37)
that is, of both velocities one is subcritical, that is to say subsonic, and the other super-
critical, that is to say, supersonic. Furthermore, the two values v1 and v2 correspond to
the velocities before and after a normal shock wave, since (5.37) is the Prandtl relation
for shock waves.7
F
v
B D

Figure 5.1: Schematic diagram showing the two possible velocities resulting from the heat
addition.
Figure 5.1 represents the pair of values corresponding to Eq. (5.33) for variable
. Their corresponding curve is a parabola. The upper branch of this parabola, AC,
corresponds to the supersonic velocities, and the lower branch, BC, corresponds to the
subsonic velocities. At point C, where both branches join, the velocity of the gases
with respect to the wave is equal to the sound velocity.
7 See R. Courant and K. O. Friedrichs: Supersonic Flow and Shock Waves. Interscience Pub. Inc., New
York, 1948, p. 147.

Let A and B be the two representative points of the two possible initial states
( = 0), compatible with the assumed values for m, i and e. Velocity is supersonic
in A and subsonic in B. The jump from A to B occurs through a shock wave, as
aforesaid. Since, as previously seen in the preceding chapter, the propagation velocity
of a detonation is supersonic, the representative point for the initial state of the mixture
is A. When increases, that is, when the reaction occurs, two alternatives are possible
either the representative point of the intermediate states of the mixture throughout the
wave moves along the upper branch AC (see Fig. 5.1), or else a shock wave produces
rst which changes the velocity from A to B and then, as the combustion progresses,
the representative point moves along the lower branch BC. A decision between both
alternatives can not be taken from purely hydrodynamic considerations. Let us analyze
both cases separately.
Suppose that the point moves on the upper branch, starting from A. This means
that the combustion initiates in the state of the unburnt gases. But in this state, the
temperature is small and the reaction velocity cannot be sufciently large to burn
the gases with the speed required by the detonation wave. Therefore, the detonation
must be initiated by a shock wave which, by making the gases pass from the state
represented by point A to the one represented by point B, compresses and heats the
gases, taking them to a state in which the reaction velocity can be sufciently large to
burn them with the required speed.
The ratio of the thickness of the shock wave to the thickness of the detonation
wave is measured by the ratio of the thermodynamic characteristic time t to the chem-
ical characteristic time c . Therefore, the thickness of the shock wave is very small
with respect to the thickness of the detonation wave.8 This means that while the gases
pass through the shock wave, the burnt fraction is insignicant. Thus the detonation
wave appears as formed by the shock wave, followed by a combustion wave.
The need of a shock wave to initiate the chemical reaction in the detonation
wave has always been acknowledged. This has been the stand-point, for instance, for
Vieille [5] and Jouguet [6] in 1900.9 These authors, however, have consider the deto-
nation always as a discontinuity, in which not only the shock wave is instantaneously
produced, but also the subsequent reaction. The ideas developed in the present study,
concerning the structure of the wave, belong to the modern theories on detonation
8 The study of the dynamic transformations within the shock wave cannot be performed with the system
obtained by neglecting the action of viscosity and thermal conductivity, whose action is essential within the
wave. See for example M. Roy: Structure de londe de choc et des ammes deagrantes. ONERA, Paris,
1952.
9 In the fundamental work of Jouguet [6] data can be found concerning the historical evolution of the
ideas relative to the classical theory of the detonation waves.

waves. Such theories were initiated by the works of Taylor in England, von Neumann
[7] in the U.S.A., Zeldovich [8] in the U.R.S.S. and Doring [9] in Germany.
Within the combustion zone that follows the initial shock wave, the gases move
along the lower branch of the curve shown in Fig. 5.1, starting from point B. The point
= 1, at which the combustion ends and the thermodynamic equilibrium is reached
after the wave, cannot lie at the right hand side of D. Therefore, the three following
cases can occur:
1) The combustion ends at a point such as E of the lower branch, in which the
velocity is subsonic. This case represents a strong detonation.
2) The combustion ends at the point C that separates the subsonic from the super-
sonic branch. In C the velocity of the burnt gases with respect to the detonation
wave is sonic. The corresponding detonation for this case is of the Chapman-
Jouguet type.
3) The representative point of the nal state is point F in the supersonic branch.
Consequently, the nal velocity is supersonic and the corresponding detonation
is weak.
We shall presently see that the last case cannot possibly occur. In fact, in order
to reach point F either one has to pass by point C, in which case the reaction between
C and F would be endothermic and therefore in contradiction with what happen in the
combustion, or else one must pass from B to D and from D to F by means of a jump
or expansion shock which is also impossible. Consequently, weak detonations are not
possible.
On the other hand, nothing opposes a strong detonation or a detonation of the

Chapman-Jouguet type. For a given case the occurrence of one or the other depends
on the boundary conditions after the wave. Let us see the inuence of these conditions.
When the detonation is strong, the velocity of the burnt gases, with respect to
the wave is subsonic. Any perturbation produced in the burnt gases will propagate
throughout them with sound velocity and can therefore reach the wave and alter its
nature. For example, an expansion of the burnt gases will reach the wave, reducing
its strength down to the point where the velocity of the burnt gases with respect to the
wave equals the sound velocity. In such a case the wave becomes insensible to the
perturbations that occur in the burnt gases. Such is the situation that normally occurs
in practice; for example, when the detonation propagates along a tube, starting either
at an open or closed end. Thus the detonations normally observed in practice are of the
Chapman-Jouguet type, due to the fact that these are the only stable ones with respect
to perturbations coming from the burnt gases.
The study of the aerodynamic eld subsequent to a detonation wave can be per-
formed by applying the Gas Dynamic methods.10 For instance, Zeldovich [10] has cal-
culated the aerodynamic eld following a detonation wave that propagates unchanged
from the closed end of a tube, initially full of detonant gas at rest. By neglecting the
friction and the heat lost through the walls of the tube, a solution is obtained which
propagates with the Chapman-Jouguet velocity. The detonation wave is followed by
an expansion region with an approximate length of 50% of the total length covered
by the detonation wave. The remaining burnt gases (approximately 40% of the burnt
mass) are at rest. By including the effect of friction and heat losses through the walls
of the tube, a stable Chapman-Jouguet wave is obtained, followed by an expansion
in which the direction of the motion of the burnt gases in the tube is reversed. Fol-
lowing the expansion there also exists in this case a region at rest in which the burnt
gases have cooled down to ambient temperature. There is photographic evidence of
the existence of this reversal motion in the expansion region.
The aerodynamic eld that follows a spherical detonation wave has been stud-
ied by G.J. Taylor [11] in England, and by Zeldovich in U.S.S.R. [12]. The calcula-
tions also demonstrate the existence of a solution which propagates unchanged with
the Chapman-Jouguet velocity. Immediately after the detonation wave there is a strong
expansion region, followed by a central nucleus at rest. Within the expansion region
more than 90% of the burnt mass is in motion. The spherical detonation waves have
been experimentally observed, for example, by Manson and Ferrie [13]. A strong det-
onation can occur, for example, when the detonation propagates inside a tube, starting
from an end closed by a piston, that moves after the wave with a subsonic velocity
with respect to the wave. In such a case the intensity of the strong detonation would
be determined by the compatibility condition obtained by expressing that the veloc-
ity of the burnt gases with respect to the wave must equal the velocity of the piston.
Then, when friction and heat losses through the walls of the tube are neglected, it
results that the strong detonation wave propagates unchanged throughout the mass of
unburnt gases.
The relation between pressure and density within the wave can be obtained
from Eqs. (5.1.a) and (5.28.a) by elimination of v. Thus resulting
m2
p+ = i, (5.38)

or else, introducing the state (p1 , 1 ) of the unburnt gases
( )
1 1
p p1 = m2 . (5.39)
1
10 See Courant and Friedrichs, ib., pp. 218 and 416 for plane and spherical waves, respectively.
This relation shows that the pressure at each point of the wave is determined
only by the corresponding value of the density. This relation plays here the same part
as, for example, in Gas Dynamics the relation p = C for the isentropic motions of
non-reacting gases.
In order to represent the result in the diagram (p, ), as done in chapter 3, the
specic volume = 1/ must be introduced in place of the density in Eq. (5.39),
which then takes the form
p p1 = m2 (1 ). (5.40)
This equation is represented in diagram (p, ), see Fig. 5.2, by a straight line joining
the representative points of the initial and nal states. On this line also lie all the
representative points of the intermediate states corresponding to the reaction zone.11
70
D
60
E
50
C Trayectory through shock wave
40
1
p/p
30 =1.0
J
20
=0.3 =0.7 E
10
=0
1 P
0.0 0.2 0.4 0.6 0.8 1.0
/1
Figure 5.2: Hugoniot curves from different values of .
Bringing forth in Eqs. (5.1.a), (5.28.a) and (5.29.a) the specic volume and the
initial conditions corresponding to the unburnt gases, there results
v v1
= , (5.41)
1
v2 v2
p+ = p1 + 1 , (5.42)
1
1 1
cp T + v 2 q = cp T1 + v12 . (5.43)
2 2
11 W. Michelson was the rst to postulate that within the reaction zone of the detonation wave the linear
relation (40) is veried.

The elimination of T , v, T1 and v1 between these equations and the state equa-
tions,
p = Rm T and p1 1 = Rm T1 , (5.44)
gives the following Hugoniot relation for each value of the degree of advancement
of the combustion
+1 1
(p p1 1 ) + (p1 p1 ) = q. (5.45)
2( 1) 2
This equation represents a family of hyperbolas. For each value of a hyper-

bola is obtained, which represents the locus of the possible states of the gas when
burnt fraction is and the initial state is (p1 , 1 ). The Hugoniot curve EJE is obtained
for the particular case = 1. This curve corresponds to all the possible states of the
burnt gases, see Fig. 5.2. Similarly, for = 0 the Hugoniot curve PD is obtained,
which corresponds to all the shock waves that can form in the unburnt gases at the
initial state represented by point P . Between both curves lie those corresponding to
the intermediate states. Two of them are shown in Fig. 5.2, corresponding to = 0.3
and = 0.7.
Now, let us consider, for example, a Chapman-Jouguet detonation. Equation

(5.40) corresponding to this detonation is represented by the straight line PJC. The
transformations that occur throughout the shock wave, preceding the combustion,
carry the gas from the initial state P to the state C after the shock wave with no combus-
tion. However, the trajectory of the gases through the shock wave is not represented
by the straight line PC. In fact, we have seen that the thickness of the shock wave
is very small and, as aforesaid, the thermal conductivity and viscosity therein cannot
be neglected. In particular, Eq. (5.40) must be substituted by the following relation,
deduced from (5.3.a), which takes into account the viscosity,12
4 dv
p p1 = m2 (1 ) + . (5.46)
3 dx
But since, through the shock wave the velocity decreases, dv/ dx is negative. There-
fore, the representative points of Eq. (5.46) lie below the straight line PC, as indicated
in Fig. 5.2.13
The representative states of the combustion that follow the shock wave, that
is, those corresponding to section BE, Fig. 5.1, are obtained by traversal segment CJ,
starting from C. The points at which this segment intersects the successive Hugoniot
12 Assuming that the transformations through the shock wave can be described by variables corresponding
to a continuum. See chapter 3, 1.

13 Hirschfelder et al. ib. page 810.
curves, represent the successive states of the mixture for the corresponding values of
the degree of advancement of the combustion.
Now, let us consider a strong detonation, which in Fig. 5.2 would be repre-
sented by a straight line such as PED. The initial shock wave produces a jump from
state P to state D. From D to E the reaction takes place and is accompanied by an
expansion and acceleration of the gases. The intermediate states are represented by
the points of segment DE.
Figure 5.2 also shows that weak detonations are not possible. In fact, once
point E is reached, point E corresponding to a weak detonation can only be reached
by means of an expansion wave between E and E, which is impossible, or else by
means of an endothermic reaction, corresponding to segment EE.
The temperature variation throughout the wave can be analyzed in the same
manner as that used for the pressure variations. For this purpose it is sufcient to
eliminate p and p1 from Eq. (5.40), by making use of (5.44). Thus, we obtain
( )2
T
= (1 + M12 ) M12 , (5.47)
T1 1 1
in which the value of the Mach number M1 of the unburnt gases has been made ex-
plicit. To each value of the Mach number M1 corresponds a different parabola, on
which lie the representative points of the successive states of the mixture within the
combustion zone of the detonation wave. In Fig. 5.3 two parabolas are shown. One
corresponds to the Chapman-Jouguet detonation and the other to a strong detonation.
The same letters are used to designate homologous points in Figs. 5.1 and 5.2.
The Hugoniot curves are obtained from Eq. (5.45) eliminating p and p1 by
means of Eq. (5.44), as has been done to obtain Eq. (5.47). There results
( )2 ( )
+1 T T 2 q +1
+ + = 0. (5.48)
1 1 T1 1 T1 1 c p T1 1 1
These curves are a family of hyperbolas. A hyperbola is obtained for each value of
. In particular, for = 1 one obtains the Hugoniot curve of the burnt gases and for
= 0 the curve of the shock waves of the unburnt gases. Both have been taken into
Fig. 5.3. Their intersection with curve of Eq. (5.47), representative of the intermediate
states, determines section CJ or DE, corresponding to the said intermediate states. The
state corresponding to a given fraction of burnt gases is given by the intersection of
curve of Eq. (5.48), corresponding to this fraction, with section CJ, or DE, as shown
in Fig. 5.3 for the values = 0.3 and = 0.7.
The preceding considerations give an idea of the structure of a detonation wave.

The initial shock wave compresses, decelerates and heats suddenly the gasses, within a
zone of negligible thickness compared to that of the detonation wave, where no appre-
ciable chemical reaction occurs. After the shock wave the combustion is initiated. The
gases expand and accelerate and the pressure decreases. Temperature rises due to the
reaction. For Chapman-Jouguet detonation and slightly strong detonations, the maxi-
20
18
16
E
=1
14
J
K E
12
=0.7
1
D
T/T
10
8
=0.3
C
6
4
=0 Trayectory through shock wave
2
P
0
0.2 0.4 0.6 0.8 1.0
/
1
Figure 5.3: Hugoniot diagram in variables (T, ).
15 50 1.0
C P/P K
13 40 1
J 0.8
11 30 J 0.6
T/T
P/P1
1
1
T/T
J
9 20 0.4
/1
C
7 10 0.2
C
5
0.0 0.2 0.4 0.6 0.8 1.0

Figure 5.4: Typical values of T , p and in a Chapman-Jouguet detonation as a function of
the fraction of burnt gases.
5.6. DEFLAGRATIONS 123
mum temperature is reached shortly before the combustion ends (as is clearly shown
in Fig. 5.3), after which a slight temperature drop (100 to 200 C for typical cases)
takes place due to the fast expansion of the combustion products. For the very strong
detonations, like DE represented in Fig. 5.3, the maximum temperature is reached at
the end of the combustion. These results appear in Fig. 5.4, which shows the values
corresponding to a typical case for a Chapman-Jouguet detonation.
5.6 Deagrations
We have said that the deagrations, also known as ames, will be the subject of a
careful study in the following chapter. Consequently, herein we shall only include a
few brief considerations concerning their possible existence, structure and propagation
velocity.
In the preceding chapter we have seen that the propagation velocity of a dea-
gration wave is always subsonic. Thereby it cannot be ascertained that in the deagra-
tion waves the condition M 2 will be satised, and the system that must be used
for the study of its structure is B, at least within the region of the ame close to the
cold boundary.14 Let us consider separately the weak and strong deagrations, and let
us start by demonstrating that the latter cannot exist.
In the strong deagration waves, the nal velocity of the gases is supersonic.
Consequently, at least in the nal zone of such waves, the condition M 2 1 is sat-
ised, and system A can be used. In particular in this zone, the variation law for the
velocity of the gases with respect to the wave is given in Fig. 5.1. However, due to
the inuence of thermal conductivity and diffusion, this law can differ within the zone
close to the cold boundary. Hence, in a strong deagration wave, the variation of the
velocity of the gases through the wave must follow a law as the one represented in
Fig. 5.5, where the curve in Fig. 5.1 representing the limiting solution of system A has
also been represented by a dash line. The velocity of the unburnt gases is represented
by point P . Starting from this point, the gases accelerate. When their Mach number
is such that condition M 2 is satised, the curve overlaps the dash line of the
limiting solution (in the gure this happens at point B). If now one must reach point
D, corresponding to the nal state where the velocity is supersonic, this can only be
achieved by passing through point C where the velocity is sonic. But the jump from C
14 It has been shown in the preceding chapter that the propagation velocity of a deagration is always equal
to or smaller than that corresponding to the Chapman-Jouguet deagration. In this case the Mach number
corresponding to the propagation velocity is always much smaller than unity. Thereby, the propagation
Mach number of the deagrations observed is always much smaller than unity. In practice, of the order of
103 .
to D could only be attained through an endothermic reaction, and this kind of reaction
is impossible in a combustion process. As a consequence, the non-existence of the
strong deagrations is concluded.
C
v

Figure 5.5: Schematic diagram showing the variation of the velocity of gases in a deagra-
tion.
On the other hand, since for a weak deagration the nal velocity is subsonic,
nothing is opposed to its existence.
In the weak deagrations diffusion and heat conduction are important and the
system to be used is system B. Diffusion and heat conduction are responsible for
the propagation of the combustion to the unburnt gases, activating their reaction by
diffusion of the active centers (atoms, radicals, etc.) and by heating. As they heat, the
gases expand and accelerate, producing a slight pressure drop throughout the wave.
Fig. 5.6 shows qualitatively the structure of a deagration wave bringing forth, by
comparison with Fig. 5.3, the difference in the mechanism that propagate the process
in each case. The system B of differential equations that must be integrated in order
to obtain the structure of the deagration wave can be simplied by eliminating x, as
done in 4. Thus obtaining
p Dw dY
= Y , (5.49)
Rm T m2 d
w dT
= cp T q + e. (5.50)
m2 d
In the deduction of this system, the state equation p = Rm T has been used in order
to eliminate the density from the diffusion equation.
5.6. DEFLAGRATIONS 125
8
v/v1=T/T1
7
5 P/(1/2) v2
1 1
T /T =8
2 1
4
D c /=1.5
p
3
2
Y
1
0
0 0.2 0.4 0.6 0.8 1

Figure 5.6: Typical values of T , v, Y and p in a deagration as a function of the fraction
of burnt gases .
In this system the reaction rate w is a known function of Y and T . The pressure
p is constant in virtue of Eq. (5.31). One must look for a solution of this system, in
the interval 0 1, satisfying the following boundary conditions
=0: Y = 0, T = T0 , (5.51)
=1: Y = 1, T = Tf . (5.52)
Since this is a system of two rst order equations, the four conditions (5.51)
and (5.52) cannot be satised unless parameters e and m take well dened values.
Parameter e is given by the expression: e = q cp Tf , which results from expressing
the condition w = 0 for = 1, that is, at the end of the combustion. As for m it must
taken well dened value, so that the previous differential system to be compatible.
This means that weak deagrations can only propagate with a well dened velocity
which depends on the state of the mixture, its reaction rate and its coefcients of ther-
mal conductivity and diffusion. The problem of determining the propagation velocity
of the ame through a combustible mixture appears, thus, as an eigenvalue problem.
The solution to this problems will be studied in the following chapter.
The same conclusions are reached when taking into account the inuence of all
the terms in the differential equations of the combustion wave, instead of considering
a limiting solution as done herein.
5.7 Transition from deagration to detonation
The preceding study presents deagrations and detonations as independent phenom-

ena. Both, deagrations and detonations can be initiated by means of various proce-
dures. For example, deagrations can be initiated by means of an electric spark within
a combustible mixture, and detonations by making a shock wave cross the detonant
mixture. Such procedure was applied, for example, by Fay [14] using a shock tube.
However, a detonation can also be produced starting from a deagration by compres-
sion of the unburnt gases and acceleration of the combustion wave. This can occur, for
example, when a detonant mixture lling a tube is ignited at the closed end. In such
a case a deagration wave initiates at the ignition point. This deagration accelerates
as it propagates along the tube, producing a succession of intermediate states, non-
stationary, which end with the establishment of a stable Chapman-Jouguet detonation,
Hereinafter, we shall restrict ourselves to a qualitative description of the process, fol-
lowing the lines of Zeldovichs work [8], who carefully studied this phenomenon. For
this purpose the ame front will be considered as a discontinuity that travels along the
tube, as done in the preceding chapter.
Let be the propagation velocity of the ame through the unburnt gases, and
S the area of the cross-section of the tube. If the ame front is plane and normal to the
tube axis, the mass of the gases burnt per unit time is 1 S, where 1 is the density
of the unburnt gases before the ame. Before burning, this mass occupies a volume
S. Since the gases expand as they burn, the volume that must occupy the said mass
at the same pressure15 is nS. Here n is the ratio of the temperature of the burnt gases
to temperature of the unburnt gases. Consequently, the increase in volume due to the
combustion is (n 1)S. Such an increase in volume sets the unburnt gases into
motion in front of the combustion wave to empty the necessary space. The total mass
of unburnt gases is not set into motion simultaneously, but progressively by a pressure
wave that travels through the mass with a velocity close to the velocity of sound. The
situation is shown schematically in Fig. 5.7. The gases are at rest within the region
between O and the ame front A. Between the front A and the pressure wave B, lie the
unburnt gases, compressed and moving towards the right-hand side. Starting from the
pressure wave B and towards tho right-hand side, lie the unburnt gases in the initial
state and at rest. The intensity of the pressure jump across the pressure wave is such
that the burnt gases are at rest. Let vp1 and vp2 be the velocities of the unburnt gases
before and after the pressure wave, measured with respect to this wave. The velocity
at which the pressure wave travels along the tube is vp1 . The velocity of the unburnt
15 The pressure drop across the ame is negligible.

5.7. TRANSITION FROM DEFLAGRATION TO DETONATION 127
O A B
n P
2
P
1
P vp1 vp2 = (n1) vp1

3
P P
3 2
P
P, v 1
(n1)
Figure 5.7: Schematic diagram of the ame propagation in a closed tube.
gases after the wave, with respect to the tube, is vp1 vp2 . The condition that the burnt
gases must be at rest is
vp1 vp2 = (n 1), (5.53)
as it can easily be veried. The velocity va at which the ame travels along the tube is
va = n, (5.54)
where is the propagation velocity of the ame throughout the gases in their state
after the pressure wave. Such propagation velocity differs from that corresponding
to the state of the gases before the ame. This state of affairs can be maintained
unchanged along the tube. For the detonation to occur it is necessary to activate the
combustion, increasing the burnt mass per second. Such an increase is necessary in
order to increase the intensity of the pressure jump across the pressure wave, up to the
point of self-ignition of the mixture after the wave, needed to maintain the detonation.
The slight compression of the mixture, produced by the compression wave, is not
sufcient to increase, appreciably, the propagation velocity of the ame. Therefore,
the increase of burnt mass per second must be due to other effects. The explanation
can be found in the inuence of the tube walls. In fact, due to friction, the distribution
of velocities in the cross-section of a tube is not uniform but maximum at the axis
and zero on the walls. Due to this non-uniformity of the velocity distribution, the
ame front curves increasing its surface. Due to this surface increase, the burnt mass
per second increases and therewith the velocity of the gases and the intensity of the
pressure jump, producing an interaction of both effects such that when the tube is
sufciently long it leads to the establishment of the detonation wave. This stand-
point is corroborated by the fact that for tubes with rough walls or small diameter the
detonation is soon establish because both effects increase the inuence of friction. On
the other hand, if the mixture is ignited at the open end of the tube, the detonation is
established with greater difculty as the burnt gases can escape through the open end
and the pressure wave that precedes the combustion wave is much weaker. In this case,
a longer tube is necessary so that the pressure wave may reach the required intensity
for the detonation to occur.
For the case previously analyzed, A.K. Oppenhein [15] has studied the succes-
sion of intermediate states that must be produced in the transition from deagration to
detonation by acceleration of the deagration wave up to the point where it overtakes
the shock wave. He demonstrated his results by means of a hydraulic analogy.
References
[1] Backer, R.: Stosswelle und Detonation. Zeitschrift fur Physic, Vol. 8, 1922.
[2] von Karman, Th.: Aerothermodynamics and Combustion Theory. LAerotecnica,
Vol. XXXIII, Fasc. 1st., 1953, pp. 80-86.
[3] Friedrichs, K. O.: On the Mathematical Theory of Deagrations and Detona-
tions. NAVORD Report 79-46, Institute for Mathematics and Mechanics, New
York University, 1946.
[4] Hirschfelder, J. O., Curtiss, C. F. and Campbell, D.E.: The Theory of Flames
and Detonations. Fourth Symposium (International) on Combustion, Williams
and Wilkins Co., Baltimore, 1953, pp. 190-211.
[5] Vieille, P.: Role des Discontinuites dans la Propagation des Phenomenes Ex-
plosifs. Comptes Rendus des Seances de 1Academie des Sciences, Vol. CXXXI,
1900, p. 413.
[6] Jouguet, E.: Mecanique des Explosifs, Paris, 1917, p. 325.
[7] Neumann, J. von: Progress Report on the Theory of Detonation Waves. N D R
C, Division B, 0 S R D, No. 549, 1942.
[8] Zeldovich, Y. B.: On the Theory of Propagation of Detonation. Journal of
Experimental and Theoretical Physics, Vol. 10, 1940, p. 542, Translated as
NACA Technical Memorandum No. 1261, 1950.
[9] Doring, W.: Annalen der Physic, Vol. 43, 1943, p. 421.
[10] Zeldovich, Y. B.: Theory of Combustion and Detonation of Gases. A9-T-45,
Air Documents Division, Air Material Command, U S A F .
5.7. TRANSITION FROM DEFLAGRATION TO DETONATION 129
[11] Taylor, G. I.: Detonation Waves. Ministry of Supply, Explosives Research Com-
mittee, R.C. 178, A.C. 639, 1941.
[12] Zeldovich, Y. B.: The Distribution of Pressure and Velocity in the Products of
a Detonation Explosion. Journal of Experimental and Theoretical Physics, Vol.
12, 1942, p. 389. Also ref. [8].
[13] Manson, N. and Ferrie, P.: Contribution to the Study of Spherical Detona-
tion Waves. Fourth Symposium (International) on Combustion, Williams and
Wilkins Co., Baltimore, 1953, pp. 486-94.
[14] Fay, J. A.: Some Experiments on the Initiation of Detonation in 2H2 -O2 Mix-
tures by Uniform Shock Waves. Fourth Symposium (International) on Combus-
tion, Williams and Wilkins Co., Baltimore, 1953, pp. 501-507.
[15] Oppenheim, A. K.: Gasdynamics Analysis of the Development of Gaseous Det-
onation and its Hydraulic Analogy. Fourth Symposium (International) on Com-
bustion, Williams and Wilkins Co., Baltimore, 1953, pp. 471-480.

Chapter 6
Laminar ames
6.1 Introduction
It was demonstrated in chapter 4 that through a combustible mixture of gases two

types of combustion waves may propagate: detonation waves and deagration waves
or ames. Chapter 5 includes the analysis of the nature of such waves, their structure
and the conditions which determine their propagation velocity. This analysis proved
that while in detonation waves the propagation velocity is determined only by a condi-
tion of mechanical stability within the zone of burnt gases, which simplies extremely
its determination, when dealing with ames their propagation velocity is determined
by an internal equilibrium between the processes of chemical reaction, heat trans-
fer (conductivity) and transport of the chemical species (diffusion), which makes the
study specially difcult.
The initial attempt to establish a theory on these bases, and in particular the
deduction of the rst formula for the computation of the propagation velocity of a
ame through a given mixture, which is the fundamental problem of combustion, is
own to Mallard and Le Chatelier [1]. Their theory is purely thermal and it establishes
a primitive balance between the heat received by the gases through conduction, when
an assumed ignition temperature is reached, and the heat content of these gases. The
concept of reaction velocity is not applied in this theory.
The work by Mallard and Le Chatelier was the starting point for the develop-
ment of the thermal theories of the ame, so called because they did not include the
diffusion of species.
The idea of chemical reaction velocity, incorporated by Crussard [2] to this

thermal model, was a denite step forward in the progress of the theory. From there
131
132 CHAPTER 6. LAMINAR FLAMES
on, the interest of the later works was centered both on the use of more realistic ex-
pressions for the reaction velocity which considered the inuence of the concentration
of species and of the temperature of the mixture, as well as on the development of
approximate methods, analytical and numerical, for the integration of the ame equa-
tions. A critical discussion on thermal theories, whose development and application
have been pursued up to our days, may be found in the work by Goward and Payman
[3]. Frank-Kamenetskii-Semenov [4], Boys-Corner [5] and von Karman [6] have spe-
cially contributed to the development of analytical methods and Hirschfelder and his
collaborators [7] to the expansion of numerical ones.
Initially these studies were performed with a very simplied kinetic model
which considered only two different chemical species (reactants and products) and
they applied a law of reaction which reected approximately the relationship between
the reaction velocity of the mixture and its temperature and composition at each point.
Later on these models were improved by taking into account the inuence of the dif-
ferent species and chemical radicals which act in the process.
The rst attempt to include the inuence of both diffusion and chemical rad-
icals on the velocity of the ame is own to Lewis and von Elbe [8] in their study of
ozone decomposition ame, which will be given further consideration in 13. This
study established the bases for the development of a complete theory on ames. Other
authors followed their steps, trying to nd a more accurate and complete formulation
for the problem. At the same time the analytical and numerical methods of the thermal
theory were extended to the new formulation. A very complete analysis of the differ-
ent theories and methods available, as well as of the assumptions used in expanding
them, may be found in the work by Evans [9].
In the last years a great effort has been applied to the development of the theory
of the ame, which has resulted in the obtaining of a complete formulation of the
problem and its boundary conditions. At the same time a great amount of attention was
given to the development of analytical and numerical methods, very approximated,
for the solution of the resulting equations, while the theory was applied to a certain
number of practical cases, although very few due to the fact that it is very difcult to
know the physico-chemical constants of the process and of the chemical scheme of
the reactions. It is practically impossible to try to give a complete bibliography of the
work achieved in these last years since it is very abundant and scattered, specially in
the United States, United Kingdom, and Russia. Therefore, it is far more practical
to refer the reader to the Proceedings of the International Symposia on Combustion,
in special starting from the third one which took place in 1948, as well as to the
volumes of the Selected Combustion Problems published by AGARD, containing the
6.1. INTRODUCTION 133
papers presented at the two International Colloquia sponsored by this organization in

1954 and 1956. The work by von Karman [10] is a complete review of the state of
knowledge on this problem up to 1956.
At the same time that this theory which considers the ame as a wave propa-
gating through a continuous medium was developed, other theories have attempted to
emphasize only partial aspects of the phenomenon. From them the most representa-
tive is the one proposed by Tanford and Pease on the diffusion of radicals from the hot
boundary [11]. It has also been objected that ames cannot be treated by the meth-
ods of the Mechanics of Continua. This objection has originated such theories as the
one proposed by van Tiggelen [12], who applies a method analogous to that used by
Semenov in studying chain reactions.
At present, however, it seems clearly established and widely accepted that the
correct method which should be applied to this problem is the one expanded in the fol-
lowing paragraphs, and that its difculty reduces to the use of the adequate chemical
kinetic scheme for each case; the knowledge of the corresponding physico-chemical
constants and of the transport coefcients of the mixtures; and nally to the obtain-
ing of satisfactory methods of solution of the difcult mathematical problem of the
integration of the ame equations.
In addition to this progress of theory, a substantial improvement has been

achieved in the eld of experimental techniques to measure the propagation veloc-
ity of the ame. An increasing number of data is now available on the inuence of
several parameters such as the composition, pressure, initial temperature of the mix-
ture, etc., on the ame propagation velocity. As a bibliography for the study of these
techniques, their limitations and applicability we recommend the works by Jost [13],
Gaydon and Wolfhard [14] and specially the one of Ref. [15]. Linnet [16] has pub-
lished a review on these methods, in which he classies the experimental techniques
most commonly used into the following ve groups:
1) Method of the bunsen burner (Gouy and Michelson). The most universal, of
which a great number of variations are used.
2) Method of the tube (Coward-Hartwell, Gerstein, etc). Not considered reliable
enough.
3) Method of the spherical ame in a constant volume bomb (Fick, Lewis-von
Elbe). Not frequently used and difcult to observe.
4) Method of the soap lm or spherical ame at constant pressure. Not frequently
used and limited to mixtures which are not sensitive to the inuence of water
vapor.
5) Method of the at burner (Egerton-Pauling, Spalding-Botha). Indicated for mix-

tures with a low ame velocity.
At the same time, several attempts have been made, with limited success, to
measure the distribution of certain variables through the ame in order to analyze
its structure. Information of this matter may be found in the above mentioned work
by Linnet and in the Proceedings of the 6th International Symposium on Combus-
tion. Initially, attempts were made to obtain temperature proles whose results show
a fair agreement with theoretical predictions. Later on, attempts were made to ob-
tain distributions of the concentration of the main species, and nally, of the radicals.
These observations are difcult due to the small thickness of the ame. Although it
may be increased by reducing pressure, then the perturbation effects of the measuring
instruments and of radiation become more important, and it is difcult to know the
magnitude of the corrections which must be introduced into the results obtained so
that the values be exact.
In the following paragraphs of this chapter, we shall deduce and discuss the
ame equations, its boundary conditions and the methods of integration most com-
monly used. The application of the theory to some practical cases will be performed
in 13, 14 and 15.
6.2 Equation for the combustion wave in the case of

two chemical species
The equations for the propagation of a plane stationary combustion wave were de-
duced in the preceding chapter, system B, under the following assumptions:
1) There are only two different chemical species, reactants and products.
2) The mixture behaves like a perfect gas.
3) The specic heat of the mixture is independent from its composition and tem-
perature.
The system obtained under these conditions, referred to the axis advancing with the
wave, and preserving the notation previously used, is as follows:
v = m (6.1)
b) Chemical reaction equation.

d
m = w(Y, p, T ) (6.2)
dx
6.3. BOUNDARY CONDITIONS 135
c) Diffusion equation.
dY
D = m(Y ) (6.3)
dx
d) Momentum equation.
p = const. (6.4)
e) Energy equation.
dT
cp T q =e (6.5)
m dx
f) State equation.
p
= Rg T (6.6)

In this system, m and e are two constants, whose values will result from the
boundary conditions. The elimination of through (6.6) reduces the system to three
rst order differential equations (6.2), (6.3) and (6.5) with three unknown , Y and T .
6.3 Boundary conditions
In order for the solution of this system to represent a stationary wave, it is necessary
that it satises the following boundary conditions, which insures the transition from an
uniform state of the unburnt gases before the wave to an uniform state of the products
in chemical equilibrium after it,
Unburnt gases, x : Y 0, 0, T T0 . (6.7)

Products, x + : Y 1, 1, T Tf . (6.8)
These conditions imply, as well, the following
dY d dT
x : 0, 0, 0. (6.9)
dx dx dx
The rst of conditions (6.9) is satised by virtue of (6.3), since and Y take
the same values both in x = + and in x = , by virtue of (6.7) and (6.8).
The third condition (6.9) is also satised by virtue of (6.5) when the following
values are assigned to q and e
q = cp (Tf T0 ),
(6.10)
e = cp T0 ,
which, furthermore, enables (6.5) to be written as
dT ( )
= mcp (T Tf ) + (Tf T0 )(1 ) , (6.5.a)
dx
in which form it will be used further on.
As for the second condition (6.9) it is also satised for x = +, since the
condition of chemical equilibrium for the products may be expressed as
w(1, f , Tf ) = 0. (6.11)
d
To the contrary, condition 0 for x , will not be satised unless
dx
w(0, 0 , T0 ) = 0, (6.12)
which, generally, will be in contradiction with the laws of Chemical Kinetics.
This circumstance implies a fundamental difculty, whose origin lies upon the
fact that it is impossible to maintain invariable the composition of the unburnt gases
located before the wave, as imposed by boundary conditions (6.7) and (6.9), when the
reaction velocity of such gases is not zero.
The existence of stationary combustion waves observed in practice may be

explained by the fact that such waves do not correspond exactly to the one-dimensional
model proposed in this work, since the mixing of reactant species forms, in general,
shortly before they reach the wave, which always looses a certain amount of heat
through the walls of the ame holder, for example, through the walls of the burner.
Furthermore, these walls act as chain breakers for the chemical reactions of the wave.
There is a possibility of eluding the above mentioned difculty by incorporat-

ing to the one-dimensional model the effect of all these complex circumstances, which
prevent the reaction of the unburnt gases in a real ame, through a slight modication
of the boundary conditions at the neighborhood of the cold limit of the ame, which
may be attained in several different ways. The practical interest of these solutions
is based on the fact that for reactions with an appreciable activation energy, as those
expected in combustion, the structure and propagation velocity of the ame are insen-
sible to a modication of the said boundary conditions, which makes it unnecessary to
dene them exactly, within a wide range of values of the dening parameters. More-
over, the proposed solutions are equivalent since they lead to the same fundamental
results. The following paragraph is devoted to a discussion of these solutions.
6.4. MODIFICATION OF THE CONDITIONS AT THE COLD BOUNDARY 137
6.4 Modication of the conditions at the cold

boundary
There are several solutions available in order to elude the difculty of the cold bound-
ary, which has passed unnoticed until recently. The history of the evolution of thought
on this subject may be followed by consulting references [17] through [19]. Herein
we will only study the two solutions currently used, which consist in introducing an
ignition temperature Ti and a ame holder.
Ignition temperature
By adopting the same assumption used in the classical theories of Combustion, that
is, by assuming that there is an ignition temperature Ti , such that reaction velocity of
the mixture is zero for T < Ti , this temperature divides the combustion wave into
two zones: a heating and diffusion zone, corresponding to temperatures under Ti , at
which no chemical reaction takes place, and a reaction zone, which corresponds to
temperatures over Ti .
The differential equations for the reaction zone are the same given in the pre-
ceding paragraph, which are summarized in the following, as well as the boundary
conditions, assuming that the origin of coordinates x = 0 is located at ignition point
T = Ti .
Reaction zone, x > 0.
d
m = w(Y, , T ), (6.2)
dx
dY
D = m(Y ), (6.3)
dx
dT ( )
= mcp (T Tf ) + (Tf T0 )(1 ) . (6.5.a)
dx
The boundary conditions for this system will be
x=0: = 0, T = Ti , (6.13)
x + : = 1, T = Tf , Y = 1. (6.14)
The equations for the heating and diffusion zone, x < 0, may be obtained
from those given 2 by introducing in them the conditions expressing the absence of
chemical reaction, that is: w = = 0. There resulting
Heating and diffusion zone, x < 0.
dY
D = mY, (6.15)
dx
dT
= mcp (T T0 ). (6.16)
dx
The corresponding boundary conditions are
x = : T T0 , Y 0, (6.17)
x=0: T = Ti . (6.18)
Since for x both dY / dx and dT / dx tend to zero, as it can be veried

by taking (6.17) into (6.15) and (6.16), the introduction of an ignition temperature
eliminates the difculty at the cold boundary. Furthermore by comparing the equa-
tions for the heating and diffusion zone with those for the reaction zone, given in the
preceding paragraph, and the boundary conditions (6.13) and (6.18), it may be readily
veried that the transition from one zone to another at point x = 0 is continuous for
all variables when adopting the additional condition that the value for Y at x = 0,
which is undetermined, be the same for both solutions
x=0: Y (0 ) = Y (0+ ). (6.19)
The objection to this way of dealing with the difculty at the cold boundary
states that the solution obtained, and in special the propagation velocity for the ame,
will depend on the value adopted for Ti , which does not actually exist. Nevertheless, it
will be proved further on, when studying the solutions for the combustion wave, that as
long as the velocity of the chemical reaction depends substantially on the temperature
of the mixture, it will happen that the solution obtained will be independent from the
values of Ti , except for values of this temperature very close to T0 or to Tf . Under
such conditions the inuence of the ignition temperature will vanish completely from
the result, when calculating the propagation velocity of the ame.
The presence of a ame holder
Another solution proposed to the problem of the cold boundary consists in imagining
the presence of a ame holder placed in front of the wave which has the double mission
6.4. MODIFICATION OF THE CONDITIONS AT THE COLD BOUNDARY 139
of absorbing a certain amount of heat Q from the ame and of acting as a lter which
allows the unburnt gases to reach the ame but prevents the combustion products from
diffusing towards the unburnt gases. By assuming that the holder is located at point
x = 0, the system of Eqs. (6.2), (6.3) and (6.5a) will still hold, but conditions (6.13)
at the cold boundary will have to be substituted by the following
dT
x=0: T = T0 , = 0, 6= 0. (6.20)
dx
To the contrary, for x < 0 the composition and state of the mixture are uniform
x<0: Y = = 0, T = T0 . (6.21)
The value of Y at x = 0 remains undetermined, but different from zero. There-

fore through x = 0 there exists a discontinuity in the composition of the mixture which
becomes possible due to the existence of the lter.
It happens here, as in the case of ignition temperature, that the structure and
propagation velocity of the ame are independent from the value of parameter Q,
provided it is not close to heat of reaction q or very close to zero, which justies the
practical value of the proposed model.
Tf Tf
T0 Ti 1 T0 1
1Y 1Y
boundary condition
Karman Hirschfelder boundary condition
Figure 6.1: Boundary conditions for ignition temperature and ame holder models.
Figure 6.1 summarizes and compares the conditions at the cold boundary for
the two solutions proposed. Since both lead to the same results, in the following we
will use only the assumption that an ignition temperature exists.
Figure 6.2 shows in a qualitatively way the form of the solutions obtained for
the propagation velocity of the ame in a given mixture when the ignition temperature,
assumed for it, is changed. It is seen that there is a wide interval AB of ignition
temperatures within which the propagation velocity of the ame is independent from
Ti . The quantitative solutions for some typical cases will be included further on.
0.8
0.6
b
S
0.4
A B
0.2
0
0 0.2 0.4 0.6 0.8 1
T0/Tf T /T
i f
Figure 6.2: Schematic diagram showing the ame propagation velocity vs ignition temper-
ature.
6.5 Propagation velocity of the ame
A simple check of the boundary conditions at the reaction zone shows that they
are superabundant. In fact, since we are dealing with a system of three equations of
rst-order, the solution of the system is determined for each value of m by boundary
conditions (6.14), except for a translation, which is determined by the additional con-
dition that for x = 0 be T = Ti as imposed by (6.13). In order to satisfy as well the
additional condition x = 0 : = 0, it is necessary that m takes a particular value: the
eigenvalue of the system which makes compatible boundary conditions (6.13) and
(6.14).
This eigenvalue will determine the propagation velocity of the ame, which by
virtue of (6.1) is given by
m
u0 = . (6.22)
0
6.6 Example
Even in the case of only two chemical species as considered in the preceding para-
graphs, the non-linear character of the ame equations and in particular the shape of
the reaction velocity add difculties to the problem to such an extent that it becomes
6.6. EXAMPLE 141
generally impossible to obtain explicit solutions for the equations. Hence, it is neces-
sary to resort to cumbersome numerical integrations or to analytical methods which
will be discussed further on. However, in order to illustrate the nature of the solutions
and the way in which the eigenvalue appears determined, the present paragraph offers
an example in which through an adequate simplication of both transport coefcients
and chemical reaction velocity, it is possible to obtain an explicit solution of the prob-
lem giving correct qualitative predictions. A possible objection to this solution could
be that the results obtained depend on the ignition temperature of the mixture, which,
as before said, does not exist. But this is due to the fact that the simplication assumed
for the reaction velocity is excessive and consequently the inconvenient will vanish for
more realistic cases, as will be proven in the following.
For this example the following assumptions will be adopted:
1) The coefcient of thermal conductivity and the product D are constant.

2) The reaction is of rst-order and its velocity is independent from temperature, in
which case it has the form1
w = 0 k(1 Y ). (6.23)
It is precisely this simplied form of the reaction velocity and, in special, the fact
that in it the activation energy is assumed to be zero, an indispensable condition
for the integrability of the system, which makes the solution depends on the
ignition temperature, as aforesaid.
When expression (6.23) is taken into (6.2), it gives
d
m = 0 k(1 Y ). (6.24)
dx
The diffusion equation subsists in the form (6.3)
dY
D = m(Y ). (6.25)
dx
Finally, when the dimensionless temperature = T /Tf is introduced, the en-

ergy equation (6.5.a) may be written
d
= 1 + (1 0 )(1 ), (6.26)
mcp dx
where is 0 = T0 /Tf .
1 See 8 of Chap. 1.
In order to nd the solution, it is convenient to eliminate x from the preceding

system. This is done by dividing (6.24) and (6.25) by (6.26), thus resulting
dY Y
=L , (6.27)
d 1 + (1 0 )(1 )
d 1Y
= . (6.28)
d 1 + (1 0 )(1 )
In this system

L= (6.29)
Dcp
is the Lewis-Semenov number of the mixture, which is constant.
is a dimensionless parameter, also constant, dened by
0 k
= . (6.30)
m2 cp
The problem lies, precisely, in determining the eigenvalue of this parameter. Once it is
known, the velocity u0 of the ame propagation may be derived from it and Eq. (6.22),
thus obtaining
k 1/2
u0 = . (6.31)
0 cp
The system of equations (6.27) and (6.26) holds only within the reaction zone, where
boundary conditions are
= i : = 0, (6.32)
=1: = = 1, (6.33)
where we have written i = Ti /Tf .
Precisely because these three conditions exist for a system of second-order, it

is necessary to look for the value of which makes them compatible.
The system is readily integrated by testing lineal solutions of the form
1 = (1 ), (6.34)
1 Y = (1 ), (6.35)
which satisfy boundary conditions (6.33) at the hot boundary.
Condition (6.32) at the cold boundary gives the following relation
1 = (1 i ). (6.36)
6.6. EXAMPLE 143
Two more relations may be obtained by taking (6.34) and (6.35) into (6.27) and
(6.28), there resulting

= , (6.37)
1 + (1 0 )

= . (6.38)
1 + (1 0 )
The three Eqs. (6.36), (6.37) and (6.38) become a determined system for the
computation of , and . The solution is
1
= , (6.39)
1 i
( )1
1 1 i 0
= 1+ , (6.40)
1 i L 1 i
( )
i 0 1 i 0
= 1+ . (6.41)
1 i L 1 i
Once is known, then Eq. (6.31) provides the following dimensionless ex-
pression for the propagation velocity of the ame

0 cp
u0 = 1/2 . (6.42)
k
Finally, since we know , Eq. (6.35) gives the value Yi for Y corresponding to
= i ,
Yi = 1 (1 i ). (6.43)
The solution for the heating and diffusion zone, x < 0, will be obtained by
integrating the following equation
dY Y
=L , (6.44)
d 0
which results from (6.27) when making = 0. The solution to this equation is
( )L
0
Y = Yi . (6.45)
i 0
Once the preceding solutions are known, one may pass to the physical space
by making use of solutions (6.34) and (6.35) and Eqs. (6.24), (6.25) and (6.26), as
follows.2
2 This far, the problem could have been solved under less drastic assumptions, i.e., not being constant
nor D, but the Lewis-Semenov number.

a) Reaction zone.
After substituting (6.34) into (6.26), we have

d i 0
= (1 ), (6.46)
d 1 0
where the following dimensionless distance was introduced
x
= , (6.47)
l
being

l= (6.48)
mcp
a characteristic length of the wave. The integration of (6.46), once it is assumed that
reaction starts at point x = 0, gives
i 0

1 = (1 i ) e 1 i . (6.49)
Likewise, one obtains for and Y

i 0

1 = e 1 i , (6.50)
( )1 i 0
1 i 0
1Y = 1+ e 1 i . (6.51)
L 1 i
1.0
0.8
Y(L=0.5)
0.6
, Y
0.4
Y(L=1)
0.2
Y(L=2)
0.2 i=0.4 0.6 0.8 1.0

=0.125
0
Figure 6.3: Distributions of Y and as a function of when the activation energy is zero.
6.7. REACTION VELOCITY 145
1.0
0.8
Y(L=2)
0.6
, , Y
Y(L=1)
0.4
Y(L=0.5)

0.2

0
0.0
10 8 6 4 2 0 2 4 6 8 10

Figure 6.4: Distributions of Y , and as functions of , when the activation energy is zero.
b) Heating zone.
Likewise, in this zone one obtains
= 0 + (i 0 ) e , (6.52)
( )1
1 i 0 1 i 0
Y = 1+ eL , (6.53)
L 1 i L 1 i
= 0. (6.54)
Figures 6.3 and 6.4 represent the solutions computed for the following typical
values: 0 = 0.125, i = 0.4 and L = 0.5, 1 and 2.
6.7 Reaction velocity
The correct expression for the reaction velocity w is, in accordance with the laws of
Chemical Kinetics,3
w = n k(1 Y )n . (6.55)
3 See Chap. 1, 8.
In this expression, n is the order of reaction and k a function of temperature, which,

generally, is of the form
( )
T
k=A eE/RT , (6.56)
Tf
where A is a constant and E is the activation energy of the reaction.
It is convenient to eliminate the density from Eq. (6.55) so that only two vari-
ables appear explicitly: the temperature and the mass fraction of products.
Let Mr and Mp be the molar masses of reactants and products, respectively,

and a the relation
Mr
a= 1. (6.57)
Mp
Then, the mean molar mass of the mixture is4
Mr
M= , (6.58)
1 + aY
and the density of the mixture may be expressed as a function of pressure and temper-
ature in the form
pMr 1
= . (6.59)
RT 1 + aY
By taking now Eq. (6.56) and (6.59) into (6.55), we nally obtain for the reac-
tion velocity
( )n ( )n ( )
pMr 1Y T
w=A eE/RT . (6.60)
RT 1 + aY Tf
It is advantageous to bring forth in Eq. (6.60) the dimensionless temperature =

T /Tf as dened in the preceding paragraph, and to introduce dimensionless activation
temperature a as dened by
E
a = . (6.61)
RTf
Now, Eq. (6.60) may be written
( )n
1Y
w = B n
ea / , (6.62)
1 + aY
where ( )n
pMr
B=A (6.63)
RTf
is a constant of the process. Eq. (6.62) is the form of the reaction velocity that will be
used in the following.
4 See Chap. 1, Eq. (1.36).
6.8. FLAME EQUATIONS 147
6.8 Flame equations
When (6.62) is taken into (6.2) , the later takes the form
( )n
d 1Y
m = B n
ea / . (6.64)
dx 1 + aY
Equation (6.3) is still written

dY
D = m(Y ). (6.65)
dx
If we substitute into Eq. (6.5.a) the dimensionless temperature, we have for it
d ( )
= mcp 1 + (1 0 )(1 ) . (6.66)
dx
In order to solve the above system it is convenient to divide (6.64) and (6.65)
by (6.66), as was done with the example treated in 6, thus obtaining
( )n
n 1 Y
1
d f 1 + aY a
= (1 0 ) e , (6.67)
d 1 + (1 )(1 0 )
dY Y
=L . (6.68)
d 1 + (1 0 )(1 )
In this system,
Bf ea
= (6.69)
m2 cp (1 0 )
is an undetermined parameter of the problem and L is the previously dened Lewis-
Semenov number of the mixture.5
The boundary conditions of system (6.67), (6.68) are the same given in (6.32)
and (6.33), namely,
Hot boundary: = 1, Y = 1, (6.70)

Cold boundary: = i , = 0. (6.71)
The problem lies in determining the value of which makes compatible these
three conditions in the system formed by the two rst-order equations (6.67) and (6.68).
1 a
5 In the numerator of (6.67) we bring forth factor ea , in lieu of e in order to simplify
calculations since a is generally much larger than unity and otherwise, we would have to deal with very
small numbers instead of doing it with numbers of the order of unity.
Once is known then the velocity u0 of the ame is given by

Bf ea
u0 = 1/2 . (6.72)
0 cp (1 0 )
2
6.9 Solution of the ame equations
As before said, the preceding system cannot by integrated analytically. Hence, it

becomes necessary to resort either to numerical solutions or else to semi-analytical
approximation methods.
The numerical methods are very cumbersome since we are dealing with an
eigenvalue problem with given boundary conditions at both extremes of the interval,
and, therefore, we would have to make numerous tentatives before reaching the exact
solution. Generally, the use of electronic computers is required and in particular they
were widely utilized by Hirschfelder and his collaborators.
In recent years, several approximate methods have been proposed generally

based in the unusual behavior of the solutions due to the presence of factor ea / in
the reaction velocity. In fact, when the reduced activation temperature a is large, as
should be expected in combustion reactions, due to this factor the result is determined,
essentially, by the form of the solution close to the hot boundary and this enables
the development of methods, as those proposed by Zeldovich-Frank-Kamenetskii-
Semenov [4] and the one by Boys-Corner [5], Adams [20], Wilde [21] and von Karman
[6], with different variations and approximations. Only very recently a comparative
study of these methods has been performed. Information on the subject will be found
in the work by von Karman [10] and, specially, in the more complete study carried
out by Millan, Sendagorta and Da Riva [22]. Figures 6.5 and 6.6, taken from this
study show a comparison between the approximate values for 1/2 ( to which the
ame velocity is proportional by virtue of (6.72)), obtained through several of those
methods, with the exact value, given by a numerical integration. This comparison was
performed for several values of the temperature of the unburnt gases and of the acti-
vation energy of the reaction. All the results shown in these gures correspond to a
Lewis-Semenov number equal to unity and to a constant mean molar mass of the mix-
ture. However, the unpublished results of the calculations performed for more general
cases indicate that the same conclusions are still valid.
Figures 6.5 and 6.6 show that the Boys-Corner method in its rst iteration
gives defect values with an important deviation. The same happens with the method
by Zeldovich et al., but giving excess values. The approximation improves with the
6.9. SOLUTION OF THE FLAME EQUATIONS 149
second iteration of the Boys-Corner method but it is very arduous to perform and,
furthermore, like the rst one it does not converge towards the correct solution for
increasing activation temperatures.
1.5
1.4
1.3
ZELDOVICH
1.2
/ exact
1/2
WILDE
1.1 SENDAGORTA
BOYSCORNER 2nd IT.
1/2
1.0

0.9 KARMAN 2
KARMAN 1
KARMAN 3
0.8
BOYSCORNER
0.7
0.6
2 4 6 8 10 12 14 16
a
Figure 6.5: Comparison between the ame propagation problem eigenvalue obtained by
different approximated methods and the exact value for 0 = 0.125.
1.3
ZELDOVICH
1.2
BOYSCORNER 2nd IT.
1.1
WILDE SENDAGORTA
exact
1/2 / 1/2
1.0
KARMAN 2
0.9
KARMAN 3 KARMAN 1
0.8
BOYSCORNER
0.7
0.6
2 4 6 8 10 12 14 16
a
Figure 6.6: Comparison between the ame propagation problem eigenvalue obtained by
different approximated methods and the exact value for 0 = 0.250.
The other methods are considerably better mainly when applied to normal acti-
vation temperatures. This is specially true about Wildes method, the second and third
alternatives of von Karmans one and that proposed by Sendagorta [23], which is the
best and does not require more work than the others.
1.0
BOYSCORNER
0.8
0.6
EXACT

0.4
0.2
KARMAN 3
0.0
0.4 0.2 0.0 0.2 0.4 0.6 0.8 1.0

Figure 6.7: Comparison of the distributions of vs obtained by two approximate methods
with the exact result.
Figure 6.7 shows, for one of the case calculated, the law of the variation of
with of the Boys-Corner approximation, Karmans third alternative and the exact
solutions. It is evident that the approximation reached with Karmans method is very
satisfactory. The following studies the nature of this method after referring the reader
to the above references for a more detailed analysis of the other methods mentioned.
If in Eq. (6.67) we bring the denominator of the right hand side into the left one,
and this equation is integrated between the cold boundary (6.71) and the hot boundary
(6.70), we obtain
( )n 1
1 0 1 1
n 1Y a
(1 ) d = (1 0 ) e d.
2 0 i f 1 + aY
(6.73)
Consequently, in order to determine , the problem reduces now to obtaining

an approximation of vs on the integral of the left side of Eq. (6.73) and an approxi-
mation of Y vs on the integral of the right side of the same equation. These integrals
will be written for shortness as

( )n 1
1
n 1Y a
I =(1 0 ) e d, (6.74)
i f 1 + aY
1
J= (1 ) d. (6.75)
0
With this notation, the fundamental Eq. (6.73) may be written

1 0
J = I. (6.73.a)
2
The following is a separate study of both approximations.
Approximation of integral I
Heat transfer is a function of temperature of the mixture and its composition

dened by the value of Y .6 Consequently, in order to calculate I the problem reduces
to nding an approximation for Y as a function of . The solution depends on the
value of the Lewis-Semenov number.
Lewis-Semenov number equal to unity
If L = 1, then Y is a lineal function of in the following form

0
Y = . (6.76)
1 0
In fact, when condition

L =1 (6.77)
Dcp
is satised, the diffusion Eq. (6.65) may be written in the form
dY
= mcp (Y ). (6.78)
dx
By adding to it the energy Eq. (6.66), multiplied by an arbitrary constant C,
d ( ( ) )
(Y + C) = mcp Y + C( 0 ) C(1 0 ) + 1 . (6.79)
dx
This equation is satised identically with the following two conditions
C(1 0 ) + 1 = 0, (6.80)
Y + C( 0 ) = 0, (6.81)
6 See Chap. 2, 4.
the rst is satised by adopting the following value for C

1
C= . (6.82)
1 0
This value when taken into Eq. (6.41) gives for Y the expression (6.76) which is, in
fact, the desired solution valid throughout the interval of temperatures as the boundary
conditions are met at both extremes.
= 0 : Y = 0, =1: Y = 1. (6.83)
0.40
a=4
0=0.125
0.35
1/2
0.30
0.25
0.20
0.5 1.0 1.5 2.0
L
Figure 6.8: Dimensionless ame velocity as a function of the Lewis number for a = 4 and
0 = 0.125.
Lewis-Semenov number different from unity
Actually the Lewis-Semenov number is only identical to unity when dealing with pure
mono-atomic gases (in which case D is the coefcient of autodiffusion). Generally
in more complex cases L differs from unity and it varies with the composition and
temperature of the mixture. However, it is frequently very close to unity and the
preceding approximation is satisfactory. Moreover, the inuence of the value of the
Lewis-Semenov number on the ame velocity is, generally speaking, quite small. This
is veried in Fig. 6.8 taken from Ref. [22] which shows, as an example, that when
L doubles its value, the ame velocity increases less than 20 per cent, in this case
corresponding to a rst-order reaction with an activation temperature a = 4, which
is small, and therefore the inuence of L is more obvious than it would be for larger
values of a .
1
a
On the other hand, due to the presence of factor e in the integral of
the right hand side of Eq. (6.73) it occurs that for large values of a , as being those
normally appearing in combustion, the value of the integral is inuenced only by the
value taken at the neighborhood of = 1 by the quantity under the integral.
1.0
a=4
L=0.5
0.8
0.6

0.4
0.2
0=0.125
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
, Y
Figure 6.9: Solutions of the ame equations 6.67 and 6.68 for L = 0.5, a = 4, 0 =
0.125, i = 0.4, n = 1, a = 0 and = 1.
In order to obtain the desired approximation Y vs it is sufcient to observe

Fig. 6.3 corresponding to the example treated in 7, since this gure shows that even
when the activation energy a is zero it happens that near = 1 is 1 1 Y ,
and also 1 1 . This fact appears more evident for reactions with a high
activation temperature as shown in Fig. 6.9 which was calculated for a more realistic
case. Eq. (6.68) may be written
dY L (1 ) (1 Y )
= , (6.84)
d 1 0 1
(1 )
1 0
1
and, in accordance with the preceding considerations, both 1 Y and are neg-
1 0
ligible close to = 1 and the following approximation may be obtained for Eq. (6.84)
dY L
'1: ' , (6.85)
d 1 0
whose solution is
L
'1: 1Y ' (1 ), (6.86)
1 0
this approximation must replace Eq. (6.76) when L is different from unity and fur-
thermore the value taken for L must be that corresponding to the temperature and
composition of the hot boundary conditions of the ame, T = Tf .
If the activation energy has a small value and an approximation of Y vs that

will be valid within a wider range of temperatures is desired, then the linear approxi-
mation (6.86) may be substituted by the following parabolic
( )L
0
Y ' , (6.87)
1 0
which for = 1 coincides with (6.86) and for ' 0 it behaves like the solution
corresponding to the heating zone ( 0). This can readily be veried. Fig. (6.10)
shows some of the curves (6.86) and (6.87).
1.0
L=0.5
L=0.75
0.8
L=1.5
0.6
L=2.0
Y
L=1.0
0.4
0.2
0.0
0.125 0.2 0.4 0.6 0.8 1.0

Figure 6.10: Linear and parabolic approximations of Y vs given by Eqs. 6.86 and 6.87
for 0 = 0.125.
The preceding considerations prove that it is generally sufcient, when cal-

culating the integral to take for /f the value unity which corresponds to the hot
boundary, or some other approximation at the neighborhood of this point. Frequently
the following approximations are used

' , ' . (6.88)
f f
1.4
1.3
= / =
=0.250
1/2
1.2
1/2
=0.125
0
1.1
1.0
2 4 6 8 10 12 14 16
a
Figure 6.11: Ratio of ame velocities corresponding to = f and = f as a function
of the activation temperature a for two different values of the initial tempera-
ture 0 .
Figure 6.11, taken from Ref. [22], sets forth the scarce inuence of the vari-
ation of with temperature on the ame velocity, specially for high activation tem-
perature. Two different solution are compared in this gure, one assuming that is
constant through the ame = f ; and another where varies linearly with tempera-
ture = f , in accordance with (6.88). The solution was computed for two different
values of the temperature of unburnt gases The deviations are under 10 per cent.
Once established that the preceding considerations solve the problem of the
calculation of integral I we shall proceed with the study of J.
Approximation of integral J
In order to approximate integral J, we must rst consider that in the normal cases, this
is to say when a 1, this integral is much smaller than the other term of the left
hand side of Eq. (6.73), that is
1 0
J . (6.89)
2
Such a property is illustrated in Fig. 6.12 which corresponds to a typical case

where the value of J is shown by a dot-area while the value for (1 0 )/2 is repre-
sented by the area of triangle ABC.
1.0 A
1 (1) d

0
0.8
0.6

0.4
(10)/2
0.2
=0.125 B
0
C
0.0
0.0 0.2 0.4 0.6 0.8 1.0

R1
Figure 6.12: Areas representing the values of J = 0
(1 )d and (1 0 )/2 in a typical
case.
Hence, it results that a rst approximation is obtained when J is neglected

respect to (1 0 )/2, in which case the value of is given by expression
1 0
= . (6.90)
2I
Actually, this approximation was introduced by Zeldovich et al., although by

means of a more complicated justication. Furthermore, the authors applied an in-
adequate approximation for the value of integral I which is the main reason for the
error resulting from their method as it was shown in Figs. 6.5 and 6.6. Professor
von Karman, after a more accurate calculation of integral I, as before said, has im-
proved the approximation obtained by Zeldovich et al. This can be seen in Figs. 6.5
and 6.6 where the curves named Karman correspond to (6.90) when using for I the
approximation developed in the preceding paragraph.
Von Karman improved the approximation for J, through an approximation of

1 vs 1 , which is obtained by studying the behavior of the differential equation
(6.67) at the neighborhod of the hot boundary in the following way.
Close to point = 1, difference 1 is an innitesimal, whose order depends

only on the order of 1 and it can be easily derived from (6.67) by introducing into
it the following approximations:
1) We neglect 1 respect to (1 0 ) (1 ), in agreement with the procedure

used in obtaining (6.85).
2) We substitute (1 Y ) by its approximation as a function of (1 ), given by
Eq. (6.86).
3) The remaining terms of (6.67) are substituted by the corresponding values at the
hot boundary.
Once these approximation are introduced into (6.67) one obtains the follow-
ing approximation for this equations, which is valid at the neighborhood of the hot
boundary
( )n
d L (1 )n
1 (1 )(1 0 ) : ' . (6.91)
d 1 0 (1 0 )(1 + a) 1
After integration the following approximation 1 vs 1 , is obtained
( ) 1 ( ) n 2
n + 1 1 0 n+1 (1 0 )(1 + a) n + 1
1 ' (1 ) n + 1, (6.92)
2 L
which when taken into (6.75) and integrated, gives for J
( ) 1 ( ) n
n+1 n + 1 1 0 n+1 (1 0 )(1 + a) n + 1
J= . (6.93)
n+3 2 L
By taking this value of J and the one for I given by Eq. (6.74) into Eq. (6.73), the
following equation results for the calculation of
( ) 1 ( ) n 1
1 0 n+1 n + 1 n + 1 1 + a n + 1 n + 1
(1 0 ) = I, (6.94)
2 n+3 2 L
which may be readily solved, either with a graphical method or by iteration, using for
this last case as a rst approximation the one given by (6.90) and as the correcting
term for successive iterations the right hand side of Eq. (6.94).
The result reached with (6.94) are represented in Figs. 6.5 and 6.6 where the
corresponding curves are named Karman 2.
The approximation for J may still be improved by writing it in the form

1
d
J= (1 ) d, (6.95)
d
d
and using for an approximation of (6.67) derived as follows
d
( )n 1
d L n a
(1 ) ' e (1 )n , (6.96)
d 1 0 (1 0 )(1 + a) f
which through integration, gives

v
( )n u
u 1
2 L 2 t
1
(1 )n a
1' e d. (6.97)
1 0 (1 0 )(1 + a) f n
This expression, when taken into the denominator of (6.67) after we have ne-
glected 1 , supplies an approximation for d/ d which introduced in (6.95) gives
the desired value for J. The result obtained from this approximation corresponds to
the curves named Karman 3 of Fig. 6.5 and 6.6.
Tho approximation for J is even better when, after Sendagorta, we calculate

d/ d in (6.95) by using for vs an approximation of the form
1
a
=Pe , (6.98)
where P is a polynomial of the form
P = 1 + a1 (1 ) + a2 (1 )2 + + am (1 )m , (6.99)
whose coefcients must be determined by studying the behavior close to = 1 in

the differential equation (6.67).
For example, in the case of a rst-order reaction (n = 1), it gives for J
J = H0 + a1 H1 + a2 H2 + + am Hm , (6.100)
where
1 1
a
Hj = (1 )j e d, (j = 1, 2, . . . m). (6.101)
i
This integral may be expressed by a law of recurrences starting from H0 , which

in turn is expressed through the exponential integral in the following way
1 1
a
H0 = e d
i
(6.102)
1 i ( )
a a
= 1 a ea E1 (a ) i e i + a ea E1 ,
i
being
ez
E1 () = dz. (6.103)
z
For example, if in the preceding case one limits approximation (6.99) to the rst two
terms,
P = 1 + a1 (1 ), (6.104)
6.10. STRUCTURE OF THE COMBUSTION WAVE 159
in which case the only undetermined parameter would be a1 to be obtained when

comparing the value for ( d/ d)=1 given by (6.98),
( )
d
= a a1 , (6.105)
d =1
with the one resulting from (6.67)
( )
d (1 0 )
= ( ) . (6.106)
d =1 d
(1 0 ) 1
d =1
The following second-degree equation will be obtained for a1
1 0
a a1 = , (6.107)
(1 0 )(a a1 ) 1
which when solved gives the value of the undetermined parameter, which in turn,
when taken into (6.100) provides the desired approximation for J.
Therefore, this approximation of J depends on , and when introduced into

(6.73.a), it supplies an equation for , whose solution gives the eigenvalue which
solves the problem.
6.10 Structure of the combustion wave
Once determined the value of which makes compatible the boundary conditions of
the ame equations, a simple numerical integration of such equations, through any of
the methods available, will give the distribution of temperature, concentrations, etc.
through the combustion wave. If the integration is performed by using system (6.67),
(6.68), the curves obtained will be as those shown in Fig. 6.9. When the structure is
desired on a physical plane it is sufcient to integrate the system of equations (6.64),
(6.65) and (6.66). However for this case one must select the position of origin of coor-
dinates which is undetermined. Fig. 6.13 shows the solution corresponding to the case
represented in Fig. 6.9. It is seen that the structure is analogous to the one represented
in Fig. 6.4, corresponding to the simplied case discussed in 6. Fig. 6.13, shows a
dot-line representation of the distribution of temperatures corresponding to the case of
pure thermal conduction, where chemical reaction is absent, which practically coin-
cides with the wave up to a = 0.5, proving that up this point the action of chemical
reaction is extremely small.
As a characteristic length of the phenomenon it was adopted, for its solution,

the same
f
l= (6.108)
mcp
applied before in 6. This is the characteristic length of the process and it is a measure
of the thickness of the ame, since although theoretically it is innite, almost the
entire variation of the temperature and composition of the gases takes place within a
narrow zone which is a few times the value of l, as shown in the solution computed in
Fig. 6.13.
By calculating the value of l corresponding to a typical case it may be veried

that it is a small fraction of a millimetre and, consequently, the thickness of the ame,
under normal conditions, has the same order of magnitude, as proven by experimental
observation.
This circumstance increases considerably the difculties of the experimental

observation of the structure of the ame, since the recording of the distribution of
temperatures, for instance, must be performed within a very narrow zone and when
attempting to measure the composition the difculties encountered are even greater.
Several attempts have been made of measurements of this nature, mainly by the re-
search team of the Applied-Physics Laboratory of the Johns Hopkins University. A
description of the present state of this problem may be found in Ref. [24]. The mea-
surements performed agree completely with theoretical predictions.
1.0
0.9
0.8
0.7
Y
0.6
Y, ,
0.5
L=0.5
0.4
=4
a
0.3
0=0.125
0.2
=0.4
i
0.1
0
2.0 1.5 1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5
=mcpx/
Figure 6.13: Structure of the combustion wave represented in Fig. 6.9. The origin of coordi-
nates corresponds to the ignition temperature, where chemical reaction starts.
6.11. IGNITION TEMPERATURE 161
6.11 Ignition temperature
We have established that, except when activation temperature a is zero or very small,
the ignition temperature i bears no inuence on the ame velocity unless it has a
value close to the temperature of the burnt gases = 1, or very close to initial tem-
perature 0 .
1.6
a=0
1.2
1/2
0.8
a=2
0.4 a=4
a=8
0.0
0.0 0.2 0.4 0.6 0.8 1.0
=0.125
0 i
Figure 6.14: Flame velocity as a function of the ignition temperature i for different values
of the activation temperature a .
This property is illustrated by Fig. 6.14, where the law of variation of 1/2
(to which u0 is proportional as before said), as a function of i is shown for a typical
case, corresponding to a rst-order reaction, with a Lewis-Semenov number equal to
unity and a temperature of the unburnt gases 0 = 0.125. The gure represents the
solutions corresponding to values of a comprised between 0 and 8 and it is seen that
except for the rst case (a = 0), there exists a wide range on values of i in which
1/2 is practically constant, as it was previously announced.7
This property is of a general nature and it justies the assumption of a ignition

temperature which eliminate the difculty of the cold boundary.
7 For 1 and very close to , it can be shown by means of asymptotic techniques that 1/2 =
a i 0

` 1 ` 1 0 1/2
(1 0 ) ln exp a , so that when i 0 there exists an exponentially
a (i 0 ) 0
thin layer in which the transition for 1/2 = constant to 1/2 = occurs, Ed.
0.05
0.04
a=8
(1)ea(1)/
0.03
0.02
0.01
l
0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.15: Curve showing the typical values of the integrand of I as a function of .
The reason for the insensibility of the solutions to i appears clearly in Fig. 6.15
where it is seen that the values of i comprised between 0 and l do not inuence in
the value of integral I, since in this zone the integrand is practically zero. To the
contrary, if i is comprised between l and 1, then one disregards in the solution the
contribution of the part corresponding to l < < i . Since 1/2 is proportional to
I, this explains the fact that 1/2 should decrease and tend to zero when is i 1.
The fact that the combustion velocity tends to when i 0 may be phys-
ically explained since then all the mass of gas burns simultaneously giving innite
wave velocity. In order to give a theoretical explanation it would be necessary to
analyze in detail the differential system.
The fact that 1/2 is independent from i enables the elimination of this value
in all the equations of the preceding paragraphs, in special in the lower limit of integral
I (Eq. 6.74), by substituting it for zero without changing the result.
6.12 General equations for the combustion wave in the

case of more than two chemical species
Hirschfelder and his collaborators were the rst to give the general equations for the
combustion wave in the case of more than two chemical species [7]. A derivation of
these equations may be found in a paper by Karman and Penner [6].
6.12. GENERAL EQUATIONS FOR THE COMBUSTION WAVE 163
Such equations are a particular case of those corresponding to the one-dimensional

stationary ow of a mixture of gases which were deduced in 8, chapter 3 of this book
and will be used as a starting point for the present study, although certain modi-
cations and simplications will be required in order to adapt these equations to the
problem in which we are now interested.
Reaction equations
The reaction equation which corresponds to species i is the Eq. (3.63) of the above
mentioned chapter
di
m = wi , (i = 1, 2, . . . , l). (6.109)
dx
where l is the number of different chemical species. In this expression wi is the
reaction velocity of species i, which, in the case of more than one chemical reaction,
is the resultant of the contributions of each one of them
r
wi = wij . (6.110)
j=1
Here, wij is the mass of species i produced per volume unit and time unit due
to reaction j, and r is the number of different reactions, counting as such, the two
opposite ones which take place in each reaction.
Velocity wij corresponding to reaction j is given by expression (1.119) of

chapter 1 which will be written herein in the following way by bringing forth in it
the molar fractions X of the species in lieu of the mass fractions which were applied
in the case of two chemical species (this change is done for the reasons stated further
on the diffusion equations)
( p )nj l
0
00 0
wij = Mi (ij ij )kj Xs sj . (6.111)
RT s=1
0 00
In this expression Mi is the molar mass of species i, ij and ij are the stoichiometric
coefcients of this species in the left and right sides of reaction equation j, nj =

l
0
sj is the order of the reaction and kj is a function of the temperature of the mixture
s=1
which, as in the case of two species, is generally of the form
kj = Aj T j eEj /RT . (6.112)
When (6.112) is taken into (6.111), the equation may be written for shortness

l
0
wij = Bj T j nj eEj /RT Xs sj , (6.111.a)
s=1
which is the form that will be used is the following. In the equation it has been taken
( p )nj
Bj = Aj . (6.113)
R
As aforesaid said in chapter 3, 8, even when there exist l reaction equations (6.109)
corresponding to i chemical species they are not all independent from one another.
In fact, the number of independent equations is the smallest of the following two: 1)
number of chemical reactions; 2) number of independent components of the mixture,
in the sense of the rule of phases.
If we take (6.111.a) into (6.109), the latter may be written in the form
di r l
0
m = 00
Mi Bj (ij 0
ij )T j nj eEj /RT Xs sj , (6.109.a)
dx j=1 s=1
which is the form that will be used in the following.
Diffusion equations
Since it is evident that

v = m, (6.114)
equation (3.68) of Chap. 3 may be written
Yj vdj = m(j Yj ), (j = 1, 2, . . . , l). (6.115)
If we now take this expression of the diffusion velocities into the equations of system
(3.75) (which in this case still holds if we nullify the term corresponding to pressure
diffusion since it is neglectable), we shall obtain the desired system of diffusion equa-
tions.
However, when written in this form, the said system has the inconvenient that
the derivatives of the mass fractions dYj / dx do not appear explicit as it would be
necessary in order to obtain the system of equations for the ame in the canonical
form. Such an inconvenience can be easily avoided by simply expressing the result
as a function of molar fractions Xi in lieu of the mass fractions. In fact, in this case
the system of diffusion equations to be used is the one given by Eq. (2.28), which
after applied to the one-dimensional ow and neglecting the pressure and temperature
diffusions takes the form
dXi Xi Xj
l
= (vdj vdi ). (6.116)
dx j=1
Dij
On the other hand, considering that8
Mj
Yj = Xj (6.117)
Mm
and
pMm
= , (6.118)
RT
the elimination of vdi , vdj , Yj and between these three last equations and (6.115),
nally gives the following system of diffusion equations in the form which will be
applied to the solution of the ame equations
( )
mRT 1
l
dXi j i
= Xi Xj , (i = 1, 2, . . . , l). (6.119)
dx p j=1 Dij Mj Mi
Energy equation
The energy equation is the one give in Eq. (3.74), when we disregard in it the terms
due to kinetic energy and viscosity, thus reducing it to the following

l
dT
m j hj = e, (6.120)
j=1
dx
where e is a constant dened by the initial or nal conditions.
Before transforming this equations into the form in which it will be applied we
shall rst consider the boundary conditions.
Boundary conditions
These conditions are analogous to those applied in the case of two chemical species,
namely,
Cold boundary, x :
dT di dXi
T T0 , i i0 , Xi Xi0 , 0, 0, 0. (6.121)
dx dx dx
Hot boundary, x +:
dT di dXi
T Tf , i if , Xi Xif , 0, 0, 0. (6.122)
dx dx dx
8 See Chap. 1, Eq.1.34
In (6.121), T0 , i0 = Yi0 and Xi0 = Yi0 Mm0 /Mi are the values of T , i and
Xi corresponding to the temperature and composition of the unburnt gases. Mm0 is
the value of Mm under these conditions.
Likewise, in (6.122), Tf , if = Yif and Xif = Yif Mmf /Mi are the tempera-
ture and composition corresponding to the adiabatic combustion products in chemical
equilibrium.
A recount of conditions, similar to the one performed in 6 for the case of two
chemical species, shows that conditions are superabundant and that their compatibility
imposes that m takes an eigenvalue which determines the propagation velocity of
the ame.
Boundary conditions (6.121) and (6.122) determine the value of constant e in

(6.120), when it is particularized for both extremes, thus obtaining

l
l
e=m j0 hj0 = m jf hjf . (6.123)
j=1 j=1
In this expression the equality of the last two terms expresses simply that the combus-
tion is adiabatic and at constant pressure.
Substituting (6.123) into (6.120) the latter may be written as
dT l
=m (j hj jf hjf ). (6.124)
dx j=1
In Chap. 1, 3, it was established that the enthalpy of a diluted gas is of the

form T
hj = h0j + cpj dT, (6.125)
T0
where h0j is the formation enthalpy9 of species j at temperature T0 and cpj is the
specic heat of the same at constant pressure.
When taking (6.125) into (6.124) the latter is written

T
dT
= m h0j j + cpj j dT
dx j T0 j
(6.126)
Tf
m h0j jf + cpj jf dT .
j T0 j
9 To avoid confusion with h , total enthalpy of species j at T , the notation is slightly different from
j0 0
that used in chapter 1. Ed.

In this equation, expression cpj j depends on temperature T and on the values for
j
mass uxes j of the species. In order to simplify this expression we adopt a mean
value cp independent from temperature and the composition of the mixture so that
Eq. (6.126) may be written in the form
dT ( )
=m h0j (j jf ) + cp (T Tf ) . (6.127)
dx j
The value for cp is derived from (6.127) when expressing the condition that
(6.127) vanishes at the cold boundary in agreement with (6.121) obtaining
0
j hj (j0 jf )
cp = , (6.128)
T f T0
or else since, as before said, jf = Yjf and j0 = Yj0 due to the fact that diffusion is
absent in both limits 0
j hj (Yj0 Yjf )
cp = . (6.129)
T f T0
If, as before done, we introduce dimensionless temperature = T /Tf into Eq. (6.127),
this may be written
d ( l
h0j )
= mcp 1 + (j jf ) . (6.130)
dx c T
j=1 p f
It is also convenient here to eliminate variable x, by dividing the reaction equa-

tion (6.109.a) and diffusion Eq. (6.119) by (6.130), as it was done for the case of two
chemical species. Thus the system of the ame equations reduces to the following
Reaction equations

r
l
00
(ij 0
ij )ij j nj eaj / 0
Xs sj
di j=1 s=1
= , (i = 1, 2, . . . , l), (6.131)
d f
l
1+ qj (j jf )
j=1
where, by analogy with the case of two chemical species, we have taken
nj
Mi f Bj Tf j
ij = , (6.132)
m2 cp
Ej
aj = , (6.133)
RTf
and
hj
qj = . (6.134)
cp Tf
In comparison with the case of only two chemical species, we observe here the appear-
ing of a set of parameters ij . However, all these parameters have only one unknown
quantity, the value of m, this is to say the ame velocity, therefore the set ij may be
expressed as a function of just one of the parameters.
Diffusion equations

l ( )
Mi
Lij Xi j Xj i
dXi j=1
Mj
= , (i = 1, 2, . . . , l), (6.135)
d
l
1+ qj (j jf )
j=1
where
RT
Lij = , (6.136)
Mi cp pDij
Boundary conditions
The boundary conditions reduce to the following
0 : i i0 , Xi Xi0 , (6.137)
1 : i if , Xi Xif . (6.138)
Like in the case of two chemical species the difculty at the cold boundary may be
eliminated by introducing an ignition temperature i whose value does not inuence
the result when the reduced activation energies aj , or at least some of them, are large
as normally happens in practical cases for the same reasons stated in detail in the two
species case.
A recount of the number of boundary conditions when compared to the number

of equations proves that the conditions are superabundant and hence, an eigenvalue of
m, this is, one of the values of ij , must exist which makes compatible the said
conditions and therefore the problem is essentially the same than in the case of two
species only.
6.13. OZONE DECOMPOSITION FLAME 169
Be ij the parameter chosen to express the rest of them as functions of it. Once
its value is known, the value for velocity u0 of the ame, by virtue of Eqs. (6.22) and
(6.132), is v
u
u Mi f Bj T j nj
t f 1/2
u0 = ij . (6.139)
20 cp
6.13 Ozone decomposition ame
The ozone decomposition ame in a mixture of this gas with oxygen is the rst ex-
ample we have chosen among the few cases of ame propagation which have been
calculated. Its rst theoretical study was due to Lewis and von Elbe [8] who thus be-
came the rst to analyze ame propagation considering, in addition to the effects of
heat transfer, those of the diffusion of reactants and products, as well as the inuence
of the radicals (oxygen atoms) in the process. For this purpose they applied the law of
chemical reaction which includes the inuence of temperature and the concentrations
of the various species on the reaction rate. The three chemical species in the process
are ozone, molecular oxygen and atomic oxygen. Lewis and von Elbe simplied the
problem by adopting the following assumptions:
1) The Lewis-Semenov number for the mixture O2 - O3 is equal to unity. This

assumption enables the expression of the concentrations of ozone and molecular
oxygen as functions of temperature.
2) The chemical processes take place in accordance with the following kinetic
scheme
O3 O2 + O, (6.140)
O + O3 2 O2 . (6.141)
The rst of these two reaction equations expresses that the concentration of
atoms of oxygen at each point of the ame is the one that would correspond
to a mixture of O, O2 and O3 in chemical equilibrium at the temperature and
composition of the point.
Having stated the problem, and if, furthermore, we take into account that the
molar fraction of atomic oxygen is much smaller than unity, its solution is straightfor-
ward.
In fact, the rst of the above assumptions allows the expression of the molar
fractions of O2 and O3 as functions of temperature, as shown in 9 of this chapter,
and equation (6.140) allows the same for the molar fraction of O.
Thus, the problem reduces to the integration of only one reaction equation
(that corresponding to O2 or to O3 ), which may be performed by means of one of the
methods enumerated in the aforesaid.
Lewis and von Elbe carried out a more arduous calculation, through the numer-
ical integration of the resulting equation. A detailed information on their calculations
may be found in the above reference or in [9]. Even when the values of the ame ve-
locity obtained from this calculation are several times larger than those experimentally
observed, considering the state of knowledge at the time, their work was published,
the fact that predicted and experimental values were of the same order of magnitude
was considered an important success.
Recently, von Karman and Penner [10] have recomputed the ame velocity for
a set of mixtures of O2 and O3 , using the kinetic scheme proposed by Lewis and von
Elbe, but applying the semi-analytical methods developed in 9. The results agree
with those obtained by Lewis and von Elbe and they shown that the inuence of the
composition of the mixture on velocity of the ame obtained through this procedure,
differ essentially from those observed by experimentation. This discrepancy is basi-
cally due to the fact that the set of chemical reactions proposed by Lewis and von Elbe
and, in particular, the distribution of oxygen atoms within the ame, are different from
the actual ones.
Hirschfelder and his associates [25] have recently calculated the same ame,
applying the following complete kinetic scheme
O3 + G O + O2 + G 1)
O + O2 + G O3 + G 2)
O + O3 2 O2 3)
(6.142)
2 O2 + G O + O3 4)
O2 + G 2 O + G 5)
2 O + G O2 + G 6)
This computation was carried out by means of an arduous numerical integration

for which electronic computers were used.
Later on, von Karman and Penner [6] performed a simplied analysis of the
problem, utilizing the same kinetic scheme and physico-chemical constants as Hirsch-
felder but applying Karmans analytical method of integration, and postulating a dis-
tribution of oxygen atoms within the ame determined through the extension of the
steady state assumption which is widely used in classical Chemical Kinetics.
The following is a summary of the von Karman-Penner study which can be

found fully described in the papers of Refs. [6] and [10].
If we assign subscripts 1, 2 and 3 to species O, O2 and O3 respectively, and

making use of the ame equations derived in 11 of this chapter, we obtain the fol-
lowing system.
Reaction equation
It is sufcient to write two equations corresponding, for example, to species O2 and

O3 thus obtaining after Eq. (6.131)
d1 [
= 11 3/2 ea1 / X3 12 5/2 ea2 / X1 X2
d f
13 3/2 ea3 / X1 X3 + 14 3/2 ea4 / X22
] (6.143)
+ 215 3/2 ea5 / X2 216 5/2 ea6 / X12
[ ]1
1 + q1 (1 1f ) + q2 (2 2f ) + q3 (3 3f ) ,
d3 [
= 31 3/2 ea1 / X3 + 32 5/2 ea2 / X1 X2
d f
]
33 3/3 ea3 / X1 X3 + 34 3/2 ea4 / X22 (6.144)
[ ]1
1 + q1 (1 1f ) + q2 (2 2f ) + q3 (3 3f ) .
Diffusion equations
Likewise, the diffusion equations corresponding to O and O3 are, according to (6.135)
( ) ( )
dX1 L12 12 X1 1 X2 1 + L13 13 X1 1 X3 1
= , (6.145)
d 1 + q1 (1 1f ) + q2 (2 2f ) + q3 (3 3f )
( )
dX3 L31 (3X1 3 X3 1 ) + L32 32 X3 2 X2 3
= . (6.146)
d 1 + q1 (1 1f ) + q2 (2 2f ) + q3 (3 3f )
The following Table 6.1shows the values of the physico-chemical constants

corresponding to the six reactions of Eq. (6.142).
Order Ej
Reaction j Aj
nj cal/mol
1 2 1/2 24.14 10.56 1012
2 3 1/2 0.00 0.230 1012
3 2 1/2 6.00 7.15 1012
4 2 1/2 99.21 2.93 1012
5 2 1/2 117.34 8.92 1012
6 3 1/2 0.00 0.482 1012
Table 6.1: Physico-chemical parameters for reactions 6.142.
The boundary conditions for this system are as follows:
a) Hot boundary.
At the hot boundary ( = 1) the concentration of ozone is zero and the concen-
tration of atoms of oxygen is very small. Therefore it can be written
=1: 1 1f ' 0, 3 3f = 0, (6.147)

X1 X1f ' 0, X3 X3f = 0. (6.148)
b) Cold boundary.
At the cold boundary since no chemical reactions has been yet produced, the
mass ux of ozone is the one corresponding to the initial mixture and the one of
atomic oxygen is zero
= i : 1 10 = 0, 3 = Y30 (datum). (6.149)
Simplication of the equations
Before proceeding to integrate the system, we will simplify it, following von Karman
and Penner. For the purpose, we shall begin by analyzing the values of the physico-
chemical constants of the reactions which appear summarized in the above Table 6.1,
starting by the activation energies.
We see in this table that for reactions 4 and 5 in (6.142), this is, those producing
atomic oxygen from molecular oxygen, the activation energies are much larger than
for the rest, and consequently the reaction velocities in which they participate will be
very small and may be neglected. Hence, we can eliminate the terms corresponding
to these reactions from equations (6.143) and (6.144).
Let us now consider, coefcients ij which are the ones that may also inuence
the order of magnitude of the reaction velocities. In order to simplify this comparison,
in the following table we give their values, referred to the value of 31 which we are
adopting as a reference according to the reasons stated in the foregoing.
11 12 13
= 0.333 = 1.162 103 Tf1 = 0.226
31 31 31
14 15 16
= 0.925 = 0.282 = 2.435 103 Tf1
31 31 31
32 33 34
= 3.486 103 Tf1 = 0.677 = 0.277
31 31 31
Table 6.2: Values of the ratios ij /31 appearing in Eqs. (6.145) and (6.146).
It appears here that 12 , 16 and 32 , which correspond to reactions 2 and 6

in (6.142), are much smaller than the others, due to the following two reasons. First,
because the corresponding coefcients Aj are smaller, as shown by Table (6.1), and
second because they are reactions requiring triple collision, as shown by Eq. (6.142),
which explains the presence of term Tf1 in these coefcients.
Hence, it results that we can also eliminate the corresponding terms in reaction
equations (6.143) and (6.144).
Finally, with reference to the denominator of these equations, it can also be

simplied, keeping in mind that:
1) In accordance with boundary conditions (6.147) it is 1f = 3f = 0.

2) Moreover, 1 is very small compared to unity throughout the ame, and there-
fore, the corresponding term may be neglected.
3) Finally, in accordance with Eq. (6.124) and considering that h02 is zero, it is
q2 = 0.
Taking into account these considerations, the said denominator reduces to the
following
1 + q1 (1 1f ) + q2 (2 2f ) + q3 (3 3f )1 ' 1 + q3 3 . (6.150)
Consequently if we introduce the above simplications into the system of re-

action equations (6.143) and (6.144), this system reduces to
( )
d1 3/2 ea1 / X3 ea3 / X1 X3
= , (6.151)
d f 1 + q3 3
( )
d3 3/2 ea1 / X2 + ea3 / X1 X3
= . (6.152)
d f 1 + q3 3
Steady state assumption
In spite of the simplication introduced into the system, it is still very difcult to
perform its integration by semi-analytical methods.
In an effort to nd further simplications, von Karman and Penner checked the

applicability of the steady state assumption for the distribution of atomic oxygen. This
assumption is expressed through nullifying the reaction velocity of the oxygen atoms,
thus obtaining
ea1 / X3 ea3 / X1 X3 = 0, (6.153)
from which we derive the following distribution of atomic oxygen throughout the
ame
a1 a3

X1 = e . (6.154)
10
9
Hirschfelder
8
KarmanPenner
7
6
1
10 X
5
4
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Figure 6.16: Mole fraction of atomic oxygen as a function of the temperature for the ozone
decomposition ame.
Figure 6.16 taken from [6] compares the distribution of X1 corresponding to

this assumption, with that obtained by Hirschfelder through numerical integration of
the ame equations, without making use of the assumption. It is seen that the result is
completely satisfactory and it plainly justies the assumption of Karman and Penner,
which is not applicable just within a narrow zone very close to the maximum temper-
ature, at which, since the concentration of ozone is practically reduced to zero there
are other reactions, different from 1 and 3 in (6.142) controlling the formation of O.
But this zone is unimportant when studying the ame structure and in particular its
propagation velocity.
The applicability of the steady state assumption is an important contribution

to the theory of ames which essentially simplies. The success reached with ozone
ames has encouraged the idea of applying this assumption to the study of similar
practical cases, as will be seen later on.
Recently, however, the applicability of such an assumption has been discussed,

among others, by Campbell [26], Giddings and Hirschfelder [27] and Spalding [28].
They based their objections rst on the fact that the crossing of the ame is so rapid
that there is not sufcient time for the formation of radicals to reach the level required
by the condition of equilibrium established by the said assumption. On the other hand,
even if such level were reached the diffusion of radicals would change substantially
the distribution. Although other studies by Gilbert and Altman [29] and by Millan
and Sanz [30] tend to conrm the applicability of the steady state assumption for the
cases studied in the foregoing paragraphs, however, the problem cannot be considered
as solved, until a general study enables the establishment of the conditions that must
be satised in a ame so that this assumption be valid.
Let us proceed with the study of the ozone ame. The application of the steady
state assumption eliminates Eq. (6.151) and reduces (6.152) to the following
d3 3/2 ea1 / X3
= 2 , (6.155)
d f 1 + q3 3
where, to simplify notation, we have written
M1 f B3
13 = = 3/2
. (6.156)
m2 cp Tf
Taking into account the preceding considerations, the diffusion equation reduces to
dX3 3
X3 3 12 X3 3
=L2 , (6.157)
d 1 + q3 3
where also, for shortness, we write
RT
L13 = L = . (6.158)
M1 cp pD13
The problem reduces now to the integration of these two equations with the following
boundary conditions
= i : 3 = 0, (6.159)
=1: 3 = X3 = 0. (6.160)
The compatibility of these three conditions determines, the eigenvalue of , as es-

tablished in 9, and once it is determined,we obtain with it the value for the ame
velocity by virtue of Eq. (6.139).
The integration of the system is very simple using Karmans method, in the
following way. In the zone of interest, this is for close to 1, it is 3 X3 and
3 1 , therefore a rst approximation of Eq. (6.157) is the following
dX3 L
' , (6.161)
d q3
where considering that for = 1 it is X3 = 0, we obtain
L
X3 ' (1 ). (6.162)
q3
If this value is taken into (6.155) and integrating between = 1 and = i ' 0 as
expressed in 9 of this chapter, we obtain the desired value of .
Von Karman and Penner have calculated the result for X30 = 0.25, 0.40, 0.50,
0.75, and 1.00. Furthermore, these authors calculated as well the ame velocity for
the same values by means of the Lewis-von Elbe chemical scheme.
A comparison of results with those obtained through experimentation by Grosse

[31] proves that the approximation of the theoretical values reached by Karman and
Penner is satisfactory since the deviations, in both cases, are under 25 per cent.
6.14 Hydrazine decomposition ame
Hydrazine combustion has been experimentally studied by Murray and Hall [32].
Tests were performed by burning mixtures of hydrazine and water vapours in a Bunsen
burner at ambient pressure. The ame propagation velocity was obtained by measur-
ing the area of the luminous cone of the ame.
The analysis of the combustion products shown that they correspond to the
overall reaction
1 1
N2 H4 H2 + N2 + NH3 . (6.163)
2 2
Murray and Hall also calculated the theoretical propagation velocity of the
ame for one of the mixtures experimentally analyzed (97.2% hydrazine). They as-
sumed that such velocity was determined by the unimolecular decomposition reaction
N2 H4 2NH2 , (6.164)
6.14. HYDRAZINE DECOMPOSITION FLAME 177
which is the initial reaction in the complex process that is stoichiometrically repre-
sented by (6.163).
Making use of the thermal theory of Zeldovich and Frank-Kamenetskii [33]

and using the specic reaction rate
k = 4 1012 e60 000/RT , (6.165)
proposed by Szwarc [34], they obtained a theoretical propagation velocity of the ame
of 110 cm/s compared to the 185 cm/s obtained from experiments.
More precise calculations were performed by Hirschfelder [35] and by Karman

and Penner [6], making use of Murray and Halls kinetic scheme with results that
conrm those obtained by Murray and Hall. Hirschfelder and Karman and Penner
also studied the inuence over the propagation velocity of the ame of the diffusion
of the different species. In this case they obtain values considerably smaller than
the experimental ones since diffusion reduces substantially the value of the ames
propagation velocity.
The following table summarizes the results of the works by Murray-Hall and
Karman-Penner.
Experimental value Theoretical values

Refs. Murray-Hall Murray-Hall Karman-Penner
Zero diffusion Zero diffusion Non-zero diffusion 10
u0 185 cm/s 110 cm/s 95 cm/s 42 cm/s
Table 6.3: Theoretical and experimental propagation velocities of the ame in a mixture of
vapour of hydrazine and water. Mass fraction of hydrazine = 0.972. Pressure =
1 atm. Initial temperature of the mixture = 150 C.
If the reaction controlling the process is the one given by Eq. (6.164) with the
specic reaction rate (6.165), the following two conditions must be satised:
1) Since chemical reaction (6.164) is of the rst order, the propagation velocity of
the ame must be inversely proportional to the square root of pressure.
2) Since the activation energy E of reaction (6.164) is 60 000 cal/mol and the ame
propagation velocity is approximately proportional to exp (E/2RTf ) accord-
ing to Eq. (6.72) where Tf is the ame temperature, when such velocity is repre-
sented as a function of 1/Tf in the logarithmic scale one must evidently obtain
a slope equal to E/2R.
10 For a Lewis number equal to unity.

With the experiments made by Murray and Hall it is not possible to verify the
rst conclusion since they were all performed at ambient pressure, and the second
conclusion was not checked by the authors. In order to verify both, Adams and Stock
[36] measured the combustion velocity of hydrazine at different pressures. For this
purpose, they stabilized the ame over a column of liquid hydrazine contained in a
capillary tube and measured the recession velocity of the liquid level as its vapour
burnt. This method is not practical to measure the absolute propagation velocity of
the ame due to the irregularity of its front but is very convenient for the comparative
study intended by these authors. The results of their experiments show that the in-
uence of pressure is approximately that corresponding to a rst order reaction. The
discrepancies could be due to the experimental method used.
As for the apparent activation energy it decreases with the drop in ame tem-
perature and it is in all cases way under the 60 000 cal/mol that correspond to the
process proposed by Szwarc and applied to the theoretical studies mentioned herein.
Such result led Adams and Stocks to conclude that the decomposition of hy-
drazine should occur through a complicated system of chain reactions of which (6.164)
is only the initial one. As a result of a detailed discussion of all possible reactions the
above mentioned authors proposed a simplied kinetic scheme [36], which summa-
rizes the actual process. This proposed scheme considers only three chemical species:
reacting species A (hydrazine), stable products C of the combustion (mixture of NH3 ,
N2 , H2 , etc.), and one radical B, that propagates the chain (which could be NH2 , H,
etc). The Adams and Stock simplied scheme is a follows
A 2 B, (6.166)
B + A B + 2 C, (6.167)
B + B + X 2 C + X. (6.168)
Of the three reactions, the rst initiates the chain, the second is the propagation reac-
tion and the third the chain breaking reaction.
Lately, Spalding [37] has calculated the propagation velocity of the ame with
the scheme proposed by Adams and Stocks and he applied an interesting method of
numerical integration developed by him. Spalding calculates two cases corresponding,
respectively, to a cold ame and to one of the hot ames experimented by Murray
and Hall. In both cases he gets results in close agreement with those experimentally
obtained by Murray and Hall. These results are summarized in Table 6.4.
Theoretical Experimental
(Spalding)
Hot ame 190 cm/s 185 cm/s (Murray-Hall)

Cold ame 12 cm/s 10 cm/s (Adams-Stocks)
Table 6.4: Propagation velocity of the ame in cm/s.
As Spalding points out the excellent numerical agreement is casual to a certain

extent since there are doubts respect to the correct values of the transport coefcients.
Moreover, Spalding did not include the inuence of the variation of such coefcients
with the temperature across the ame. More interesting is the fact that Spaldings
results predict correctly the inuence of the combustion temperature on the velocity
of the ame.
Spalding also concludes that the propagation velocity of the ame that would
be obtained when calculating the radical distributions by means of the steady state
assumption would be much too large. He bases this conclusion upon the fact that
the distribution of radicals obtained with the steady state assumption is considerably
larger, at temperatures close to combustion temperature, than the distribution given by
correct calculation.
Millan and Sanz [30] calculated the propagation velocity of hydrazine decom-
position ame applying the same simplied model proposed by Adams and Stocks.
Two cases were considered, the rst assuming steady state for concentration of radi-
cals and solving the ame equations through Karmans method; the second, for which
an approximate calculation method was developed to obtain ame velocity without
considering the steady state assumption. The authors concluded that the results ob-
tained applying the steady state assumption to the radical concentrations are satis-
factory when compared to those corresponding to a more correct distribution for the
same.
Simultaneously, Gilbert and Altman [38] obtained the ame propagation ve-
locity corresponding to complete kinetic model proposed by Adams and Stocks [36]
through the application of the Boys-Corner method in nding the solution of the ame
equations and calculating the concentration of radicals throughout the ame under the
steady state assumption introduced by Karman and Penner [6].
Recently [39] a new analysis of the problem was conducted using the complete
kinetic model of Adams and Stocks, and adopting the steady state assumption for the
determination of the concentration of radicals, with Karmans method in calculating
the solution.
Kinetic model of the reaction of hydrazine decomposition
The mechanism of decomposition proposed by Adams and Stocks is as follows
N2 H4 2NH2 1)
N2 H4 + NH2 NH3 + N2 H3 2)
N2 H3 N2 + H2 + H 3)
(6.169)
H + N2 H4 2NH3 + NH2 4)
H + NH2 + X NH3 + G 5 c)
H + H + X H2 + G 5 d)
In these equations, G represents any one of the particles of the mixture.
In order to identify the molecules of the components of the mixture the follow-
ing subscripts will be used
1 N2 H4 reactant
2 NH3 product
3 N2 product
4 H2 product (6.169.a)
5 NH2 radical
6 N2 H3 radical
7 H radical
According to the law of the mass action, the velocity reaction wi of the radicals
must be expressed as a function of the mole concentration in the following way.
w5
For NH2 : = 2k1 cX1 k2 c2 X1 X5 + k4 c2 X1 X7 k5c c3 X5 X7 , (6.170)
M5
w6
For N2 H3 : = k2 c2 X1 X5 k3 cX6 , (6.171)
M6
w7
For H : = k3 cX6 k4 c2 X1 X7 k5c c3 X5 X7 2k5d c3 X72 . (6.172)
M7
Likewise, the hydrazine decomposition velocity is
w1
= k1 cX1 + k2 c2 X1 X5 + k4 c2 X1 X7 . (6.173)
M1
Here, Xi are the mole fractions of the different species, ki the specic reaction veloc-
ities given by Eq. (6.112) (subscript of k indicates the corresponding reaction in ac-
cordance with the number assigned to them in the preceding paragraph) and c = /M
is the mole concentration per cm3 .
The following values, also applied by Gilbert and Altman [38], are adopted for
specic velocities
k1 = 4 1012 e60 000/RT s1 (6.174)
k2 = 1013 e4 600/RT cm3 mol1 s1 (6.175)
k4 = 1013 e7 000/RT cm3 mol1 s1 (6.176)

k5c = 5 1015 cm6 mol2 s1 (6.177)
16 6 2 1
k5d = 10 cm mol s (6.178)
Reaction velocity of hydrazine under to steady state assumption for

radicals
The steady state assumption is expressed by making to reaction velocities of the radi-
cals referred by 5, 6 and 7 in (6.187.a) equal to zero. Thereby
k2 cX1 X5 = k3 X6 , (6.179)
cX1 (k2 X5 k4 X7 ) = 2k1 X1 k5c c X5 X7 , 2
(6.180)
cX1 (k2 X5 k4 X7 ) = k5c c2 X5 X7 + 2k5d c2 X72 . (6.181)
Since mole fraction X7 of radical H is small throughout the reaction with

k5c cX7 2k1
respect to reactant X1 , it occurs that is negligible when compared to .
k2 X1 k2 cX5
Hence Eqs. (6.180) and (6.181) may be written
2k1
k2 X5 k4 X7 = , (6.182)
c
2k5d cX72
k2 X5 k4 X7 = . (6.183)
X1
Consequently, the concentration of radicals H is supplied by

k1
X7 = X1 . (6.184)
k5d c2
From (6.182) and (6.179) the following expressions may be deduced for mole
fractions X5 of NH2 and X6 of N2 H3 :
2k1 + k4 cX7
X5 = , (6.185)
k2 c
2k1 + k4 cX7
X6 = X1 . (6.186)
k3
If the values for X7 and X5 given by (6.184) and (6.185) are substituted into
the hydrazine reaction velocity given by (6.173), the following is obtained
w1 1/2
= cX1 (3k1 + kX1 ), (6.187)
M1
where ( )1/2
4k42 k1
k= = 4 1011 e37 000/RT . (6.188)
k5d
It is easily seen that the rst term in the parenthesis of Eq. (6.187) may be neglected
respect to the second. Hence it nally results
w1 3/2
= kcX1 . (6.189)
M1
Let us assume that the proportion between concentrations of products NH3 , N2 and
H2 is the one corresponding to the complete reaction, and the concentrations of rad-
icals are very small, then the following relations will be obtained between the mole
fractions and the mass fractions of the main species
X1 + 2X2 = 1, (6.190)
Y1 X1
= , (6.191)
M1 M
Y2 X2
= . (6.192)
M2 M
Furthermore, the mean mole mass of the mixture will be
( )
1 1
M = M1 X1 + X2 M2 + M3 + M4 . (6.193)
2 2
From the system of equations (6.190) to (6.193) it results
32
1 Y2
X1 = 17 . (6.194)
32
1 + Y2
17
The reaction velocity of ammonia results to be
3/2
32
1 Y
w2 w1 17 .
2
= = kc
32
(6.195)
M2 M1
1 + Y2
17
Energy equation
When mean specic heat is assumed to be constant and the inuence of radicals on
the energy of the mixture is neglected, the energy equation may be written
( )
dT 17
= qr 2 cp (Tf T ), (6.196)
m dx 32
where qr is the reaction heat per gram of ammonia produced. If reduced temperatures
T T0
= and 0 = (6.197)
Tf Tf
are introduced into Eq. (6.196) and taking into consideration the conditions at the cold
boundary, = 0 , 2 = 0, this equation may be written
d
= (1 0 )(1 ) (1 ), (6.198)
mcp dx
where
32
= 2 . (6.199)
17
Chemical reaction and diffusion equations
The reaction equation for ammonia is

d2
m = w2 . (6.200)
dx
The diffusion equation corresponding to the same is given by Ficks law and written
dY2 m
= (Y2 2 ). (6.201)
dx D2m
Coordinate x is eliminated from the above two equations through Eq. (6.198), and the
following system results
d 32 w2
= , (6.202)
d 17 m2 cp (1 0 )(1 ) (1 )
dY Y
=L , (6.203)
d (1 0 )(1 ) (1 )
where
32
Y = Y2 , (6.204)
17
and

L= (6.205)
Dcp
is the Lewis-Semenov number.
Taking (6.195) into (6.202)

( )3/2 1
1Y a
e
d 1+Y
= , (6.206)
d (1 0 )(1 ) (1 )
where ( p )
= 128 1011 ea . (6.207)
RT m2 cp
If a linear variation of with temperature is assumed
= f , (6.208)
we obtain from Eq. (6.207).

( )
p f
= 128 1011 ea . (6.209)
RTf m2 cp
Then the problem reduces to the integration of the system of Eqs. (6.203) and (6.206),
with the following boundary conditions
= 0 : = 0, (6.210)
=1: = 1, Y = 1. (6.211)
The value of which makes compatible both boundary condition will be the
eigenvalue for the system, and it determines the ame propagation velocity.
Flame velocity
The ame velocity is given by

( )1/2
m 1 p f
u0 = = 128 1011 ea 1/2 , (6.212)
0 0 RTf cp
where is the eigenvalue referred to.
When Eq. (6.206) is written

( )3/2 1
1 0 1 1
1Y a
(1 ) d = e d, (6.213)
2 0 0 1+Y
the problem reduces (following the idea of Karmans method) to nding an approxi-
1
mation of vs for the calculation of integral (1 ) d, and an approximation of
0
Y vs for the computation of the integral of the right hand side.
The parabolic approximation given by Eq. (6.87) is adopted for Y vs

( )L
0
Y = , (6.214)
1 0
which satises Eq. (6.203) for = 0, and for = 1 is tangent to the straight line
1
1Y =L , (6.215)
1 0
which, in turn, is also tangent to the solution of Eq. (6.216) as it can easily be veried
when considering that for ' 1 is
1 1 Y, (6.216)
and
1
1
. (6.217)
1 0
When approximation (6.214) is taken into Eq. (6.213), it results
1
1 0
(1 ) d = IK , (6.213.a)
2 0
where
1
(
0 L
) 3/2
1
1
1 0 a
IK = ( ) e d. (6.218)
0 L
0 1 +
1 0
1
The approximation of vs for the computation of (1 ) d can be ob-

tained from the solution of Eq. (6.206) for values of close to 1. Through the proce-
dure applied in Ref. [30] , it results11
[ ( )3/2 ]2/5
1 4/5 4 L
= (1 ) . (6.219)
1 0 5 2
Therefore
[ ( ) ]2/5
1 3/2
5 4 L
(1 ) d ' (1 0 ) 2/5 . (6.220)
0 9 5 2
If Eqs. (6.218) and (6.220) are taken into Eq. (6.213) the following equation will be
obtained for the eigenvalue
[ ( ) ]2/5
3/2
1 0 10 4 L
= 1 2/5 , (6.221)
2IK 9 5 2
which is solved through iteration or else graphically.

11 Essentially, it reduces to neglecting (1 ) respect to (1 ) in the denominator of Eq. (6.206) and to
apply the linear approximation given by Eq. (6.215) in order to eliminate Y in the numerator of the same.
In order to perform the numerical calculation of the ame velocity the follow-
ing values given by Hirschfelder were applied. These same values were also used by
Gilbert and Altman in his work:
0 = 9.17 104 gr/cm3 ,

T0 = 423 K,
Tf = 1933 K,
cal (6.222)
= 0.00067 ,
cm s K
cal
cp = 0.6623 ,
gr K
L = 1.33.
With them, the reduced activation temperature results to be
37 000
a = = 9 632. (6.223)
1.933 1.9872
The numerical integration of Eq. (6.218) gives
IK
= 270.9 105 . (6.224)
1 0
When the precedent values are taken into (6.221)
1/2 = 8.2962 102 , (6.225)
which, when taken into (6.212), nally results
1/2
u0 = ' 209 cm/s. (6.226)
3.97
The agreement between this value and the 200 cm/s given by experimental results is
better than it could be expected considering the dubiousness of many of the numerical
values used.
Inuence of pressure
As shown in Eq. (6.212) the ame velocity varies with pressure in accordance with
the following law
p1/2
u0 p1/2 , (6.227)
0
in agreement with the predictions of theory for rst-order reaction ames Eq. (6.75)
and as veried by the experiments carried-out by Adams and Stocks [36].
6.15. FLAME PROPAGATION IN HYDROGEN-BROMINE MIXTURES 187
6.15 Flame propagation in Hydrogen-Bromine mixtures
The present example contains a calculation of the velocity of the hydrogen-bromine

ame performed with a variation of the method developed by von Karman and Pen-
ner [40]. The method presented here differs from earlier calculations because the
inuence of dissociation of bromine on the ame velocity is evaluated properly. Von
Karman and Penner included the inuence of the atoms of bromine on the reaction
rate but when calculating the distribution of the mole fraction of Br2 , H2 and HBr,
they neglected the mole fraction of Br atoms. However, for temperatures close to the
maximum ame temperature the dissociation of Br2 reduces substantially the value
for the mole fraction of Br2 , which has an important inuence on the ame veloc-
ity. Hence it seemed desirable to compute the burning velocity correctly, especially
in hydrogen-rich ames where the inuence of dissociation is most important. Von
Karman and Penner also neglect the inuence of the energy of dissociation of Br2 ,
which should be taken into account since it may be of the same order of magnitude as
the energy transferred by convection through the ame.
In the following, a method is developed for the correct calculation of the mole
fraction of Br2 and of the inuence of its dissociation energy on the velocity of the
ame, within the limitations of the steady-state assumption for the distribution of
bromine and hydrogen atoms. This method will be applied to computation for the
four hydrogen-rich ames previously considered by von Karman and Penner. The re-
sults are compared with those obtained by these authors and are shown to differ by not
more than 25% even for the hottest ame.
A detailed study of the structure of the ame is also presented. It will be shown
that dissociation of Br2 reduces the ame velocity because of a decrease in the number
of molecules of bromine which exist near the hot ame boundary. On the other hand,
it will be seen that the inuence of the dissociation energy can be neglected if thermal
convection is ignored because these two effects tend to cancel.
The procedure developed here follows closely the work performed in Ref. [41]
and could also be useful for computations on other types of ames.
Flame Equations
Since there are only ve different chemical components, namely, HBr, Br2 , H2 , Br
and H, the unknowns are:
1) Five ux fractions, i .
2)
Expressions for the reaction rates wi are obtained from chemical kinetics. Be-
tween the ve chemical components, the following independent chemical reactions
exist [42]
k
Br2 + X
1
2Br + X, (6.233)
k
Br + H2
2
HBr + H, (6.234)
k
H + Br2
3
HBr + Br, (6.235)
k
HBr + H
4
H2 + Br, (6.236)
k
2Br + X
5
Br2 + X. (6.237)
These equations give the following expressions for the reaction rates w1 , w4
and w5
( )2
p
w1 = M1 2 (k2 X3 X4 + k3 X2 X5 k4 X1 X5 ), (6.238)
RTf
( )
p p 1 2 M4
w4 = M4 1 (2k1 X2 2k5 X4 ) w5 , (6.239)
RTf RTf M5
( )2
p
w5 = M5 2 (k2 X3 X4 k3 X2 X5 k4 X1 X5 ). (6.240)
RTf
Energy Equation
The energy equation is
d ( 5
h0 )
= mcp ( 1) + (i if ) i . (6.241)
dx i=1
c p Tf
Diffusion Equations
Between mole fractions Xi the following relation exists
X1 + X2 + X3 + X4 + X5 = 1. (6.242)
Consequently, four additional equations are needed in order to complete the determi-
nation of all of the mole fractions. The remaining equations are those for the diffusion
of four of the components, and they are obtained from Eq. (6.119)
( )
RTf 1
5
dXi j i
=m Xi Xj , (i = 1, 2, 3, 4). (6.243)
dx p j=1 Dij Mj Mi
Equations (6.228) through (6.232), (6.241), (6.242) and the four expressions
contained in Eq. (6.243) form a system of eleven equations for the computation of
the eleven unknowns of the ame. In the following section we simplify the problem
by introducing the steady-state approximation for the concentration of bromine and
hydrogen atoms.
The steady-state approximation
The following is the expression of the steady-state assumption for bromine and hydro-
gen atoms, respectively
w4 = 0, (6.244)
w5 = 0. (6.245)
If Eqs. (6.239) and (6.240) are combined with Eqs. (6.244) and (6.245), we ob-
tain for the mole fractions of Br and H the following relations in terms of the principal
components and of the temperature

k1 RTf
X4 = X2 , (6.246)
k5 p
and
k2 k1 RTf X3 X2
X5 = . (6.247)
k3 k5 p k4
X2 + X 1
k3
From Eqs. (6.238), (6.242) and (6.246) the following expression is deduced for w1
( )2
p
w1 = 2M1 2 k3 X2 X5 . (6.248)
RTf
Using in Eq. (6.248) the value for X5 given in Eq. (6.247), we obtain
( )3/2 3/2
p k1 3/2 X3 X2
w1 = 2M1 k2 . (6.249)
RTf k5 k4
X 2 + X1
k3
Equation (6.249) expresses the rate of formation of HBr as a function of the mole
fractions of the principal components and of temperature. Introduction of Eq. (6.249)
into Eq. (6.230) leads to the result
( )3/2 3/2
d1 p k1 3/2 X3 X2
m = 2M1 k2 . (6.250)
dx RTf k5 k4
X2 + X1
k3
The steady-state assumption implies the hypothesis that the diffusion of radicals may
be neglected. Therefore, the ux fractions of Br and H must be proportional to their
corresponding mole fractions, viz.,
M4
4 = X4 , (6.251)
Mm
M5
5 = X5 . (6.252)
Mm
The mole fraction of hydrogen, X5 , is always negligibly small. Therefore, in

view of Eq. (6.252), the corresponding ux fraction may also be neglected.
The preceding assumption and considerations allow a considerable simplica-

tion of the ame equations. Since
X5 1 and 5 1, (6.253)
Eqs. (6.228) and (6.229) reduce to
1 + 2 + 3 + 4 = 1, (6.254)
X1 + X2 + X3 + X4 = 1. (6.255)
The mole fraction of Br, X4 , is expressed as a function of X2 through. Eq. (6.246).

The rate of formation of HBr is given by Eq. (6.250). The energy equation, see
Eq. (6.254), now reduces to
d ( )
= mcp 1 + q1 (1f 1 ) q4 (4f 4 ) , (6.256)
dx
where ( )
M4 0 M5 0 1
q1 = h + h h01 (6.257)
M1 2 M1 3 cp Tf
and
1
q4 = (h04 h02 ) (6.258)
cp Tf
are, respectively, the reduced reaction heats per unit mass corresponding to reactions
1 1
H2 + Br2 HBr + M1 q1 cp Tf , (6.259)
2 2
and
Br + Br Br2 + M2 q4 cp Tf . (6.260)
Finally, by virtue of Eqs. (6.251) and (6.252), only two of the four diffusion equations,
see Eqs. (6.243), remain. Moreover, the terms involving X5 and 5 may be ignored.
Selection is made of the two equations corresponding to X1 and X3 because that

corresponding to X2 presents some difculties, as will be seen later on. Thus
( )
RTf 1
4
dX1 j 1
= X1 Xj , (6.261)
dx p j=1 D1j Mj M1
and
( )
RTf 1
4
dX3 j 3
= X3 Xj . (6.262)
dx p j=1 D3j Mj M3
Hence the system of the ame equations has been reduced to the ve rela-
tions given in Eqs. (6.229), (6.246), (6.251), (6.254) and (6.255) and to the four
differential equations (6.250), (6.256), (6.261) and (6.262) for the nine unknowns i
(i = 1, 2, 3, 4), Xi (i = 1, 2, 3, 4) and . If, furthermore, the differential equations
(6.250), (6.261) and (6.262) are divided by Eq. (6.256), the coordinate x is eliminated
and is introduced as independent variable. The results are
( )3/2
d1 p 3/2 k1
= 2 2M1 k2
d m cp RTf k5
( )
3/2 k4
X3 X 2 X2 + X 1 1
k3
. (6.263)
1 + q1 (1f 1 ) q4 (4f 4 )
( )
4 1 j 1
X1 Xj
dX1 RTf j=1 D1j Mj M1
= , (6.264)
d mcp p 1 + q1 (1f 1 ) q4 (4f 4 )
( )
4 1 j 3
X3 Xj
dX3 RTf D
j=1 3j Mj M3
= . (6.265)
d mcp p 1 + q1 (1f 1 ) q4 (4f 4 )
Solution of the reaction equations

k1 k4
Von Karman and Penner [40] give the following expressions for k2 and
k5 k3

k1
k2 = 0.8 1014 e1 / , (6.266)
k5
and
k4 1
= = 0.119, (6.267)
k3 8.4
where
40 200
a = (6.268)
RTf
is the reduced activation temperature for the formation of HBr. These authors assume
also that the coefcient of thermal conductivity is independent of the composition
of the mixture and that it varies proportionally with the square root of the temperature,
viz.,

= f . (6.269)
If Eqs. (6.266), (6.267), and (6.269) are introduced into Eq. (6.263), we obtain
1
a
1 X3 X2 (X2 + 0.119X1 )1 e
3/2
d1
= , (6.270)
d 1 + q1 (1f 1 ) q4 (4f 4 )
where
( )3/2
f p
= 1.6 1014 M1 ea (6.271)
m2 cp RTf
is the eigenvalue which determines the propagation velocity u0 of the ame through
the relation
( )3/4
7 1 f M1 p
u0 = 1.265 10 ea /2 1/2 . (6.272)
0 cp RTf
The eigenvalue can be obtained by applying the method of von Karman

and Penner [6] which involves integration of Eq. (6.270) between the cold and hot
boundaries of the ame in the following form
1
1f ( ) 3/2 a
21f 1
X 3 X2 e
q1 1 + q4 (4f 4 ) d1 = d. (6.273)
2 0 0 (X2 + 0.119X1 )
The problem lies now in obtaining:
1) Approximate expressions for (1 ) and (4f 4 ) as function of (1f 1 )

for the computation on the integral on the left hand of Eq. (6.273).
2) Approximate expressions for X1 , X2 , and X3 as functions of , for the compu-
tation of the integral appearing on the right hand side.
Von Karman and Penner [40] performed this computation by assuming that
X4 as well as 4 may be neglected (compared with unity). In this case, the diffusion
equations (for Lewis number close to unity) show that X1 , X2 , and X3 are linear
functions of 1 near = 1; a study of Eq. (6.270) shows that 1f 1 is a function
of of the form
1f 1 = (1 )n (6.274)
near = 1, where the exponent n takes the following values
X3,0 < 0.5 : n = 1,

7
X3,0 = 0.5 : n= , (6.275)
4
5
X3,0 > 0.5 : n= .
4
The approximation given in Eq. (6.274) holds only for values of very close
to unity. In fact, the values for 1f 1 given by such an approximation decrease very
rapidly with 1 , reaching the value zero for values of very close to unity. A better
approximation for 1f 1 can be obtained as follows. In Eq. (6.270), (1 ) as well
as q4 (4f 4 ) are much smaller than q1 (1f 1 ) for the interesting range of values
of .13 Therefore, a good approximation for Eq. (6.270) can be obtained if these terms
are neglected, thus yielding
1
3/2 a
d1 X3 X 2 e
(1f 1 ) = . (6.276)
d q1 (X2 + 0.119X1 )
Equation (6.276) may be integrated from the hot boundary forward. The result is

1
1f 1 = q1 f (0 ) d0 , (6.277)
2
where
1
3/2 a
X3 X2 e
f () = . (6.278)
(X2 + 0.119X1 )
It can be seen that Eq. (6.277) is an approximation for (1f 1 ) which is valid
for a range of temperatures far more extensive than that covered by Eq. (6.274).
The improved approximation is necessary when the velocity of the ame is

computed, as it is done here, without neglecting the inuence of 4 and X4 . In fact, if
an approximation similar to Eq. (6.274) is used in this case, when computing the inte-
gral on the left-hand side of Eq. (6.273), the contribution of this integral would be con-
siderably overestimated, as can be veried easily. The approximation of Eq. (6.277)
gives

d1 f ()
= . (6.279)
d 2q1 1 0 ) d 0

f (
13 However a very small interval near = 1 must be excluded.

This expression permits the integral of the left-hand side of Eq. (6.273) to be written
as follows
1f
( )
1 +q4 (4f 4 ) d1
0
1(
) f ()
(6.280)
= 1 + q4 (4f 4 ) d.
2q1 0 1

f (0 ) d0
By using these expressions in Eq. (6.273), the following equation for is obtained
1
21f
q1 f () d
2 0
1(
M4 ) f ()
(6.281)
1 + q4 (X4f X4 ) d = 0.
2q1 0 Mm 1

f (0 ) d0
Here Eq. (6.251) was used in order to eliminate 4 .14 For simplicity the following
notation will be used
1
1 1 3/2 a
X 3 X2 e
I= f () d = d, (6.282)
0 0 (X2 + 0.119X1 )
and 1(
M4 ) f ()
J= (1 ) + q4 (X4f + X4 ) d. (6.283)
0 Mm 1 0 ) d 0

f (
If these expressions are used in Eq. (6.281) and the resulting equation is solved, the

following is obtained for
( )2
q1 21f 1 J 1 J
= + + . (6.284)
2I 8q1 I 2 2q1 I
This value, when substituted in Eq. (6.272), gives the corresponding ame velocity.
Computation of X2 and X4
Equation (6.284) shows that when computing it is only necessary to know the values
for I and J. In order to compute I, one must know f () whereas for the calculation
of J one must know f () and X4 as functions of . Finally, Eq. (6.278) shows that
to obtain f (), one must know X1 , X2 , and X3 as function of . Consequently, the
problem reduces now to the computation of X1 , X2 , X3 , and X4 as functions of .
14 In Eq. (6.281) when changing from to X , it has also been assumed that M
4 4 m is constant through
the ame. Actually its variation is very small.
This can be done by means of the diffusion relations given in Eqs. (6.264) and (6.265),
the steady-state Eq. (6.246), and Eq. (6.255).
Let X1f and X3f be the nal values for X1 and X3 . These variables can be
expressed as
X1 = X1f 1 , (6.285)
X3 = X3f + 3 , (6.286)
where 1 and 3 are functions of which approach zero when approaches one.
When Eqs. (6.246), (6.285) and (6.286) are combined with Eq. (6.255), the
following equation is obtained for the determination of X2

k1 RTf
X1f 1 + X3f + 3 + X2 + X2 = 1. (6.287)
k5 p
The computation that follows holds for hydrogen-rich ames where X2f =
X4f = 0 and the following condition is consequently satised
X1f + X3f = 1. (6.288)
Equation (6.287) can now be written as

k1 RTf
X2 + X2 (1 3 ) = 0. (6.289)
k5 p
On solving Eq. (6.289), one obtains for X2

( )2
k1 RTf 1 k1 RTf
X2 = + 1 3 . (6.290)
4k5 p 2 k5 p

k1
Von Karman and Penner [40] give for the expression
k5
r
k1
= 1.676 e , (6.291)
k5
where
22 605
r = . (6.292)
RTf
Finally, combining Eqs. (6.291) and (6.290) it is obtained
( )2
X2 = g()2 + 1 3 g() , (6.293)
where
r

RTf
g() = 0.838 e . (6.294)
p
Likewise, when Eq. (6.291) is substituted into Eq. (6.246), one nds
r

RTf
X4 = 1.676 e X2 . (6.295)
p
Equations (6.293) and (6.295) permit the computation of X2 and X4 as soon as 1

and 3 are known. Nevertheless, before initiating the calculation of these quantities
let us study the behavior of X2 as a function of , since this relation inuences more
than any other the value of the ame velocity, as shown by Eq. (6.282).
For 1, the following conditions are satised

r RTf
g() 0.838 e 6= 0,
p (6.296)
1 3 0.
Therefore, for near 1

1 3
1, (6.297)
g()2
it is now simple to show from Eq. (6.293) that, near = 1, X2 behaves as follows
2r
1 p 1
X2 ' e (1 3 )2 . (6.298)
2.808 RTf

RTf
On the other hand, when the difference 1 increases, the term 0.838 er /
p
decreases very rapidly since r 1, whilst the term 1 3 increases. Therefore,
when the difference 1 is not very small compared to unity, X2 behaves as follows
X2 ' 1 3 . (6.299)
In the different behaviors of Eqs. (6.298) and (6.299) lie essentially the inu-
ence of the dissociation of bromine on the propagation velocity of the ame. In fact,
if the term X4 is neglected in Eq. (6.255) when computing X2 , as was done in [40],
then the approximation given in Eq. (6.299) holds for all temperatures. On the con-
trary, when the inuence of X4 is taken into account, the values for X2 near = 1
are far smaller for than those given by Eq. (6.299), since then X2 varies as the square
of (1 3 ), as shown by Eq. (6.298). Since the values for X2 are smaller, the value

for the integral I is also smaller, as shown by Eq. (6.28). Finally, if I decreases,
increases as shown by Eq. (6.284). Therefore, according to Eq. (6.272), u0 decreases.
Hence, dissociation of bromine reduces the ame propagation velocity basically be-
cause such dissociation reduces the molar fractions of Br2 at temperatures close to
Tf . Dissociation also inuences the value for J, as shown by Eq. (6.283). However,
this inuence is small and it acts opposite to the inuence of thermal conductivity; the
value for J is very small and its inuence is negligible. This last conclusion will be
veried when numerical calculations are performed later on.
Introducing Eq. (6.298) into Eq. (6.295) we nd, for ' 1, the following
behavior of X4

r RTf
X4 ' 1.676 e (1 3 ), (6.300)
p
thus X4 is a linear function of (1 3 ) and, for ' 1, X4 X2 .
Evaluation of X1 and X3
The evaluation of X1 and X3 can be easily performed from the diffusion relations
Eqs. (6.264) and (6.265).
Equations (6.229) and (6.254) permit us to express 2 and 3 as functions of 1

and 4 as follows
M4
2 = (1f 1 ) 4 , (6.301)
M1
and
M1 M4
3 = 3f + (1f 1 ). (6.302)
M1
By introducing these expressions, together with Eqs. (6.285) and (6.286) , into
Eq. (6.264) and (6.265), and if 4 , X2 and X4 are expressed as functions of (1 3 )
by use of the relations given in Eqs. (6.251), (6.298) and (6.300), two equations are
obtained which depend only on , (1f 1 ), 1 and 3 . These expressions can be
developed in series of these variables near = 1. The results are
[( )
dX1 f RTf M4 X1f M1 M4 X1f 1 X3f
= + +
d pcp q1 M1 M2 D12 M1 M3 D13 M1 D13
( ) ]
1 3 1
+O , , + higher order terms , (6.303)
1f 1 1f 1 1f 1
[( )
=
d pcp q1 M1 M2 D23 M1 M3 D13 M1 D13
( ) ]
1 3 1
+O , , + higher order terms . (6.304)
1f 1 1f 1 1f 1
Hence, for ' 1, we nd

( )
' + + 1 , (6.305)
d pcp q1 M1 M2 D12 M1 M3 D13 M1 D13
( )
' 3 , (6.306)
d pcp q1 M1 M2 D23 M1 M3 D13 M1 D13
which leads to the following approximations for X1 and X3
X1 ' X1f 1 (1 ), (6.307)

X3 ' X3f + 3 (1 ). (6.308)
Thus
1 3 = (1 3 )(1 ) = 2 (1 ), (6.309)
where
2 = 1 3 . (6.310)
In short, the following approximations for X1 , X2 , X3 and X4 are obtained

near = 1
X1 = X1f 1 (1 ), (6.311)
( )2
X2 = g()2 + 2 (1 ) g() , (6.312)
X3 = X3f 3 (1 ), (6.313)
( )
X4 = 2g() g()2 + 2 (1 ) g() . (6.314)
Evaluation of f (), I and J
When computing f (), the only values of X1 , X2 and X3 needed are those for near
1, since the reduced activation temperature a is much larger than unity. The relevant
expressions are given in Eqs. (6.311) through (6.313). Once f () is known, the
integral I can be computed by numerical of graphical integration. Likewise, knowing
f () and X4 , the integral J can be obtained either numerically or graphically.
Computation of the ame velocity
The preceding results may be summarized by the following method for the computa-
tion of the velocity of the ame.
1) Values for 1 , 3 and 2 are obtained from Eqs. (6.305), (6.306) and (6.310).
2) Introducing these values into Eqs. (6.307), (6.308) and (6.312) one obtains X1 ,
X2 and X3 as functions of .
3) Substituting the results into Eq. (6.278), the value of f () is obtained.
4) From the numerical or graphical integration of f () between 0 and 1, the value
of the integral I, dened in Eq. (6.282), is obtained.
5) The numerical or graphical integration of f () between and 1 gives the value
1
of f (0 ) d0 .

1
6) We introduce this value of f () d, the value of f (), computed in (6.230),

and the value for X4 , given by Eq. (6.314), into Eq. (6.283). Then we inte-
grate Eq. (6.283), either numerically or graphically, between 0 and 1 in order to
obtain the value of J.
7) When both the values for I and J are substituted into Eq. (6.284) the value for

is obtained.
8) Finally, u0 is computed from Eq. (6.272).
It has been stated that the inuence of J on the value of u0 is negligible. This
fact allows a considerable simplication of the method since it eliminates steps 5) and
6). By neglecting J in Eq. (6.284), the following simplied equation results

q1 21f
= . (6.315)
2I
Results
The method developed herein has been applied to the computation of the four cases
calculated by von Karman and Penner for hydrogen-rich ames. The results obtained
are given in Table (6.5), in which are also listed the values obtained by von Karman
and Penner.
X3,0 0.55 0.60 0.65 0.70

Von Karman and Penner 30.6 42.3 34.2 23.6
u0 calculated from Eq. (6.315) 22.9 33.7 30.5 21.9
u0 calculated from Eq. (6.284) 23.3 34.2
Table 6.5: Flame velocity (cm/s) for hydrogen-rich H2 -Br2 mixtures.
The physico-chemical constants given by von Karman and Penner were used
in our calculations. The values of these constants, as well as those for the integrals
and parameters needed for evaluation of u0 , are summarized in Table 6.6.
In order to analyze the inuence of convection and of dissociation of bromine

upon the value of u0 , the ame velocity was computed for two cases, using the correct

relation given in Eq. (6.284) for . The results are listed in Table 6.5 and 6.6, which
show that the correction is negligible small due to the cancellation of convection and
dissociation energy effects.
X3,0 0.55 0.60 0.65 0.70

X2,0 0.45 0.40 0.35 0.30
X1,f 0.90 0.80 0.70 0.60
X3,f 0.10 0.20 0.30 0.40
1f 0.997 0.994 0.990 0.984
Tf K 1660 1585 1460 1324
105 f (cal/cm-s-K) 19.8 25.0 29.4 33.1
cp (cal/g-K) 0.105 0.116 0.132 0.151
M (g/mole) 73.1 65.2 57.3 49.4
a 12.2 12.75 13.9 15.28
r 6.81 7.13 7.83 8.55
0 0.1945 0.204 0.221 0.244
1 1.532 1.745 1.762 1.684
2 1.380 1.560 1.560 1.470
3 0.152 0.185 0.202 0.214
q1 0.807 0.800 0.787 0.768
q4 1.63 1.55 1.475 1.42
103 I
0.667 1.601 2.639 3.395
24.54 13.82 12.09 10.47
u0 (cm/s) 551 529 369 228
u0 (cm/s) 22.9 33.7 30.5 21.8
3
10
J -0.87 -0.94
from Eq. (6.284) 24.03 15.49
u0 (cm/s) from Eq. (6.284) 23.34 34.15
Table 6.6: Parameters for the computation of ame velocities.
The results given in Table 6.5 show that dissociation of bromine reduces the
ame velocity considerably. Values for u0 have been plotted in Fig. 6.17 which also
shows the results calculated by von Karman and Penner [40] and by Gilbert and Alt-
man [29] . Also listed are experimental data obtained by Anderson and his collabora-
tors [43]. It will be seen that the inuence of dissociation of bromine on u0 predicted
by Gilbert and Altman is exaggerated. This conclusion is not surprising if one consid-
ers that these authors estimated the inuence of dissociation from the results obtained
through a thermal theory. Then they applied the reduction factor obtained from a ther-
60 von Krmn Penner

von Krmn Milln
Gilbert Altman (Dissoc. Negl.)
Gilbert Altman (Dissoc. Incl.)
50 Boys Corner (Diss. Negl.)
(1) Flame cone area method
(2) Flame cone angle method
(3) Tube method
40
S (cm/s)
30 (1)
(2)
b
(3)
20
10
0
0.50 0.55 0.60 0.65 0.70
X
3,0
Figure 6.17: The quantity Sb as a function of X3,0 for hydrogen-rich H2 Br2 mixtures.
mal theory to a diffusion theory in which the inuence of dissociation was neglected.
On the other hand, the excellent agreement between their results and the experimental
ones is not too relevant since the former were obtained by applying the method of Boys
and Corner, which gives considerably smaller values for u0 than those obtained from
the application of a more correct method, as will now be veried. With the purpose of
estimating the uncertainty due to the use of a poor method of computation, the value
of u0 for X3,0 = 0.70 was computed with the method of Boys and Corner, as applied
by Gilbert and Altman, and with the method of von Karman and Penner. In both cases
the inuence of the dissociation of bromine was neglected and the physico-chemical
constants of von Karman and Penner were used, so that the difference in numbers was
only due to the difference in method. The following are the results obtained.
Boys-Corner method, as applied by Gilbert and Altman: u0 = 13.8 cm/s,

Von Karman and Penner method: u0 = 23.6 cm/s.
Therefore, if Gilbert and Altman had applied a better method than that of Boys and
Corner, they would have obtained considerably larger values for u0 . The agreement
between their values and those obtained from experiments must be considered to be
largely fortuitous.
1.00
1f 1
0.75
X1
X (dissociation neglected)
2
0.50
X3
0.25
X2
X4
0.0 0.1 0.2 0.3

1
Figure 6.18: Composition and ux proles for X3,0 = 0.65.
Structure of the Flame
The distributions of X1 , X2 , X3 , X4 and 1f 1 corresponding to one of the cases

considered are shown in Fig. 6.18.
In order to show the inuence of X4 on the distribution of X2 , the gure also

indicates the values for X2 obtained when X4 is neglected. Obviously, the inuence
of X4 is very important, especially for values of near 1. On the other hand, this is
the most important inuence of neglecting X4 . It may be seen that small deviations of
X1 and X3 from the linear law have little inuence upon the value of u0 , since X1f
and X3f are different from zero,
References
[1] Mallard, E. and Le Chatelier, H.: Recherches sur la Combustion des Melanges
Gateaux Explosives Ann. Mines, Vol. 8 series 4, 1883, p. 274.
[2] Crussard, L.: Les Deagrations on Regime Permanent dans les Mileux Conduc-
teurs. Compt. Rend., Vol. 158, 1914, pp. 125, 340.
[3] Coward H. F. and Payman W.: Chem. Rev., Vol. 21, 1937, p. 359.
[4] Zeldovich, Y. B. and Semenov, N.: Kinetics of Chemical Reactions in Flames.
NACA Tech. Memo. No. 1084, 1946.
[5] Boys, S. F. and Corner, J.: The Structure of the Reaction Zone in a Flame. J.
Proc. Roy Soc. London, 1949, A1-97,90.
[6] von Karman, Th. and Penner, S. S.: Fundamental Approach to Laminar Flame
Propagation. Selected Combustion Problems, Vol. I, AGARD, 1954, pp. 5-41.
[7] Hirschfelder, J. O., Curtiss, C. F. and Campbell, E.: The Theory of Flames and
Detonations. Fourth Symposium (International) on Combustion, Williams and
Wilkins Co., Baltimore, 1953.
[8] Lewis, B. and von Elbe, G.: On the Theory of Flame Propagation. Phys., Vol.
2, 1934, pp. 537-546.
[9] Evans, N. W.: Current Theoretical Concepts of Steady-State Flame Propaga-
tion. Chem. Rev., Vol. 51, 1952.
[10] von Karman, Th.: The Present Status of the Theory of Laminar Flame Propa-
gation. Sixth Symposium (International) on Combustion, Reinhold Publishing
Corp., New York, 1957.
[11] Tanford, C. and Pease, R.: Equilibrium Atom and Free Radical Concentrations
in Carbon Monoxide Flames and Correlation with Burning Velocities. J. Chem.
Phys., Vol. 15, 1947, pp. 431-433.
[12] van Tiggelen, A.: Chemical Theory of the Speed of Flame Propagation. Bull.
Soc. Chim. Belg., Vol. 58, 1949, p. 259.
[13] Jost, W.: Explosion and Combustion Processes in Gases. McGraw-Hill, New-
York-London, 1946.
[14] Gaydon A. G. and Wolfhard, H. G.: Flames, Their Structure, Radiation and
Temperature. Chapman Hall, London, 1953.
[15] Ladenburg, R. W., Lewis, B., Pease, R. N. and Taylor, H. S.: Physical Measure-
ments in Gas Dynamics and Combustion. Vol. IX of High Speed Aerodynamics
and Jet Propulsion, Princeton University Press, 1954.
[16] Linnet, J. W.: Methods of Measuring Burning Velocities. Fourth Symposium
(International) on Combustion, Williams and Wilkins Co., Baltimore, 1953,
pp. 20-35.
[17] Emmons, H. W., Harr, J. A. and Strong, P.: Harvard University, Contract No.
AT (30-1)-497, 1950.
[18] Adamson, T. C.: On the Theory of One Dimensional Flame Propagation. Jet
Propulsion, January-February 1952.
[19] von Karman Th. and Millan, G.: The Thermal Theory of Constant Pressure
Deagration. Biezenos Anniversary Volume, Delft, Holland, 1953.
[20] Adams E. N., quoted by Henkel, M. J., Spaulding, W. P. and Hirschfelder J. O.:
Theory of Propagation of Flames, Part II Approximate Solutions. Third Sym-
posium (International) on Combustion, Williams and Wilkins Co., Baltimore,
1949, pp. 127-135.
[21] Wilde, K. A.: J. Chem. Phys., Vol. 22, 1954, p. 1788.
[22] Millan G., Sendagorta, J. M. and Da Riva, I.: Comparison of Analytical Meth-
ods for the Calculation of Laminar Flame Velocity. ARDC Contract AF 61
(514), 997, 1957.
[23] Sendagorta, J. M.: Method for the Computation of the Propagation Velocity of
a Plane Laminar Flame. ARDC Contract AF 61 (514), 997, 1957.
[24] Fristrom, R. M.: The Structure of Laminar Flame Fronts. Sixth Symposium
(International) on Combustion, Reinhold Publishing Corp., New York, 1957.
[25] Hirschfelder, J. O., Curtis, J. P. and Campbell, D. E.: The Theory of Flame
Propagation. IV Rep. No. OM-756. Univ. Wisconsin, 1952.
[26] Campbell, E. S.: Theoretical Study of the Hydrogen Bromine Flame. Sixth
Symposium (International) on Combustion, Reinhold Publishing Corp., New
York, 1957.
[27] Giddings, J. C. and Hirschfelder J. O.: Flame Properties and the Kinetics of
Chain-Branching Reactions. Sixth Symposium (International) on Combustion,
Reinhold Publishing Corp., New York, 1957.
[28] Spalding, D. B.: The Theory of Flame Phenomena with a Chain Reaction. Phil.
Trans. Roy. Soc., London 1956.
[29] Gilbert, M. and Altman, D.: The Chemical Steady State in HBr Flames. Sixth
Symposium (International) on Combustion, Reinhold Publishing Corp., New
York, 1957.
[30] Millan, G. and Sanz, S.: Hydrazine Decomposition Flame. ARDC Contract No.
AF 61 (514)-734-O, 1956.
[31] Strong, A. G. and Grosse, A. V.: The Ozone to Oxygen Flame. Sixth Sym-
posium (International) on Combustion, Reinhold Publishing Corp., New York,
1957.
[32] Murray, R. O. and Hall. A. R.: Flame Speeds in Hydrazine Vapour and in
Mixtures of Hydrazine and Ammonia with Oxygen. Trans. Faraday Soc., Vol.
57, 1951, pp. 743-751.
[33] Zeldovich, Y. B.: Theory of Flame Propagation. NACA Tech. Memo. No.
1282, 1951.
[34] Szwarc M. J.: Chem. Phys. Vol. 17, 1949, p. 505.
[35] Hirschfelder, J. O. and Curtiss O. F.: J. Phys Chem. Vol. 57.

[36] Adams, G. K. and Stocks, G. W.: The Combustion of Hydrazine. Fourth Sym-
1953, pp. 239-248.
[37] Spalding, D. B.: The Theory of Flame Phenomenon with a Chain Reaction.
Phil. Trans. Roy. Soc. London, 1956.
[38] Gilbert M. and Altman D.: Hydrazine Decomposition Flame. ORDCIT Project
Contract No. DA-04-495, Ord. 18 Progress Report No. 20-278, 1956.
[39] Millan, G. and Sendagorta J. M.: Hydrazine Decomposition Flame. ARDC
Contract No. AF 61-(514)-997, 1957.
[40] von Karman, Th. and Penner S. S.: The Theory of One-Dimensional Lami-
nar Flame Propagation for Hydrogen-Bromine Mixtures. Part I. Dissociation
Neglected. Tech. Rep. No. 16, Contract No. DA 04-495-0rd. 446, 1956.
[41] von Karman Th. and Millan G.: The Theory of One-Dimensional Laminar
Flame Propagation for H2 -Br2 Mixtures. Part II. Dissociation Included. Tech.
Rep. No. 16, Contract No. DA-04-495-Ord.-446, 1956.
[42] Mileson, D. F.: The Thermal Theory of Laminar Flame Propagation for Hydrogen-
Bromine Mixtures. Tech. Rep. No. 6, Contract No. DA 04-495-Ord.-446, 1954.
[43] Anderson, R. C.: Hydrogen-Halogen Flames. AGARD Combustion Panel Meet-
ing, Oslo, 1956.
Chapter 7
Turbulent ames
7.1 Introduction
There is experimental evidence that turbulence increases the propagation velocity of

the ame through a combustible mixture. This is a very important fact in technical
applications since it allows a considerable reduction of the space and time required to
burn a given mass. The inuence of turbulence over combustion was rst recognized
by Mallard and Le Chatelier in 1883 [1]. However the attempts made to determine the
causes of this inuence as well as to estimate quantitatively its value are quite recent.
The basic problem lies in determining the propagation velocity of the ame through a
combustible mixture in turbulent motion knowing the characteristics of the turbulence
and the state and composition of the mixture.
From the experimental stand-point, several techniques are available for the de-
termination of the ame velocity. However the basis for such techniques are not as
solid as those applied to the case of laminar ames and the measurements taken are
not as numerous or systematic. Among them, the technique more commonly used
consists in photographing a ame, obtained for example in a bunsen burner, then mea-
suring the area of the combustion front and dividing it by the ow rate of the burner
as is done for the case of laminar ames. One of the difculties of such techniques
is the fact that the combustion front of a turbulence ame is not well dened since
the long exposure photographs show a thick luminous zone (see Fig. 7.1a) whereas
a short exposure Schlieren photograph reveals a very irregular and wrinkled struc-
ture (see Fig. 7.1b). These circumstances impose the introduction of an arbitrariness
in estimating the ame area. Damkohler [2], for instance, adopted as surface of the
turbulent combustion front the one that limits internally the luminous zone. Later on,
207
208 CHAPTER 7. TURBULENT FLAMES
(a) Time exposure photograph (b) Instantaneous Schlieren photograph

Figure 7.1: Stoichiometric natural gas-air ame, Re = 25 000, burner tube diameter =
5.08 cm (by courtesy of Bureau of Mines).
preference was given to the use of the mean surface between those internally and ex-
ternally limiting the luminous zone [3] or else to the surface where luminosity reaches
its maximum intensity [4].
Additional difculty lies in the fact that the paths of the gas particles are not
known. Therefore it is risky to identify the fraction of unburnt mixture corresponding
to each element of the ame front. Such difculty has often been eluded by measuring
the mean velocity obtained when the total ow rate of the mixture is divided by the
area of the ame front. The propagation velocity of turbulent ames has been mea-
sured with these or similar techniques by Damkohler [2] , Bollinger and Williams [3],
Scurlock [4], Williams, Hottel and Scurlock [5], Karlovitz, Denniston and Wells [6],
Wohl, Shore, von Rosemberg and Weil [7], Leason [8], Bowditch [9], Wohl and Shore
[10], Mickelsein and Ernstein [11], etc.
Figure 7.2 taken from Ref. [3] shows, as an example, the results of the mea-
surements performed by Bollinger and Williams in mixtures of several hydrocarbons
and air. For each case the mixture giving the maximum velocity was used. This gure
also shows the laminar velocities corresponding to the same mixtures. It is seen that
the effect is maximum in the acetylene ame where the value of the laminar ame ve-
300
250
TURBULENT FLAME SPEED, cm/s
200
150
100
50
0 3
0 10 20 30 40 x10
REYNOLDS NUMBER OF PIPE FLOW
Figure 7.2: Variation of ame speed with Reynolds number of ow.
locity is half of that for a Reynolds number 3 104 . Other measurements have given
turbulent velocities considerably higher than these observed here.
From the theoretical stand-point several attempts have been made to explain the
activation of combustion due to turbulence. The rst attempt was made by Damkohler
[2] who pointed-out two causes for the action of turbulence. One is the increase in the
transport coefcients (conductivity and diffusion) due to turbulent diffusivity. The sec-
ond cause is the distortion of the laminar ame front due to the turbulent oscillations
of the velocity which would increase, considerably, the effective surface of the com-
bustion front. The importance of either one of these two factors would depend in each
case on the relation between the scale of turbulence and the thickness of the laminar
ame. In technical applications this ratio is generally very large, hence the main cause
for the activation of combustion would be in this case the distortion of the ame front.
Figure 7.1b seems to conrm this point of view. The major part of the later works on
the subject have attempted to estimate the importance of either one of these causes,
preserving Damkohlers model. In 1952 von Karman and Marble [12] proposed a dif-
ferent model which considered the turbulent ame as a zone whose structure should
be dened by mean values of the characteristic variables (temperature, reaction rate,
etc.) to which were superimposed static oscillations of turbulent nature. Later, this
model was adopted and developed by Sommereld and his collaborators [13], who
obtained some experimental evidence (not sufcient) that the approximation is cor-
rect. The comparison between theoretical and experimental results is not conclusive
enough in any case to establish either one of the proposed theories, which are actually
in a preliminary phase.
Consequently the present study will only be a brief exposure of the basic prin-
ciples and fundamental results of these theories.
7.2 Turbulent combustion theories
So far, the only type of turbulence whose inuence on combustion has been studied is
the isotropic. This turbulence is characterized by two magnitudes: its scale l and its
intensity v 0 .1 Damkohler studies the inuence of turbulence on the ame by compar-
ing l and v 0 with the corresponding magnitudes of the laminar combustion wave of the
mixture; these being thickness dl of the ame and the laminar propagation velocity
ul . When performing this comparison the following cases arise.
1) The turbulence scale is small when compared to the thickness of the laminar
ame
l
l (7.1)
dl
In this case the action of turbulence is reduced to an activation of the transport coef-
cients conductivity and diffusion at the ame). Let

x= (7.2)
cp
and
lv 0 , (7.3)
1 For an exposure of the principles of turbulence, see, i.e., H.L. Dryden: A review of the Statistical Theory
of Turbulence. Quart. Applied Math., Vol. 1, 1943, pp. 7-42.

7.2. TURBULENT COMBUSTION THEORIES 211
be the laminar and turbulent thermal diffusivities, respectively. The turbulent propa-
gation velocity ut of the ame is obtained from ul when x is substituted by . Since

ul is proportional to x, it results for ut

ut
= (7.4)
ul x
Shelkin [14] also considers the inuence of laminar diffusivity on the turbulent ame.
In this case, must be substituted in (7.3) by x + , thus resulting

ut
= 1+ (7.5)
ul x
This expression has the advantage over (7.4) that when turbulence approaches zero,
ut approaches ul .
Case 1) very seldom arises since under normal conditions the thickness of the
laminar ame is only of some tenths of a millimeter.
2) Scale of turbulence and ame thickness are of the same order of magnitude
l
' 1. (7.6)
dl
Damkohler does not consider this case. However Shelkin has proven that the
action of turbulence does not reduce then to an activation of the transport coefcients
but it also inuences the combustion time of the mixture, which not only depends on
the reaction velocity of the species but on the rapidity of the turbulent mixing. To date,
no formula is available for this case.
3) Scale of turbulence is large when compared to ame thickness

l
1. (7.7)
dl
In this case combustion takes place through a laminar front but turbulence wrin-
kles the laminar ame front thus increasing its surface. Thereby, the effective com-
bustion surface of the mixture is increased. Two possibilities should be considered
depending on the intensity of the turbulence.
3.a) The intensity of turbulence is very large with respect to the laminar prop-
agation velocity of the ame
v0
1. (7.8)
ul
This case was not analyzed by Damkohler. The following consideration are
owned to Shelkin. When condition (7.8) takes place the structure of the combustion
B A
BURNED UNBURNED
GASES GASES
B A
Figure 7.3: Structure of the turbulent ame in the case l dl and v 0 ul , according to
Shelkin.
zone shows islands of unburnt gas which have been dragged by the strong turbulent os-
cillations towards the region of combustion products without time to burn, see Fig. 7.3.
In this case, the zone between AA0 and BB 0 can be considered as a combustion wave
which advances with a propagation velocity ut . The propagation velocity of a com-
bustion wave is determined by an equilibrium between transport processes, which are
characterized by combustion time . Independently from the mechanism which deter-
mines the values for and , a dimensionless analysis shows that ut must be of the
form

ut . (7.9)

For the case shown in Fig. 7.3, is clearly the turbulent diffusivity. Combustion
velocity is determined by the mixing rapidity and therefore is proportional to the
time of turbulent mixing
l
0. (7.10)
v
When the expressions for and given by (7.2) and (7.10) are taken into (7.9)
ut v 0 , (7.11)
Hence, the propagation velocity is proportional to the intensity and independent from
the scale of turbulence. Furthermore, ut results to be independent from laminar ve-
locity ul . Such conclusion contradicts experimental evidence.
3.b) The intensity of turbulence is of the same order of magnitude or smaller

than the propagation velocity of the laminar ame
v0
. 1. (7.12)
ul
In this case the islands disappear and the action of turbulence reduces to deforming
l
BURNED UNBURNED

GASES GASES
Figure 7.4: Structure of the turbulent ame in the case l dl and v 0 . ul , according to
Shelkin.
the laminar ame front thus increasing its surface as shown in Fig. 7.4. Let l be the
effective surface of combustion and the surface of access to the ame of the unburnt
gases. Velocity ut is given in this case by expression
ut l
= . (7.13)
ul
The problem lies then in estimating the value for l , for which several models have
been proposed.
Shelkin assumes that l is formed by a system of cones whose base is propor-

tional to the square of the scale of turbulence l2 and whose height is proportional to
distance v 0 l/ul travelled by a mass of gas with a diameter l due to turbulent oscilla-
tions during the time l/ul needed by the laminar ame to cross this mass. Here, l /
is the ratio of the lateral to the base area of the cones, and from (7.13), it results
( 0 )2
ut v
= 1+k , (7.14)
ul ul
where k is a numerical coefcient of the order of magnitude unity. Therefore, ut is
also independent from the scale of turbulence. When turbulence is weak, Eq. (7.14)
shows that its effect is of the second order, whilst for very intense turbulence Eq. (7.14)
reduces to (7.11). Through a different analysis M. Tucker [15] obtains the following
expression for the case of weak turbulence
( 0 )2
ut v
= 1 + k() , (7.15)
ul ul
which is valid if v 0 /ul l. In this formula, k() is a coefcient depending on ratio
of temperature Tf of the burnt gases to temperature T0 of unburnt gases. It is seen
that when v 0 /ul l. Eqs. (7.14) and (7.15) agree.

Karlovitz [16] reasons as follows. Let X = X 2 be the root mean square
displacement of a particle due to turbulent oscillations. X is a increasing function of
time counted from the point at which measurements were initiated. Let t = l/ul be
the time taken by the laminar ame to cross a turbulent vortex. The total path travelled
by the ame in this time is X + l and when this expression is divided by t one obtains
for ut
ul
ut = X + ul , (7.16)
l
That is to say
ut X
=1+ . (7.17)
ul l
The value for X is given by Taylors formula
t
dX 2 02
= 2v Rt dt, (7.18)
dt 0
where Rt is the correlation coefcient between the velocities of the same particle at
two different instants. Karlovitz uses from Rt the following expression
tv 0

Rt = e l0 , (7.19)
where
l0 = v 0 Rt dt (7.20)
0
is the Lagrangian scale of turbulence. By substituting (7.19) into (7.18) and with
t = l/v 0 it results for X
( [ ( )])
v0 aul 1 v0
X = l 2a 1 0 1 exp , (7.21)
ul v a ul
Here a is the ratio from the Lagrangian to the Eulerian scales of turbulence. Karlovitz
assigns to a a value 1 while Scurlock and Grover [17] and Wohl [18] choose 1/2.
Taking (7.21) into (7.17) one obtains for ut
( [ ( )])
ut v0 aul 1 v0
= 1 + 2a 1 0 1 exp . (7.22)
ul ul v a ul
For very intense turbulence (7.22) reduces to

ut v0
' 1 + 2a , (7.23)
ul ul
which is valid provided that
v0
1.
ul
Eq. (7.23) is in contradiction with (7.12) and for weak turbulence is reduces to
ut v0
'1+ , (7.24)
ul ul
valid for
v0
1,
ul
which also contradicts (7.16).
Wohl has shown that (7.22) can also be deduced from (7.13) through purely
geometric considerations by assuming that the distortion of the ame front consists in
the formation of a system of prisms instead of the cones proposed by Shelkin.
Scurlock and Graver [17] calculate l by assuming with Shelkin that the lam-
inar surface is formed by a set of cones whose base is proportional to the square l2
of the scale of turbulence and whose height is proportional to the root mean square
displacement X of the turbulent oscillations suffered by a ame element. Therefore,
as in Eq. (7.14), ut is given by expression

ut X 2
= 1+k 2 . (7.25)
ul l
The problem lies in computing X. Its magnitude, according to Scurlock, is

governed by three different mechanisms:
1) Turbulent diffusivity tends to increase indenitely the value of X as time elapses.

Figure 7.5 shows two consecutive positions of the ame front in Scurlocks
model after the instant at which the ame was plane.
Initial flat flame Flame after passage Flame after passage

of short time of longer time
Figure 7.5: Schematic diagram showing wrinkling with passage of time of an initially at
ame element exposed to turbulence.
2) Laminar combustion tends to absorb oscillations reducing the value of X. Karlo-

vitz and his collaborators [19] were the rst to acknowledge this fact which is
represented in Fig. 7.6
3) The ame generates turbulence2 and this tends to increase the value of X.
2 See 4.
ARBITRARY WAVY FLAME FRONT
BURNED GAS FRESH GAS
FLAME FRONT AFTER t TIME

Figure 7.6: The effect of laminar ame propagation on the evolution of a turbulent ame
front.
Since the locus of an element of the ame at two different instants are related
to two different particles of the mixture, when calculating the inuence of turbulent
diffusivity on X, a combined time-space coefcient of correlation Rtx should be used
substituting coefcient time-correlation Rt in Taylors formula [17].
The inuence of the laminar propagation of the ame in the value of X may
be obtained through purely geometric considerations. Finally, the inuence of the
turbulence generated by the ame is taken into consideration by substituting into the
expression of X, the intensity v 0 of turbulence of the unburnt gases ow by the resul-
tant of it, plus the intensity of the turbulence originated by the ame.
The combination of these three effects gives a differential equation which in

turn supplies the law of variation for X as a function of time t during which the
particle has been exposed to turbulent oscillations. This differential equation replaces
Taylors equation (7.18) for this case. Through integration of this equation the value
for X is obtained, and when taken into Eq. (7.22) it gives a formula for the propagation
velocity of the turbulent ame.
Now we must determine the time t that each element of the ame has been
exposed to the action of turbulence. Scurlock considers only ames inclined respect
to the ow such as the one obtained with a ame-holder or from the ring of a bunsen
burner. He identies t with the time taken by a gas particle to cross the ame front
from the holders section to the point under consideration. Let vt by the component
tangential to the ame front of the velocity of the unburnt gases and s the distance to
the ame-holder measured along the ame front. The following expression is obtained
for t
s
ds
t= . (7.26)
0 vt
The selection of t, which is arbitrary to a certain extent, is justied by the

theoretical and experimental studies carried out by Markstein [20] on the propagation
of disturbance along inclined ame fronts. These studies show that such disturbances
propagate with the tangential velocity of the ow as they become amplied.
BURNED BURNED BURNED

GASES GASES GASES
MEAN POSITION MEAN POSITION MEAN POSITION

OF FLAME OF FLAME OF FLAME

INSTANTANEOUS INSTANTANEOUS INSTANTANEOUS
FLAME FRONT FLAME FRONT FLAME FRONT
UNBURNED UNBURNED UNBURNED

GASES GASES GASES
U U U
BURNER BURNER BURNER
RIM RIM RIM
Figure 7.7: Cross section of stabilized unconned Bunsen ames constructed on basis of
theory to demonstrate under typical conditions the predicted individual effects
of eddy diffusion, ame propagation and ame generated turbulence.
Figure 7.7 taken from the work by Scurlock and Grover, shows the inuence of
the above mentioned facts for the case of an open ame stabilized in a bunsen burner.
The following values were adopted for computations: diameter of the burner = 5 cm,
velocity on the gases in the burner= 200 cm/s, u1 = 10 cm/s, v 0 = 10 cm/s, intensity
of turbulence 5%, scale of turbulence= 0.25 cm, and width of the ame= 2X.
7.3 Turbulence generated by the ame
Markstein [21] has shown that the disturbances originated at a certain point of an in-
cline ame propagate and amplify along the same and eventually generate turbulence.
In their studies of turbulent ames stabilized in tubes, Willians, Hottel and Scurlock
[5] recognized the existence of turbulence originated by the ame. They considered
this turbulence as a consequence of the strong gradients of velocity which originate
through the ame in this case.3 Scurlock and Grove have given a formula for the
maximum intensity v 0 of the possible turbulence. Such formula was deduced from el-
ementary considerations on mass and momentum conservation across the ame before
and after the mixing of burnt gases. This formula is
( )
0 Km u 2 u
vm = (vu u2l ) 1 . (7.27)
3 b b
p p
Here, Km = 1, where is the relation between pressure jumps across
p0 p0
the ames before and after the turbulent mixing, vu is the velocity of the unburnt gases
normal to the mean ame front and u /b is the ratio between densities of unburnt and
burnt gases. Formula (7.27) gives a maximum limit to the turbulence that could be
originated in the ame but does not allow the computation of the possible turbulence
for each case. This arises the problem of introducing an additional undetermined
element in the theory which adds difculties to the comparison between theoretical
0
and experimental results. In practice vm can be considerably larger than the turbulence
of the incident stream.
Karlovitz [16] when comparing the values for ut predicted by his theory with
those obtained from experimenting with ames stabilized in bunsen burner veried
that experimental velocities were much larger than those calculated and he considered
this discrepancy to be due to the inuence of the turbulence originated by the ame.
Karlovitz explains the production of turbulence as follows: through the laminar front
u
an increase in velocity takes place of the order of magnitude of 1. In a laminar
b
ame this increase has a constant direction but in a turbulent ame it oscillates since
it must be normal to the temporary position of the ame front at all instants which is
variable. The statistic oscillations of this increase are the source of the turbulence.
Through a not too legitimate calculation Karlovitz deduces the following ex-
pression for the maximum intensity of turbulence produced by the ame
( )
0 1 u
vm = 1 ul . (7.28)
3 b
3 See chapter 10.
7.4. COMPARISON WITH EXPERIMENTAL RESULTS 219
It is also impossible here to estimate which fraction of the intensity given by (7.28)
will actually turn into turbulence in each case.
7.4 Comparison with experimental results

As aforesaid the available experimental evidence is not sufcient to establish either
one of the proposed theories in a denite way precluding the others. We owe the most
complete set of experiments to Bollinger and Williams [2] and to Wohl and Shores [7]
referred to herein. Some of the results obtained show laws of variation of ut which
approach those predicted by Karlovitz and by Scurlock and Grover. However, their
effects appear which are either not predicted by theory or in contradiction with it. An
example of effects not predicted is the inuence of the mixtures composition, since
it is veried that the effect of turbulence is not the same for rich mixtures than for
poor ones, even when operating under identical conditions and with the same laminar
propagation velocity. This is especially true for certain combustibles (like butane) and
this fact cannot be explained with any of the proposed theories. Wohl believes this
effect to be due to the different laminar stability in poor and rich mixtures. An ex-
ample of effects appearing in contradiction with theory is the inuence of the scale of
turbulence, which after Scurlocks predictions, should be considerable and which ac-
cording to experimental results is very small. Consequently it is necessary to increase
experimental evidence before nal decision may be reached.
References
[1] Mallard, F. and Le Chatelier, H.: Ann. de Mines. Vol. 8, Sev. 4, 1884, p. 274.
[2] Damkohler, G.: The Effect of Turbulence on the Flame Velocity in Gas Mixtures.
NACA Tech. Mem. No. 1112, 1947.
[3] Bollinger L. M. and Williams, D. T.: Effect of Reynolds Number in the Turbulent-
Flow Range on Flame Speeds of Bunsen-Burner Flames. NACA Tech. Note
No. 1707, Sept. 1948.
[4] Scurlock, A. C.: Flame Stabilization and Propagation in High Velocity Gas
Streams. Meteor Report No. 19, July 1948.
[5] Williams, G. C., Hottel, H. C. and Scurlock, A. C.: Flame Stabilization and
Propagation in High Velocity Gas Streams. Third Symposium (International)
on Combustion, Williams and Wilkins Co., Baltimore, 1949, pp. 21-40.
[6] Karlovitz, B. Denniston, D. W. and Wells, F. E.: Investigation of Turbulent
Flames. Journal of Chemical Physics, Vol. 19, No. 5, May 1951.
[7] Wohl, K., Shore, L., von Rosemberg, H. and Weil, C. M.: The Burning Ve-
locity of Turbulent Flames. Fourth Symposium (International) on Combustion,
Williams and Wilkins Co., Baltimore, 1953.
[8] Leason, D. B.: Turbulence and Flame Propagation in Premixed Gases. Fuel,
Vol. XXX, No. 10, Oct. 1951.
[9] Bowditch, F.: Some Effects of Turbulence on Combustion. Fourth Symposium
pp. 620-635.
[10] Wohl, K. and Shore, L.: Experiments with Butane-Air and Methane-Air Flames.
Industrial and Engineering Chemistry, April 1955.
[11] Mickelsein, W. R. and Ernstein, N. E.: Propagation of a Free Flame in a Tur-
bulent Gas Stream. NACA Tech. Note No. 3456, 1955.
[12] von Karman, Th. and Marble, F.: Combustion in Turbulent Flames. Fourth
Symposium (International) on Combustion, Williams and Wilkins Co., Balti-
more, 1953, p. 923.
[13] Sommereld, M., Reiter, S. H., Kebely, V. and Mascolo, R.W.: The Structure
and Propagation Mechanism of Turbulent Flames in High Speed Flow. Jet
Propulsion, August 1955.
[14] Shelkin, F. I.: On Combustion in a Turbulent Flow. NACA Tech. Mem. No.
1110, Febr. 1947.
[15] Tucker, E.: Interaction of a Free Flame Front with a Turbulence Field. NACA
Tech. Note No. 3407, 1955.
[16] Karlovitz, B.: A Turbulent Flame Theory Derived From Experiments. Selected
Combustion Problems, Vol.I, AGARD, 1954, pp. 248-262.
[17] Scurlock, A. C. and Grover, J. F.: Experimental Studies on Turbulent Flames.
Selected Combustion Problems, Vol.I, AGARD, 1954, pp. 215-247.
[18] Wohl, K.: Burning Velocity of Unconned Turbulent Flames Industrial Engi-
neering Chemistry, April 1955, p. 825.
[19] Karlovitz, B.: Open Turbulent Flames. Fourth Symposium (International) on
Combustion, Williams and Wilkins Co., Baltimore, 1953, pp. 60-67.
[20] Markstein, G. H.: Discussion on Turbulent Flames. Selected Combustion Prob-
lems, Vol.I, AGARD, 1954, pp. 263-265.
[21] Markstein, G. H.: Interaction of Flow Propagation and Flame Disturbances.
Third Symposium (International) on Combustion, Williams and Wilkins Co.,
Baltimore, 1949, pp. 162-167.
Chapter 8
Ignition, ammability and

quenching
8.1 Introduction
In the preceding chapter we have analyzed the properties of a plane wave propagat-
ing in a stationary regime, through an indenite combustible mixture whose state and
composition are uniform. Although the problem is far from being completely solved,
the essential characteristic of those waves are well understood, and a precise formula-
tion of the same is available as well as several methods to solve the resulting equations,
including some practical solutions, which may be considered as fairly acceptable when
compared to experimental results.
However, it must be said that the stage of knowledge is not as favorable for
other fundamental questions relative to ames such as: the problem of ignition of
a combustible mixture; the possibility of a ame to propagate through a mixture of
given state and composition; and the inuence of the vicinity of the walls on the
characteristics of the ame. Even when a great amount of attention has been applied
during recent years to the study of these problems, the progress achieved, specially
from the theoretical stand-point, has been considerably less than for the case of an
indenite plane wave, to the extreme that the causes of some of the phenomena are
still unknown as well as their governing laws.
The present chapter contains a brief description of each of the above mentioned
problems, accompanied by a selected bibliography which will enable a more detailed
study on each one.
221
222 CHAPTER 8. IGNITION, FLAMMABILITY AND QUENCHING
8.2 Ignition
Considering a combustible mixture capable of propagating a ame, in order for this

to happen it is necessary to supply it with a certain amount of energy, located within a
small volume and reduced interval of time, to initiate the wave. The study of this pro-
cess, both theoretically and experimentally, has been the subject of numerous reports.
Among the various sources of energy that may be utilized, the most adequate is the
electrical spark because it is capable of steering great amounts of energy in reduced
spaces and times. For this reason it is the most widely used and has been the subject
of more systematic and complete experimental studies, of which a description may be
found in Refs. [1] and [2].
The fundamental problem of ignition lies on determining the minimum energy

required to ignite a mixture of given composition, at known pressure and temperature.
The results of experiments have pointed out that to each mixture it corresponds a well
dened minimum ignition energy, which increases very rapidly with the decrease in
pressure. Such energy results to be independent from the distance between electrodes,
provided this distance exceeds a given minimum value under which it grows very
rapidly and nally the ame cannot propagate.
Several theories have been attempted for the calculation of this minimum en-
ergy with fair success. The present state of knowledge on this problem may be con-
sidered as comparable to the situation existing 20 years ago respect to theories on
ames.
All the theories developed are based on the following model. Let us consider
a small volume V of gas to which energy H is instantaneously communicated, rising
its temperature from T0 to Tf . Starting from this instant, the gas contained in V
tends to cool, by thermal conductivity, heating the gas layers surrounding V . On the
other hand, the chemical reaction which produces in the heated mass releases heat,
which tends to compensate the cooling effect. From the balance between this two
phenomena, it will depend that the ame may progress towards a wave of the type
described in Chap. 6, or, to the contrary, that it extinguishes.
The available theories differ in the development of this idea and on the condi-
tions applied to determine the minimum volume V and the necessary energy H. For
instance, Lewis and von Elbe [3] assume that volume V is the one corresponding to
the quenching1 distance and that the energy is determined by the excess enthalpy of
the ame. The validity of this concept does not appear clearly justied, although the
1 See 4.
8.2. IGNITION 223
comparison between the minimum values of the calculated energy and those measured
experimentally show a surprising agreement [3].
Fenn [4] performs an elementary computation which is actually more of a di-

mensionless analysis of the problem. He reasons as follows: be r the radius of volume
V and let us assume that combustion takes place through a second-order reaction of
the form
w = A2 Y (1 Y )eE/RT . (8.1)
Then, the heat released per unit time in volume V due to the combustion of the mixture
will be
4
Q = r3 qA2 Y (1 Y )eE/RTf , (8.2)
3
where Tf is the temperature of gases, and q the heat of reaction.
In turn, the heat lost by conductivity through the surface limiting V , will be
4r2 (Tf T0 )
Q0 = , (8.3)
cr
where cr is the thickness of the combustion wave, which separates hot from cold
gases.
The condition of minimum ignition energy for propagation, will be
Q = Q0 , (8.4)
since if it is Q < Q0 the mass cools and the ame extinguishes.
By taking (8.2) and (8.3) into (8.4) we obtain the following expression for the
radius of V
3(Tf T0 )
r= . (8.5)
cqA2 Y (1 Y )eE/RTf
Finally, the minimum energy H is the one needed to rise the mass contained in V from
temperature T0 to Tf which is
4 3
H= r cp (Tf T0 ), (8.6)
3
where for r one must use expression (8.5).
Recently Swott [5] has extended this theory to the study of the problem of
the ignition of owing mixtures including as well the effects of turbulence. Other
studies on the matter, specially directed to the ignition in combustion chambers, with
extensive bibliography, will be found in Ref. [6].
The preceding theories are of the thermal type, since they ignore the inuence
of diffusion, which could be important, specially the diffusion of radicals.
A correct stating of the problem would require rst an adequate formulation of

the same, in an analogous way to the one used in Chap. 6 for the stationary wave. In
the second place it would be necessary to perform an analysis of the type of solutions
for the system of equations obtained under the set of initial and boundary conditions
corresponding to the model previously described.
If we consider the more simple case of an indenite plane wave corresponding

to a one-dimensional problem, in which case the initial energy must be stored between
two parallel layers, it may be easily veried that the system of equations of Chap. 6,
corresponding to a stationary wave, keeping the same notation, must be substituted by
the following:
(v)
+ = 0. (8.7)
t x
b) Energy equation.
( ) ( )
T T T
cp +v = + qw. (8.8)
t x x x
c) Diffusion equation.
( ) ( )
Y Y Y
+v = D + w. (8.9)
t x x x
Thus we obtain a system of three equations for the three unknowns T , v and Y .
Since the process takes place at constant pressure, is already determined, as function
of T .
With reference to initial and boundary conditions corresponding to the model

under consideration, if 2d is the width of the heated slab and the origin of coordinates
is xed at the central point of the slab, we will have:
a) Initial conditions (t = 0).
0 < x < d : T = Tf ,
d < x < : T = T0 , (8.10)
0<x<: v = Y = 0.
b) Boundary conditions (by symmetry).

T Y
x=0: =v= = 0.
x x
Furthermore an additional condition must be introduced expressing that w must
be zero for T = T0 , as it was done for the case of a stationary wave.
8.3. FLAMMABILITY LIMITS 225
The preceding system of equations allows an easy simplication, by using the

stream function , frequently applied to the problem of Fluid Mechanics.
The said function is dened by the following conditions

= , v = , (8.11)
x t
through which Eq. (8.7) is identically satised. When taking Eq. (8.11) into Eqs. (8.8)
and (8.9), this system reduces to the following
( )
T T q w
= + , (8.12)
t cp cp
( 2 )
Y D Y w
= + . (8.13)
t cp
Thus the variable v has been eliminated.
The integration of the above system has not yet been performed. However,
Spalding [7] has obtained graphical solutions of Eq. (8.12) disregarding diffusion.
These solutions show the existence of a critical value dcr for d under which the wave
extinguishes, while for d > dcr it propagates indenitely.
There are some experimental observations on the evolution followed by an

ignited mass before the ame establishes [8].
8.3 Flammability limits
The theoretical studies performed so far have shown that the combustible mixture is
always capable of maintaining a ame. Experimentally, however, it has been veried
that it is not so, since there are numerous mixtures which, in practise, are not able to
propagate a ame, even when theory predicts otherwise.
In fact, experiments disclose that when analyzing the behavior of a combustible

mixture under pressure and temperature constant, but changing its composition, for in-
stance a mixture of fuel and air, the ame velocity is maximum for a denite value
of the composition which is normally very close to the stoichiometric one, and it de-
creases for mixtures either rich or loan, up to values known as inammability limits of
the mixture, beyond which it cannot propagate a ame. Moreover the ame velocity
does not vanish at the inammability limits but it generally reaches values of a few
centimeters per second. Inammability limits depend also on pressure and they are
difcult to determine experimentally because it is necessary to operate under marginal
conditions, on which the experimental techniques might have an inuence. Tables
and graphics may be found in Ref. [1], giving the inammability limits of several
mixtures of hydrocarbons with air, under different pressures. Ref. [3] supplies an
abridged review of the problem. The Proceedings of the Fourth International Sym-
posium on Combustion include, as well, several papers on the subject, Coward and
Jones, [9], reviewed the state of knowledge up to 1952, in a report including an ex-
tensive experimental and bibliographic material. Two recent, very interesting, reviews
are those by Egorton [10] and Linnot-Simpson [11].
At present, the situation of the problem is such, that the causes for the ex-
istence of inammability limits are still unknown and, even more, it is questioned
whether they actually exit as quantities governed only by the state and composition
of the mixture or, to the contrary, they depend upon the characteristics of the experi-
mental device used. Lewis and von Elbe [3] suppose that the existence of such limits
may be due to the internal stability of the combustion wave, which then would be un-
stable beyond the limits. Several attempts to study the problem from this stand-point
have been made, [12] and [13], in addition to the one by Lewis and von Elbe. Of
all, the most satisfactory is own to Werner and Rosen [14], who analyze the internal
stability of the wave with respect to small disturbances of temperature by means of a
linearization of the system of Eqs. (8.7), (8.8) and (8.9) around their stationary solu-
tion. These authors solved the resulting system after introducing several simplifying
assumptions regarding the disturbances of the composition of the mixture and of the
mass ux. They reach the conclusion that the stability of the wave depends on the
behavior of the disturbances of Y with respect to those of T . Although this study
must be considered as uncompleted it suggests the possibility for some waves to be
internally unstable, which could explain the existence of inammability limits. An
application of this theory to the ozone-oxygen ame is given in Ref. [15], the result
of which shows that the ame is stable for ozone rich mixtures and unstable for loan
ones. However, the experiments carried out by Streng and Grosse [16] disclose that
the ame is stable even for loan mixtures.
Recently Spalding [17] has analyzed the possible inuence of heat losses of
the ame on its behavior. He reaches the conclusion that there exist two different
ame velocities, and that the smaller is unstable. When heat losses augment the ve-
locities approach one another and nally coincide. Spalding identies the coincidence
of velocities with the inammability limits. Although this work does represent an im-
portant contribution, yet an unpublished work, being carried out at the Combustion
Laboratory of the INTA, reveals that his conclusions are not free from criticism.
8.4. QUENCHING 227
8.4 Quenching
So far studies have dealt with ames propagating through unlimited mixtures, not
taking into account the inuence of the proximity of walls,
A set of phenomena exist, of great importance, relative to the behavior of

ames close to walls which have been the subject of such attention in recent years.
Among those phenomena are quenching, blow-off and ash-back.
Experiments prove that when the diameter of a tube full of a combustible gas
is sufciently small, it is impossible for a ame to propagate through it. This phe-
nomenon, which has been recognized for a long time and which has been applied to
the prevention of explosions, is known as quenching.
The way in which the walls of the tube act to prevent ame propagation is
not well understood at present, The most direct explanation is to attach the effect to
the cooling of the ame due to the proximity of the walls. However, it is though
that chemical action of the wall when acting as chain breaker may also be signicant.
Analogously to what happened for some time with ame theory, if one considers either
cooling or diffusion as the only acting effect, a thermal or diffusive theory will
be obtained. Examples of thermal theories are those by Lewis-von Elbe [1] and by
Karman-Millan [18], and on diffusion theories the one by Simon-Bellos [19].
The difference between this case and the unlimited ame theory lies on the fact
that both the formulation of the equation and boundary conditions and their solution
are far more difcult. Hence, the solutions may only be attempted through drastic
simplications.
A great number of measurements have been carried out with the purpose of
determining the quenching distance in ducts of different cross-sections and the inu-
ence on it, of the mixture. For an additional study of this question as well as on the
phenomena of blow-off and ash-back we refer the reader to Refs. [20] through [23].
In particular references [20] and [23] include an extensive bibliography.
References
Academic Press, New York, 1951.
[2] Lewis, B, Pease, R. N. and Taylor, H,S.: Combustion Processes. Vol. II of High
Speed Aerodynamics and Jet Propulsion, Princeton University Press, 1956.
[3] Lewis, B. and von Elbe, G.: Fundamental Principles of Flammability and Igni-
tion. Selected Combustion Problems, Vol. II, AGARD, 1956, pp. 63-72.
[4] Fenn, J. B.: Lean Inammability Limit and Minimum Spark Ignition Energy.
Industrial and Engineering Chemistry. Vol. Dec., 1951, pp. 2865-68.
[5] Swott, C. C.: Spark Ignition of Flowing Gases. IV. Theory of Ignition in Nontur-
bulent and Turbulent Flow Using Long Duration Discharges. NACA Research
Memorandum No. R M E54F29a, 1954.
[6] Wigs. L. D.: The Ignition of Flowing Gases. Selected Combustion Problems,
Vol. II, AGARD, 1956, pp. 73-82.
[7] Spalding, D. B.: Some Fundamentals of Combustion. Butterworths Scientic
Publications, 1955.
[8] Olsen, E. L., Gayhart, E. L. and Edmoneon, R. B.: Propagation of Incipient
SparkIgnited Flames in HydrogenAir and PropaneAir Mixtures. Fourth
more, 1953, pp. 144-148.
[9] Coward, H. F. and Jons, G. W.: Limits of Flammability of Gases and Vapors.
U.S. Bureau of Mines. Bull., 503, 1952.
[10] Egerton, A. C.: Limits of Inammability. Fourth Symposium (International) on
Combustion, Williams and Wilkins Co., Baltimore, 1953, pp. 4-13.
[11] Linnet, J. W. and Simpson, O. J. S.: Limits of Inammability. Sixth Symposium
[12] Richardson, J. M.: The Existence and Stability of Simple OneDimensional,
SteadyState Combustion Waves. Fourth Symposium (International) on Com-
bustion, Williams and Wilkins Co., Baltimore, 1953, pp. 182-189.
[13] Layzer, D.: Journal of Chemical Physics, Vol. 22, 1954, p. 222.
[14] Wehner, J. F. and Rosen, J. B.: Temperature Stability of the Laminar Combus-
tion Wave. Combustion and Flame, Sept. 1957, pp. 339-345 .
[15] Rosen, J. B.: Stability of the Ozone Flame Propagation. Sixth Symposium
[16] Streng, A. G. and Grosse, A. V.: The Ozone to Oxygen Flame. Sixth Symposium
[17] Spalding, D. B.: A Theory of Inammability Units and FlameQuenching .
Proc. Roy, Soc. London, Ser. A, Vol. 240, No. 1220, 1957, pp. 83-100.
[18] von Karman, Th. and Millan, G.: Thermal Theory of Laminar Flame Front near
a Cold Wall. Fourth Symposium (International) on Combustion, Williams and
8.4. QUENCHING 229
[19] Simon, D. M. and Belles, F. E.: An Active Particle Diffusion Theory of Flame
Quenching for Laminar Flames. NACA Research Memorandum No. RM
E51L18, 1952.
[20] Wohl, K.: Quenching FlashBack, BlowOff. Theory and Experiment. Fourth
more, 1953, pp. 68-89.
[21] Berlad, A. L. and Potter, A. B.: Effect of Channel Geometry on the Quenching
of Laminar Flames. NACA Research Memorandum RM E54C05, 1954.
[22] Berlad, A. L.: Flame Quenching by a Variable Width RectangularChannel
Burner as a Function of Pressure for Various PropaneOxygenNitrogen Mix-
tures. Journal of Physical Chemistry, Vol. 58, 1954, pp. 1023-1026.
[23] Massey, B. S. and Lindley, B. O.: Flame Quenching. Journal of the Royal
Aeronautical Society, Jan. 1958, pp. 32-42.

Chapter 9
Flows with combustion waves
9.1 Introduction
As previously seen in chapter 6, in a combustible mixture at ambient of higher pres-

sure, the thickness of the ame is of the order of a fraction of millimeter. This length
is generally small when compared to those of interest in aerothermodynamic prob-
lems. Therefore, when studying these problems it is justied to assume that the ame
thickness is zero. In such a case the ame may be considered as a surface of disconti-
nuity of the pressure, temperature and velocity. The same idea is applied, for example,
in the study of gas dynamic problems involving shock waves. The problem has been
studied from this view point mainly by Emmons and his co-operators [1], [2], [3], [4].
In this chapter the general lines of Emmonss work are followed. First the conditions
that must be satised across the ame are established and then some general proper-
ties deduced. In the chapter that follows, the method is applied to the study of the
aerothermodynamic eld originated by a ame stabilized in a combustion chamber.
This problem has been studied by Scurlock, Tsien, and Fabri-Siestrunck-Foure.
9.2 Conditions that must be satised by the jump across

a ame front.
As aforesaid, the relations deduced herein are applicable only if the thickness of the
ame front is small when compared to a characteristic length of the phenomenon under
study.1 The ame thickness must also be small compared to the radius of curvature
1 For instance, this does not occur in rareed mixtures where the thickness of the ame can be large.
231
232 CHAPTER 9. FLOWS WITH COMBUSTION WAVES
of the ame. In fact, if the radius of curvature and the ame thickness are of the
same order of magnitude the ame speed depends not only on the thermodynamic
conditions of the unburnt gases, but also on the shape of the ame in the neighborhood
of the point under consideration. Such is the case in the ame tip of a Bunsen burner.
Here, the ame speed is increased due to the favorable effect of the curvature on heat
transfer and diffusion of the active particles towards the unburnt gases.
V n1 V t1
p1 , T 1 , V t2
1
V n2
p2 , T2 , 2
Figure 9.1: Schematic diagram of a surface element of a ame.
Thus, let it be a ame front of small thickness and small curvature as indicated
in Fig. 9.1, and let us consider a surface element of this front. By applying the conti-
nuity, momentum, and energy equations to the gases at both sides of this element, as
it is done in the study of the invariants across a shock wave2 , the following conditions
are obtained which relate the state of the unburnt and burnt gases at both sides of the
elements:
1) Continuity equation.
1 vn1 = 2 vn2 . (9.1)
2) Momentum equation.
a) Normal component.
2 2
1 vn1 + p1 = 2 vn2 + p2 . (9.2)
b) Tangential component.
vt1 = vt2 . (9.3)
3) Energy equation.
1 2 1
v1 + h1 = v22 + h2 . (9.4)
2 2
Subscripts 1 and 2 refer to conditions before and after the ame, respectively,
v is the velocity of the gases relative to the ame front, assumed stationary, and vn
2 See chapter 6.
9.2. CONDITIONS THAT MUST BE SATISFIED BY THE JUMP ACROSS A FLAME FRONT. 233
and vt are the normal and tangential components of this velocity. Moreover and p
are, as usual, the density and pressure of the gases, and h is the total specic enthalpy,
that is, (as seen in chapter 1), the thermal enthalpy hT plus the formation enthalpy hf ,
h = hT + hf , (9.5)
where hT is expressed, in terms of the heat capacity cp at constant pressure and the
absolute temperature T , as follows
T
hT = cp dT, (9.6)
0
which, for the special case cp = constant reduces to
hT = cp T. (9.7)
At each side of the ame front, the pressure, density and temperature are related
by the state equation, which is assumed to be that of the perfect gases, that is
p1 p2
= R1 T1 and = R2 T 2 , (9.8)
1 2
where
R R
R1 = and R2 = (9.9)
Mu Mb
are the specic constants of the unburnt and burnt gases, respectively, and Mu and Mb
their respective average molecular masses.
Since the tangential component of the velocity is continuous throughout the

ame, there results that the incoming and outgoing velocities and the vector normal to
the combustion front are coplanar.
When the state of the unburnt gases and the incline 1 of the incident ow
relative to the ame front, see Fig. 9.2, are known, the system of equations (9.1),
(9.2), (9.3), (9.4) and (9.8) determines the state of the burnt gases.
Let be the ame propagation velocity corresponding to the composition of

the unburnt gas mixture in the thermodynamic state dened by the values p1 and 1
of pressure and density. Obviously we have
= vn1 , (9.10)
and for the incline 1 of the ame front relative to the incident ow
vn1
sin 1 = = . (9.11)
v1 v1
V t1 V n1 2
1
V1
1
= V n1
V n2 V2
V t2 = V t1
Figure 9.2: Velocity components at both sides of the front.
Therefore, 1 is determined by the velocity v1 of the incident ow relative to the

front and by the burning velocity of the mixture. This velocity depends only on the
composition, pressure and density or temperature, and must be considered as a datum,
determined either theoretically or experimentally (see chapter 6).
In order to simplify calculations, it is assumed in the following that the thermal

enthalpy can be expressed as indicated in (9.7). Let
q = hf 1 hf 2 (9.12)
be the difference between the formation enthalpies of the unburnt and burnt gases. By
substituting equations (9.7) and (9.12) into (9.4), this equation can be written
1 2 1
v1 + cp1 T1 + q = v22 + cp2 T2 . (9.13)
2 2
Let Ts1 and Ts2 be the stagnation temperatures of the unburnt and burnt gases
respectively. By virtue of equation (9.13) they are related by
cp1 Ts1 + q = cp2 Ts2 , (9.14)
and their ratio n is given by

( )
Ts2 q cp1
n= = 1+ . (9.15)
Ts1 cp1 Ts1 cp2
For the particular case cp1 = cp2 , we have
q
n=1+ . (9.16)
cp1 Ts1
9.3. NORMAL FLAME FRONT 235
9.3 Normal ame front
Let us consider a normal ame front as indicated in Fig. 9.3. In such a case, the
equations that relate the conditions before and after the ame reduce to
1 v1 =2 v2 , (9.17)
1 v12 + p1 =2 v22 + p2 , (9.18)
1 2 1
v1 + cp1 T1 + q = v22 + cp2 T2 . (9.19)
2 2
V1 V2
p1 , T 1 , p2 , T2 , 2
1
Figure 9.3: Schematic diagram of a normal ame front.
Since, v1 is equal to the ame propagation velocity in the unburnt gases, (which
is normally of the order of 1 m/s) and v2 is only a few times larger, both v1 and v2
are very small when compared to the sound speed in the unburnt and burnt gases
respectively. Therefore, in equation (9.19) the terms containing the kinetic energy of
the unburnt and burnt gases can be neglected. There results
cp1 T1 + q ' cp1 Ts1 + q = cp2 Ts2 ' cp2 T2 . (9.20)
That is
T1 ' Ts1 , T2 ' Ts2 . (9.21)
Therefore, by virtue of Eq. (9.15)
T2 Ts2
' = n. (9.22)
T1 Ts1
p2 p1
Furthermore, as results from Eq. (9.18) the pressure drop is very small
p1
compared to unity, and, consequently, for the calculation of the thermodynamic state
of the burnt gases it may be assumed to be zero. The density ratio is then given by
2 T1 1
= = . (9.23)
1 T2 n
The ratio of velocities is

v2
= n, (9.24)
v1
obtained from Eqs. (9.23) and (9.17).
Therefore, once the density and temperature of the unburnt gases and the value
of n are known, the temperature of the burnt gases is determined by equation (9.22),
its density by equation (9.23), and the pressure drop across the ame by
p2 p1
= 1 M12 (1 n), (9.25)
p1
where 1 is the ratio of the heat capacity at constant pressure to the heat capacity at
constant volume of the unburnt gases, and
v1
M1 = (9.26)
a1
is the ow Mach number for the unburnt gases, where

p1
a1 = 1 (9.27)
1
is the sound speed in these same gases.
9.4 Inclined ame front
Here, vn1 and vn2 are still small compared to the sound speed of the unburnt and
burnt gases respectively. Therefore, the pressure drop across the ame front is still
small, see Eq. (9.2), and the ratio between densities on each side of the ame front is
determined by the ratio of temperature T2 to temperature T1 . Let be this ratio, there
results
T2 1 vn2
= = = . (9.28)
T1 2 vn1
Let
= 2 1 (9.29)
be the velocity deviation across the ame front (see Fig. 9.2) . A simple calculation
gives for
( 1) tan 1
tan = . (9.30)
1 + tan2 1
This ratio has been taken into Fig. 9.4 for different values of . As it can easily be
seen, the maximum deviation max corresponds to
1
tan 1 = . (9.31)

9.4. INCLINED FLAME FRONT 237
60
= 12
8
50
6
5
40
4
( )
3
o
30
20 =2

max
10
0
10 30 50 70 90
o
1 ( )
Figure 9.4: Velocity deviation across the ame front as a function of 1 .
Its value is given by ) (

1 1
tan max = . (9.32)
2
Even though the normal velocity is always small, the tangential velocity vt , which
by virtue of equation (9.3) is the same at both sides of the front, can be large. There-
fore, it must be taken into account when writing the energy equation. As a result, the
temperature at each side of the front can differ, considerably, from the corresponding
stagnation temperature. This is exactly the opposite to what occurs in the case of a
normal front.
Therefore, two different cases are to be considered, either
vt ' O(vn ), (9.33)
in which case
T1 ' Ts1 , T2 ' Ts2 , ' n, (9.34)
or
vt vn , (9.35)
in which case
v12 ' v22 ' vt2 , (9.36)
and equation (9.13) reduces to
cp1 T1 + q = cp2 T2 . (9.37)

By virtue of equation (9.15) and of the relation

Ts1 1 1 2
=1+ M1 , (9.38)
T1 2
the following expression is obtained for
( )
1 1 cp1
=n+ n M12 , (9.39)
2 cp2
which shows that for large values of M1 , that is, when the ame front is very inclined
to the incident ow, can appreciably differ from n.
The values of /n as a function of M1 , for cp2 = cp1 and different values of

1 1 n 1
= , have been represented in Fig. 9.5.
2 n
1.20 =0.20
1.18 =0.18
1.16 =0.16
=(11)(n1)/2n
1.14 =0.14
1.12 =0.12
/n
1.10 =0.10
1.06
1.06
1.04
1.02
1.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
M1
Figure 9.5: Values of the ratio /n as a function of incident Mach number M1 .
Relation (9.39) is only valid for very inclined ame front, that is, for not too
small values of M1 . For values of M1 close to zero, it must be substituted by
' n. (9.40)
The rst case, that is to say, the case of slow ows (M1 1) with ame
fronts, has been examined in detail by Gross and Esch [4], reaching the following
conclusions:
1) If , and q are constant and the motion is irrotational before the ame, after
the ame it continues being irrotational.
9.5. ENTROPY JUMP ACROSS THE FLAME FRONT 239
PRODUCTS
FLAME FRONT
REACTANTS
Figure 9.6: Straight-line ame ow eld according to Gross and Esch for /v1 = 0.1 and
= 7.
2) Opposite to what occurs with fast ows, if the ow is slow pressure drop across
the ame front cannot be neglected.
In order to analyze the motion, the ame front can be considered as a surface with a
distribution of sources. For example, in the case of a plane motion, the strength of the
( 1)
source per unit length of the front is . Now by expressing the condition that
2
in the unburnt gases the velocity normal to the front must be , an integral equation
is obtained which must satised at the front. This integral equation determines the
ame shape which a priori is unknown. Gross and Esch give approximate solutions
for certain cases. For example, for the case in which the ame front reduces to a
straight segment inclined to the incident ow. Fig. 9.6, taken from the said work,
shows the shape of the streamlines in this case.
9.5 Entropy jump across the ame front
The entropy of the unburnt gases is given by the expression3

1/
p1 1
S1 = S01 + cp1 ln , (9.41)
1
3 See chapter 1.
where S01 depends only on the composition of the mixture. Likewise, the entropy of
the burnt gases is
1/
p 2
S2 = S02 + cp2 ln 2 , (9.42)
2
Therefore, the entropy jump S across the ame, is
1/ 1/
p2 2 p 1
S = S2 S1 = (S02 S01 ) + cp2 ln cp1 ln 1 . (9.43)
2 1
As previously seen, the pressure drop across the ame is very small. Therefore, in
(9.43) we can take
p2 ' p1 = p. (9.44)
Making use of this simplication and of the relation (9.28) between p1 and p2 , the
following expression is obtained for S
p1/12
S = S2 S1 = (S02 S01 ) + cp2 ln + (cp2 cp1 ) ln , (9.45)
1
being
cp2 cp1
12 = , (9.46)
cv2 cv1
where cv1 and cv2 are the heat capacities at constant volume of the unburnt and burnt
gases respectively.
In the particular case

cp2 = cp1 = cp , (9.47)
the entropy jump across the ame reduces to
S = (S02 S01 ) + cp ln . (9.48)
Equation (9.39) shows that can vary along the ame front when M1 varies. There-
fore, if the local conditions of the ow vary, the entropy jump across the ame front
can vary considerably from one point to another. Due to this variation of the en-
tropy jump along the ame front, the motion after the front can be rotational, even
if the motion before the front is potential. This is similar to what occurs in the case
of supersonic motions with shock waves, when their incline varies from one point to
another.4
If the motion before the ame is isentropic, the value depends only on the
incline 1 of the ame front. In fact, by using the relation,
vn1
tan 1 = = , (9.49)
vt vt
4 See A. Ferri: Elements of Aerodynamic of Supersonic Flows. Mac Millan Comp., New York, 1949, p.
57.
9.5. ENTROPY JUMP ACROSS THE FLAME FRONT 241
that gives the incline of the ame front, and the relation
vt
M1 ' , (9.50)
a
which, as seen in the preceding paragraph, is valid for the very inclined ame fronts,
Bernoulli equation for unburnt gases can be written
( )2
1 1 2 a01
1+ M1 = M12 tan2 1 , (9.51)
2
where a01 is the sound speed at the stagnation point of the unburnt gases.
By eliminating M1 between this equation and equation (9.39), can be ex-

pressed as a function of the ame incline, as follows
cp1
n
cp2
=n+ ( )2 . (9.52)
2 a01
tan2 1 1
1 1
When the ame incline is increased, 1 decreases, and relation (9.52) show that
increases. Therefore, the entropy jump S increases when the ame incline is in-
creased. Such behavior is the opposite to that of a shock wave where the entropy jump
decreases when the wave incline is increased. See Fig. 9.7 [5] . The inuence of
this variation of the entropy jump on the ow will be studied in the following para-
graph. The simplied expression (9.48) for the entropy jump will be used, and since
the constant S02 S01 has no inuence on the ow, we shall express in short
S = cp ln . (9.53)
2 S
S 2
(a) Flame (b) Shock wave
Figure 9.7: Entropy jump across a ame front and a shock wave.
9.6 Vorticity across the ame
In the ow of a perfect gas, where viscosity and mass forces are neglected, the varia-
tion of the rotational = v as a function of the stagnation temperature Ts and
specic entropy S can be expressed as follows

+ v = cp Ts T S. (9.54)
t
For isoenergetic (Ts = const.) and stationary (/t = 0) ows, only this
case will be considered in the present study,5 equation (9.54) reduces to the following
(Croccos Theorem)
v = T S. (9.55)
From this equation, the jump of the rotational, across the ame, can be computed
when the values of v, T and S are known. The jump of the rotational is obtained by
(9.55) to both sides of the ame. For the calculation, we shall adopt on each point
of the ame front a cartesian rectangular coordinate system, dened in the following
way:
Axis n: Normal to the ame front.
Axis t: Intersection of the plane tangent to the ame with the plane of the in-
cident and emergent velocities v1 and v2 .
Axis : Normal to the (n, t) plane forming a positive trihedron.

The velocity components relative to this system, will be vn , vt and 0. The
vorticity components n , t and . Therefore, those of the vector product v will
be vt , vn and (vn t vt n ). Consequently, equation (9.55) breaks down into
the following three equations6
S
vt = T , (9.56)
n
S
vn = T , (9.57)
t
S
vn t vt n = T . (9.58)

The jump of the normal component of the vorticity n can be computed di-
rectly. The above equations are not necessary for the computation. In fact, since vt
5 If the ow before the ame is isoenergetic and if the heat q released in the combustion is constant, as
will be assumed hereinafter, the ow after the ame is also isoenergetic, as results from (9.14).
6 The derivative S/t in the tangential direction must not be confused with a time derivatives.
9.6. VORTICITY ACROSS THE FLAME 243
is continuous across the ame, by applying Stokes theorem to both faces of a surface
element of the ame, the following is obtained
n1 = n2 , (9.59)
that is, the normal component of the vorticity is continuous across the ame.
The jump of can then be obtained from (9.57) by writing this equation for
each side of the ame and forming the difference. Thus, when taking into account that
vn1 = , vn2 = , T2 = T1 , S2 = S1 + S there results
[ ]
2 T1 1 S1 (S)
2 = + + . (9.60)
t t
Likewise by using (9.58) we obtain for t2
[ ]
t1 T1 1 S1 (S)
t2 = + . (9.61)

On the other hand, equations (9.57) and (9.58) , when written for conditions before
the ame, give
S1
1 = T1 , (9.62)
t
S1
t1 vt n1 = T1 . (9.63)

S1 S1
By eliminating and between these two equations and (9.60) and (9.61), we
t
nally obtain for 2 and t2 ,
T1 (S)
2 = 1 + , (9.64)
t
1 T1 (S)
t2 = t1 n1 cot 1 . (9.65)

In particular, if the motion before the ame is irrotational (n1 = t1 = 1 =

0), the vorticity produced by the ame results
n2 = 0, (9.66)
T1 (S)
t2 = , (9.67)

T1 (S)
2 = . (9.68)
t
Furthermore, if the entropy jump can be expressed as (9.53), we have
cp T1 1
t2 = , (9.69)

cp T1 1
2 = , (9.70)
t
or else, if the motion is rotational, the equations resulting from (9.64) and (9.65) when
S is substituted therein for its value (9.53).
In the special case of a plane motion, the only component different from zero
of the vorticity is , and its value after the ame is given by (9.64) if the motion
before the ame is rotational, and by (9.68) if it is potential. In Fig. 9.7.a, the rotation
sense of the vorticity generated by the ame has bean indicated, in the case of a plane
motion, assuming that the motion before the ame is potential. In Fig. 9.7.b, the sense
corresponding to a shock wave is shown for the same case.
References
[1] Emmons, H. W., Ball, G. A. and Maier, A. D.: Development of a Combus-

tion Tunnel. Army Ordenance Project Report, Harvard University, Cambridge
Mass., 1954.
[2] Emmons, H. W.: Fundamentals of Gas Dynamics. Sec. E, Vol. III of High
Speed Aerodynamics and Jet Propulsion. Princeton University Press. 1958.
[3] Gross, R. A.: Combustion Tunnel Laboratory Interim Technical Report No. 2.
Harvard University, June 1952.
[4] Gross, R. A. and Esch, R.: Low Speed Combustion Aerodynamics. Jet Propul-
sion, March-April 1954, pp. 95-101.
[5] von Karman, Th.: Aerothermodynamics and Combustion Theory. LAerotecnica,
Vol. XXXIII, Fasc. 1st., 1953, pp. 80-86.
Chapter 10
Aerothermodynamic eld of a
stabilized ame
10.1 Introduction
As an example of the application of the aerothermodynamic method that was outlined

in the preceding chapter for the study of gas ows, in the present chapter we shall
study the characteristics of the ow in a combustion chamber with a stabilized ame.
y
+h F
u0 0
x
h F
Figure 10.1: Schematic diagram of a stabilized ame front in a two-dimensional combus-

tion chamber.
The problem is outlined in Fig. 10.1. To simplify calculation we shall consider

the case of a two dimensional chamber of constant width 2h and unit thickness normal
to the plane of the gure. An uniform ow of homogeneous pre-mixed combustible
enters the combustion chamber. The velocity, pressure, density and temperature of
the mixture are u0 , p0 , 0 and T0 , respectively. The ame front FOF is stabilized at
point O by one of the methods studied in the next chapter. As expressed in the pre-
245
246 CHAPTER 10. AEROTHERMODYNAMIC FIELD OF A STABILIZED FLAME
ceding chapter, this ame front can be considered as a discontinuity surface between
unburnt and burnt gases.
The problem lies in determining:
a) The width of the ame as a function of the fraction of gas burnt.

b) The pressure drop along the chamber.
c) The velocity distribution in the different sections of the chamber.
d) The shape of the ame.
The following simplifying assumptions are introduced:
1) Combustion efciency is the same at all points, that is, the heat released in the
combustion per unit mass of fuel is independent from the state of the gas before
the ame.
2) Heat capacity at constant pressure cp is independent from temperature, and has
the same value for the unburnt and burnt gases.
3) Unburnt and burnt gases behave as perfect gases. The constant Rg of their state
equation has the same value for both gases
p
= Rg T. (10.1)

4) The ame propagation velocity is constant along the front and very small when
compared to the gas velocity.
5) Unburnt and burnt gases are ideal uids, their viscosity and thermal conductivity
are negligible.
Furthermore, in this study only the case of stationary ow will be considered.

Thus stated, this problem has been studied by A.C. Scurlock [1] and [2], who intro-
duced further simplication by assuming both unburnt and burnt gases to be incom-
pressible uids.
Since the ame speed is small compared to the gas speed, the angle between the
front and the streamlines is very small. Therefore, the streamlines are approximately
parallel to the chamber axis. Furthermore, due to the negligibility of the pressure drop
across the ame front, it can be assumed that pressure is constant at each cross section
of the chamber. Moreover, gas speed can be substituted by its component parallel
to the chamber axis. Therefore, from these assumptions an almost one-dimensional
theory can be worked out.
By using the aforementioned simplifying assumptions Scurlock obtained nu-

merical solutions for the equations of motion in some typical cases. In order to per-
form numerical computations, Scurlock substitutes the differential equations for nite
differences.
Some of the results, taken from Ref. [1], are given in Figs. 10.2 to 10.5. These
results correspond to the case where the density ratio of the unburnt gases to the burnt
gases is 6, and the ratio of the ame speed to the gas velocity at the chambers inlet is
= 0.10.
y/h A B C
1
0
x/h
1
0 1 2 3 4 5 6 7 8 9 10 11
P / P t
0 1 2 3 4 5 6 7 8 9 10 11 x/h
Figure 10.2: Streamlines of ow through a ame front in a two-dimensional chamber and

pressure drop along it.
Fig. 10.2 shows the shape of the ame front and of the streamlines, as well
as the pressure drop p along the chamber referred to the total pressure drop pt ,
between the inlet and outlet sections.
Fig. 10.3 shows the velocity distribution for the cross-sections A, B and C
indicated in Fig. 10.2. These sections correspond to the point at which combustion
starts (section A), to the point where the burnt fraction is 0.25 (section B) and to the
point where this fraction is 0.75 (section C). In this gure all velocities are referred to
the velocity u0 of the gas at the inlet of the chamber. Fig. 10.3 shows that:
1) While the velocity of the unburnt gases is constant at each cross section, the
velocity of the burnt gases increases from the ame front towards the chamber
axis where it is maximum.
y/h
1
A B C
0
U/U 0
1
0 1 2 3 4 5 6 7 8 9 10
Figure 10.3: Velocity proles for three cross-sections of the two-dimensional chamber (see
Fig. 10.2).
1.0
0.9
0.8
0.7
0.6
/h
0.5
0.4
0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
f
Figure 10.4: Flame width as a function of the burnt fraction for incompressible ow and
= 6.
2) Due to the pressure drop produced by combustion, the mass of gas accelerates
downstream of the chamber.
Fig. 10.4 shows the variation law of the ame width, referred to that of the
chamber, as a function of the burnt fraction.
Fig. 10.5 shows the fraction of pressure drop along the chamber, referred to
the total pressure drop between the inlet and outlet sections, as a function of the burnt
fraction.
1.0
0.9
0.8
0.7
0.6
P / Pt
0.5
0.4
0.3
0.2
0.1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
f
Figure 10.5: Pressure drop as a function of the burnt fraction for incompressible ow and
= 6.
G.A. Ball, [3] and [4], has studied the same problem by applying the relaxation
method.
The assumption that density, for both unburnt and burnt gases, is constant,
is permissible only when the variations of the Mach number along the chamber are
small. Due to the acceleration produced by the pressure drop, the local Mach number
close to the walls and near the end of the chamber can be large, even for small entrance
Mach numbers. As a consequence a choking of the chamber can occur preventing the
complete combustion of the gas, as will be shown later. This choking phenomenon
cannot be predicted by Scurlocks hydrodynamic theory.
H.S. Tsien in the U.S.A. [5], and J. Fabri, R. Siestrunck and C. Foure in France,
[6], [7] and [8], have studied (simultaneous but independently) the inuence of gas
compressibility on the problem.
By assuming at each section a linear velocity distribution for the burnt gases
(see Fig. 10.3), H. S. Tsien has demonstrated that an approximate analytical solution
of Scurlocks problem can be easily obtained. The results are very close to those
obtained by Scurlock. The linear velocity a distribution is a good approximation to the
prole calculated by Scurlock. Moreover, Tsien has extended the analysis, taking into
account compressibility effects, by applying the same idea of a quasi one-dimensional
theory. He also assumes a linear velocity distribution for the burnt gases, showing that
in this case choking of the chamber can occur.
Fabri, Siestrunck and Foure have developed a similar theory. They also include
the effect of density variations, but do not postulate a linear velocity prole for burnt
gases. Their stating of the problem leads then to an integral equation for the stream
function. For some typical cases they have integrated this equation numerically. Fur-
thermore, they extended the analysis to other types of chambers. For instance, the
cylindrical chamber with circular cross section and different arrangements of the ame
stabilizer.
As a result of their studies Fabri, Siestrunck and Foure also predict the occur-
rence of choking when the velocity at the inlet section is larger than a critical velocity.
In the following sections the approximate method of Tsien will be described

rst and then the method developed by Fabri-Siestrunck-Foure.
10.2 Tsien method
p p
h
B

1
1
C
y u1
1
u0 D
u
y

2
ue
0
A
Figure 10.6: Notation for the Tsien method.
Figure 10.6 contains the necessary elements for Tsiens approximate calcula-
tion.
By applying the continuity equation to section AB, the following equation is

obtained y 1
1 u1 (h y1 ) + u dy = 0 u0 h. (10.2)
0
By introducing = y1 /h and U = u1 /u0 it can be written in the following dimen-

sionless form
1
u (y)
U (1 ) + d = 1. (10.2.a)
0 0 0 u0 h
10.2. TSIEN METHOD 251
As aforesaid, Tsien assumes that the velocity distribution of the burnt gases
between A and C is linear, that is, u (Fig. 10.6) is given by the expression
( )
y
u = u1 + (ue u1 ) 1 , (10.3)
y1
where ue is the velocity on the axis of the chamber.
Furthermore, Tsien assumes that the ratio of the density of the unburnt gases
to that of the burnt gases at each point of the ame front is constant.1 Then, it can be
easily shown that the density of the burnt gases is constant between A and C and equal
to 1 /
1
= . (10.4)

In fact, in D we have
01
= . (10.5)
02
But due to the fact that the expansions of both unburnt and burnt gases between
pressure p0 in D and p in C are isentropic, there result
( )1
01 p0
= (10.6)
1 p
and
( )1
02 p0
= . (10.7)
p
The combination of (10.5), (10.6) and (10.7) leads to (10.4).
Since pressure and density of the burnt gases are constant between A and C, the
temperature T of the burnt gases must also be constant between A and C, and equal to
T2
T = T2 = T1 = Te , (10.8)
where Te is the temperature at A.
The Bernoulli equation, when applied to the unburnt gas between u0 and u1 ,
gives
1 2 1
u0 + cp T0 = u21 + cp T1 . (10.9)
2 2
If T1 is eliminated by combining this equation with the equation
( ) 1
1 T1 1
= (10.10)
0 T0
1 Relation (9.36) in the preceding chapter proves that this ratio actually varies with the ame front incline.
of the reversible adiabatics, it results for 1 /0
( ) 1
1 1 2 2 1
= 1 M0 (U 1) = , (10.11)
0 2 0

where M0 = u0 / ( 1)cp T0 is the Mach number at the inlet section of the com-
bustion chamber.
On the other hand, Bernoullis equation applied to the burnt gases between O
and A, taking (10.8) into account, gives
1 2 1
u + cp T0 = u2e + cp T1 . (10.12)
2 0 2
Finally, the elimination of T0 and T1 between (10.9) and (10.12) gives for ue
ue
= 1 + (U 2 1). (10.13)
u0
When (10.3), (10.11) and (10.13) are substituted into (10.2.a) and the integra-
tion performed, the following expression for as a function of U is obtained
( ) 1
1 2 2 1
U 1 M0 (U 1)
2
= 1 ( ) . (10.14)
(2 1)U 1 + (U 2 1)
2
The fraction f of the burnt gases is given by the expression
0 u0 h 1 u1 (h1 y1 )
f=
0 u0 h
1 u1
=1 (1 )
0 u0
( ) 1
1 2 2 1
=1 U (1 ) 1 M0 (U 1) . (10.15)
2
System (10.14) and (10.15) allow computation of the fraction f of burnt gas as a
function of the ame width , through parameter U . Fig. 10.7 shows several of the
results given by Tsien in his work for the case = 6.
The main result of Tsiens work is the following: for each value of there is a
critical Mach number Mcr for the ow at the inlet section, such that when M0 < Mcr
the combustion is complete and the ame reaches the chamber walls. On the other
hand, if M0 > Mcr the maximum burnt fraction fmax is smaller than unity and the
ame cannot reach the walls. This is known as the choking phenomenon. It shows
10.2. TSIEN METHOD 253
M =0, 0.1
0
1.0
M =0.2
0
=6.0
0.8
0.6
0.50

M =0.4 M0=0.4
0
0.4
0.25
M =0.6
0
0.2 M =0.6
0 0.00
0.0 0.1 0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
f
Figure 10.7: Flame width as a function of the burnt fraction for density ratio = 6. For
each value of the initial Mach number M0 , a dot indicates the maximum values
of and f .
that, aside from the difculties arising from the problem of ame stabilization at high
speed,2 other difculties are to be expected when trying to burn gas at high speed, due
to the interaction between aerodynamic and combustion processes.
Choking occurs due to the fact that the contraction of the unburnt gases orig-
inated by downstream acceleration is not sufcient to compensate the expansion of
the gases as they burn. Compressibility acts because the contraction corresponding
to a given acceleration is smaller in a compressible uid than in an incompressible
one. This is the reason why the choking phenomenon does not exist in Scurlocks
hydrodynamic theory.
The critical Mach number that corresponds to each value of can be obtained
by expressing the condition that the curve of f as a function of U , obtained by elimi-
nating between (10.14) and (10.15), must have a maximum at the point f = 1. That
is, when the conditions
df
= 0 and f = 1, (10.16)
dU
or their equivalents
d
=0 and = 1, (10.17)
dU
are satised.
2 The problem of stabilizing a ame at high speed is studied in the next chapter.
10 1.0
M
=1
8 0.8
M
cr
Mcr, M=1
6 0.6
=1
U
4 0.4
2 0.2
U=1
0 0.0
0 2 4 6 8 10 12

Figure 10.8: Critical conditions for the complete combustion as a function of the density
ratio .
Figure 10.8, in which the values of Mcr thus calculated have been taken from
[5], shows that the choking effects impose an important limitation to the velocity at
which gases enter the combustion chamber. Choking effects can be avoided by using
an expanding combustion chamber. It is interesting to point out that when the maxi-
mum Mach number is calculated at critical conditions, that is, at the initiation of the
choking phenomenon, this number is slightly larger than one. This value corresponds
to the unburnt gases at the point where the ame reaches the chamber wall. Therefore,
in the neighbourhood of this point, there exists a small area where the ow is slightly
supersonic. Details of the calculation can be found in Tsiens work. Furthermore,
Tsiens analysis shows that Scurlocks hydrodynamic theory gives satisfactory results
when the entry Mach number is not too close to the critical.
Recently, G. Ernst [9] has applied Tsiens method to the same problem as well
as to the study of other types of combustion chambers and other arrangements of the
stabilizer. Ernst substitutes Tsiens linear velocity prole for a parabolic one with
an exponent 1.33, which is the one that best ts the proles obtained by Scurlock.
Moreover, instead of assuming a constant density for the burnt gases at each cross
section of the chamber, he includes the effect of the slight density variation of the burnt
gases. However, Ernsts results only differ slightly from those obtained by Tsien.
10.3. METHOD OF FABRI-SIESTRUNCK-FOURE 255
10.3 Method of Fabri-Siestrunck-Foure
The ow in the combustion chamber is two dimensional and stationary. Therefore, a

stream function exists. This function is dened by the following relations

,
u =0 u0 h (10.18)
y

v =0 u0 h . (10.19)
x
Thus dened, is non-dimensional.
The x axis is a streamline to which the value = 0 is assigned. The upper

wall of the chamber, y = h, is also a streamline for which = 1, as results from the
integration of (10.18). Therefore, in the upper half of the chamber, varies from the
value zero, at the axis, to the value one, at the wall.
A linearization of the problem, by assuming that all velocities are small com-
pared to u, leads to the conclusion that pressure is constant at each cross section of
the chamber. Therefore, since the ow of unburnt gases is isentropic, their density,
temperature and velocities are equally constant at each cross section. Moreover, the
value of either one of these magnitudes determines the values for the others.
B

1
C
u

D u

E u

x
0 A
Figure 10.9: Notation for the method of Fabri-Siestrunck-Foure.
Let us consider a cross section AB, at a distance x from the stabilizer, as shown
in Fig. 10.9. Here, evidently

1 h
dy = 1, (10.20)
h 0
which, by virtue of (10.18), can be written in the following form
1
0 u0
d = 1. (10.21)
0 u
Let us consider the streamline that crosses the ame front at section AB
(Fig. 10.9) at point C. At this section, the streamline separates the unburnt from the
burnt gases. To the rst correspond values 0 of the stream function between and 1.
To the latter correspond values 0 between 0 and . By separating in (10.21) both
intervals, the following is obtained;
1
0 u0 0 u0
d 0 + d 0 = 1, (10.22)
0 00 u00 u
where 00 and u00 are the corresponding values of and u, on the streamline 0 at point
E on section AB.
As previously said, both density and velocity of the unburnt gas are constant
at each cross section of the chamber. If 1 and u are their values at section BC, the
second integral in (10.22) can be integrated, obtaining
1
0 u0 0 u0
d 0 = (1 ). (10.23)
1 u 1 u
This value, when carried into (10.21), gives

0 u0 0 u0
d 0 = 1 (1 ). (10.24)
0 00 u00 1 u
This equation can be written

1 u 1 u
1+ = d 0 . (10.25)
0 u0 0 00 u00
Hereinafter all velocities will be referred to the maximum velocity umax =
2cp Ts,1 of the unburnt gases, and the ratio named U . Equation (10.25) is then
written as follows

1 U 1 U
1+ = d 0 . (10.26)
0 U0 0 00 U 00
1 U
The problem lies now in expressing as a function of U and of the di-
00 U 00
mensionless velocity U at the point D where the streamline 0 crosses the ame,
0
Fig. 10.9. Let us see how it can be done.
Proceeding in the same manner as done for the deduction of equations (10.4)
and (10.8) in Tsiens method, that is, by comparing the expansions along and 0 ,
the following is obtained
1 T 00
0 = 00 = . (10.27)
T1
10.3. METHOD OF FABRI-SIESTRUNCK-FOURE 257
Moreover, it is immediately obtained that

n U0
2
0 = , (10.28)
1 U 02
from which results
T 00 n U 0
2
1
= . (10.29)
00 T1 1 U 0 2
Bernoullis equation when applied to E, gives
T 00
U 00 +
2
= n. (10.30)
Ts,1
Likewise, Bernoullis equation when applied to C for the unburnt gases, gives
T1
U2 + = 1. (10.31)
Ts,1
By combining (10.29), (10.30) and (10.31), the following is obtained for U 00

nU 2 U 0 2 (n 1 + U 2 )
U 00 = . (10.32)
1 U 02
Substituting (10.29) and (10.32) into (10.26), and making use of relation
( ) 1
1 1 U2 1
= , (10.33)
0 1 U02
the following equation is nally obtained
( ) 1
U 1 U2 1
1+
U0 1 U02
(10.34)
(n U 0 )
2
U 0
= ( ) d .
n n1+U 2
0
(1 U 0 2 ) U 2 U 0 2
n
This expression is an integral equation which determines as a function of U . Once

the solution is known, the value of gives the burnt fraction as a function of the
velocity U of the gases at the point where the streamline crosses the ame front.
The ame width = y/h is given by the expression

0 U0
=1 (1 ). (10.35)
1 U
To determine the shape of the ame front, the abscissa = x/h corresponding to
must also be known. This abscissa is given by the following expression

0 U0
= 0
d 0 , (10.36)
0 1
which is deduced from (10.19) by making v ' . When computing this expression
the following relation must be used
( ) 1
01 1 U0
2 1
= (10.37)
0 1 U02
as well as the values 0 = (U 0 ) given by the solution of (10.34).
Equation (10.34) can be integrated numerically by substituting the integral for

a summation with a nite number of terms. Fabri, Siestrunck and Foure have, thus,
calculated the solutions corresponding to several typical cases. Some of their results
can be found in the references included herein.
When calculating the solution it is found that for each value of n there is a
corresponding value U0,cr of U0 such that if U0 > U0,cr the curve of = (U )
presents an horizontal tangent for a value of smaller than one. This means that in
such a case the burnt fraction must be smaller than unity. Therefore, the existence
of a critical Mach number for the inlet ow is also obtained here, thereupon choking
occurs. Furthermore, it can be proved that the value of U0,cr is equal to the value

( ) 1
U0,cr = n n1 , (10.38)
+1
even by the one-dimensional theory that results from the assumption that at each cross-
section of the chamber the distribution of velocities is uniform. For such a case the
value of the nal Mach number of the burnt gases is unity.3
Fig. 10.10, taken from Ref. [8], shows a solution of (10.34) that corresponds
to the case n = 6 for the three following values of U0 : subcritical value U0 = 0.05,
critical value U0 = U0,cr = 0.087, and supercritical value U0 = 0.100. In the latter,
the burnt fraction would be only max = 0.8.
The experimental evidence available is not enough to judge the good approxi-
mation of these methods.
10.4 Cylindrical chambers
As Fabri, Siestrunck and Foure have demonstrated [6] the case of a cylindrical cham-
ber of circular cross-section reduces to the two-dimensional problem studied in the
preceding paragraph. In fact, in such case equation (10.18), which denes the stream
3 See chapter 3, 9.
10.5. CHAMBER WITH SLOWLY VARYING CROSS-SECTION 259
0.5
0.4
U0 = 0.100
0.3
U
U0 = 0.087
0.2
U0 = 0.050
0.1
0
0.2 0.4 0.6 0.8 1.0

Figure 10.10: Velocity proles as a function of the burnt mass fraction for = 6 and three
values of the initial velocity (subcritical, critical and supercritical).
function, must be substituted by
1 R2
u = 0 u0 , (10.39)
2 r r
where R is the radius of the chamber and r is the distance to its axis. Coefcient 1/2
is introduced so that stream function changes from value zero at the axis to value 1
at the chamber wall.
The change of variable

( r )2y
= , (10.40)
R h
transforms (10.39) into (10.18) keeping conditions = 0 at y = 0 and = 1 at
y = 1. The remaining calculations are identical in both cases. Hence, the problem
reduces to the two-dimensional case. Fabri, Siestrunck and Foure have also studied the
case where stabilizer is at the chamber wall. The annular stabilizer has been studied
by Ernst [9]. All these cases reduce to the two-dimensional problem by an adequate
change of variables.
10.5 Chamber with slowly varying cross-section
The preceding method can be applied to an approximate study by substituting equa-

tions (10.18) or (10.39) for an equation that includes the variation of the cross-section
along the chamber axis. The same is done for the approximate study of gas motions
within ducts with slowly varying cross-section.
References
[1] Scurlock, A. C.: Flame Stabilization and Propagation in High-Velocity Gas

Streams. Meteor Report No. 19, Fuels Research Laboratory, M.I.T., May 1948.
[2] Williams, G. C., Hottel, H.C. and Scurlock, A. C.: Flame Stabilization and
Propagation in High Velocity Gas Streams. Third Symposium (International)
on Combustion, Williams and Wilkins Co., Baltimore, 1949, pp. 21-40.
[3] Ball, G.: A Study of a Two-Dimensional Flame. Combustion Tunnel Laboratory
Report, Harvard University, Cambridge Mass., 1951.
[4] Emmons, H. W.: Fundamentals of Gas Dynamics. Section F, vol. III of High
Speed Aerodynamics and Jet Propulsion, Princeton University Press, 1956
[5] Tsien, H. S.: Inuence of Flame Front on the Flow Field. Journal of Applied
Mechanics, June 1951, pp. 188-194.
[6] Fabri, J., Siestrunck, R. and Foure, C.: Sur le Calcul du Champ Aerodynamique
des Flammes Stabilisees. Comtes Rendus des Seances de lAcademie des Sci-
ences, Paris, 1951, p. 1263.
[7] Fabri, J., Siestrunk, R. and Foure, C.: Phenomene dObstruction Intervenant
dans la Stabilisation des Flammes. La Recherche Aeronautique, No. 25, 1952,
pp. 21-27.
[8] Fabri, J., Siestrunk, R. and Foure, C.: On the Aerodynamic Field of Stabi-
lized Flames. Fourth Symposium (International) on Combustion, Williams and
[9] Ernst, G.: Propagation a Faible Vitesse dune Flamme dans un Ecoulement
Compressible. Technique et Science Aeronautiques, 1955, pp. 1-12.
Chapter 11
Similarity in combustion.
Applications
11.1 Introduction
The methods of dimensional analysis and physical similarity have been applied with
very good results to the study, both theoretical and experimental, of classical Aerother-
modynamics. Recently, numerous attempts have been made to extend these methods
to the study of Aerothermochemistry. In some instances, like in the work by Schultz-
Grunow listed in Ref. [1], such an extension was intended in order to facilitate the
establishment of a correlation between several experimental results. Other attempts
tried to nd a rational basis for the experimentation with models as well as rules to
apply the results obtained through such models to the processes at normal scale.
Before treating a problem of this nature, it is necessary to determine the di-

mensionless parameters that characterize the process. In the phenomena whose study
is the purpose of Aerothermochemistry, such parameters are the same studied by clas-
sical Aerothermodynamics, but increased by the chemical parameters introduced by
Damkohler [2] when he applied physical similarity to the study of chemical reactors.
We owe to Penner [3] the rst study on the application of the laws of physical sim-
ilarity to Aerothermochemistry. A concise work on the subject, analyzing the origin
and signicance of these parameters, is own to von Karman and listed in Ref. [4].
Recently, Weller [5] has performed a review of the practical applications. Several ap-
plications of this theory to the study of technological problem of combustion will be
found in Penners work and in the corresponding sections of the Proceedings of Sixth
International Symposium on Combustion and of Second Colloquium on Combustion,
AGARD.
261
262 CHAPTER 11. SIMILARITY IN COMBUSTION. APPLICATIONS
In the following we will rst deduce the dimensionless parameters of Aerother-

mochemistry starting from the fundamental equations governing motion. Thereafter
two practical application will be performed, one to the rule of the scaling of rockets
and another to the problem of ame stabilization.
11.2 Dimensionless parameters of Aerothermochemistry
The physical similarity between two analogous processes consists in the proportional-
ity between corresponding magnitudes (velocities, temperatures, etc.) through space
and time. In the rst place, it implies a geometrical similarity and, moreover, propor-
tionality of times.
The origin of this similarity and the required conditions for it to occur may be
conceived as follows. Let us consider the equations of the process and write them in
dimensionless form, referring each one of the variable quantities comprised in them
(length, pressure, velocity, temperature, etc.) to a characteristic value of the same.
Thus, we shall obtain a system of equations between dimensionless variables, with
coefcients that will also be dimensionless combinations of such characteristic val-
ues. All the combinations of these values for which the coefcients are invariable will
correspond to processes represented by identical systems of equations. If, further-
more, we keep invariable the dimensionless expressions of the initial and boundary
conditions, then the solution of the system thus obtained will correspond to a phys-
ically similar set of processes. We will pass from one another of these processes
by introducing changes into the characteristic values which keep invariable the said
coefcients of the equations and their initial and boundary conditions. Hence, these
coefcients are the dimensionless parameters of the physical similarity. Therefore, the
problem reduces to nding, among them, those independent from one another, and to
select the most simple expressions possible for the same.
Let us see now in what way this can be attained, utilizing, for simplicity and
clearness, the equations corresponding to an one-dimensional stationary ow with
only two chemical species, since the generalization to other cases is straightforward
and it only requires multiplying the number of parameters.
The system of equations that we must apply to this study was deduced in 2 of
Chap. 5, but it is reproduced here in order to assist the reader:
a) Equation of mass conservation.
v = m. (11.1)
11.2. DIMENSIONLESS PARAMETERS OF AEROTHERMOCHEMISTRY 263
b) Continuity equation.
d
v = w. (11.2)
dx
4 dv
p + v 2 = i. (11.3)
3 dx
d) Energy equation.
( )
1 2 dT 4 dv
v v + cp T q v = me. (11.4)
2 dx 3 dx
e) Diffusion equation.
D dY
Y + = 0. (11.5)
v dx
Be l0 , 0 , v0 , w0 , p0 , 0 , cp0 , T0 , 0 and D0 characteristic values of the corre-
sponding quantities and let us change the variables of the preceding system as follows
x = l0 x0 , = 0 0 , v = v0 v 0 , w = w0 w0 , p = p0 p0 ,
(11.6)
= 0 0 , cp = cp0 c0p , T = T0 T 0 , = 0 0 , D = D0 D 0 .
Thus all the prime quantities are dimensionless. For and Y this change is
not necessary since both are already dimensionless and their variation eld has been
selected from zero to one.
When these new variables (11.6) are introduced into the preceding system it
may be written as follows:
a) Equation of mass conservation.

( )
m
0 v 0 = . (11.7)
0 v0
b) Continuity equation. ( )
0 v0 d
0 w0 = w0 . (11.8)
l0 w0 dx
( ) ( ) 0
( )
p0 0 0 02 4 0 0 dv i
p + v = . (11.9)
0 v02 3 0 v0 l0 dx0 0 v02
d) Energy equation.
[ ( 2 ) ( ) ] ( )
1
0 0 v0 02 0 0 q 0 T0 dT 0
v v + cp T 0 0
2 cp0 T0 cp0 T0 l0 0 v0 cp0 T0 dx
( ) 0
( )
4 0 v02 dv me
0 v 0 = . (11.10)
3 l0 0 v0 cp0 T0 dx 0 v0 cp0 T0
e) Diffusion equation.
( )
D0 D0 dY
Y + = 0. (11.11)
v0 l0 v 0 dx
The set of dimensionless coefcients Pi characteristic of the system is
0 v0 p0 0 v02
P1 = , P2 = , P 3 = , P 4 = ,
l0 w0 0 v02 0 v0 l0 cp0 T0
(11.12)
q 0 0 v0 D0
P5 = , P6 = , P7 = , P8 = .
cp0 T0 l0 0 v0 cp0 l0 0 cp0 T0 v0 l0
The physical similarity requires that the values of all these coefcients be equal for
the two processes compared, when using as characteristic values of the variables for
their calculations those at corresponding points and instants.
Such equality guarentees the identity of the left-hand sides of Eqs. (11.7) to
(11.11) in both phenomena. If, furthermore, we guarentee as well the equality of the
right-hand sides, which correspond to the boundary conditions of the problem, we will
have insured the identity of the solution, thus reaching the physical similarity of the
phenomena. Let us proceed to discuss the set of parameter of Eq. (11.12).
We readily observe that P7 is equal to the product of P3 by P4
P7 = P3 P4 , (11.13)
while the remaining parameters are independent from one another. Consequently, the
number of independent dimensionless parameters is seven. Let us see which are these
seven.
It is clear that P3 is reciprocal to the Reynolds number and may be substituted

by it
0 v0 l0
1) Re = . (11.14)
0
The combination of P2 and P4 gives the following parameter P40 which may
substitute P4
p0
P40 = . (11.15)
0 cp0 T0
p0
Yet, the state equation = Rg T0 allows P40 to be written
0
Rg cp0 cv0 1
P40 = = =1 . (11.16)
cp0 cp0
Consequently, the constancy of P40 is equivalent to the constantcy of relation (11.16)
between specic heats, which supplies the second characteristic dimensionless param-
eter
11.2. DIMENSIONLESS PARAMETERS OF AEROTHERMOCHEMISTRY 265
cp0
2) = . (11.17)
cv0
Keeping in mind that velocity a0 of sound in the mixture, at pressure p0 and

with density 0 , is a0 = p0 /0 , we can promptly see that P2 may be written in the
form
1
P2 = , (11.18)
M02
being
v0
3) M0 = , (11.19)
a0
the Mach number of the ow, which will be used as a third characteristic dimension-
less parameter in lieu of P2 .
The combination of P3 and P6 gives the following value

P3 0 cp0
4) Pr = = , (11.20)
P6 0
which is the Prandtl number, fourth dimensionless parameter the substitutes P6 .
Likewise, the combination of P3 and P8 gives

P3 0
5) Sc = = , (11.21)
P8 0 D0
which is the Schmidt number, fth dimensionlees parameter in lieu of P8 .
The aforegoing parameters are the same used in classical Aerothermodynamics

and the characteristic constants of the chemical reaction have no bearing on them. This
constants act through two other parameter, P1 and P5 . The rst may be written
v0 ch0
P1 = , (11.22)
l0
where
0
ch0 = (11.23)
w0
is a characteristic chemical time. The reciprocal of (11.22) is the rst parameter of
Damkohler,
l0
6) Da1 = , (11.24)
v0 ch0
and it is the sixth dimensionless parameter of the process.
Finally, P5 is the second parameter of Damkohler

q
7) Da2 = , (11.25)
cp0 T0
this being the seventh dimensionless parameter of the process.
The interpretation of the meaning of Damkohler two parameters is simple. The

rst is a measurement of the relationship between the mean mechanical time required
by a particle to travel the characteristic lenght l0 at the characteristic velocity v0 and
the time required by the chemical reaction of the mixture to take place. The second
parameter of Damkohler is a measure of the relationship between the heat produced
by unity of mixture when burning at constant pressure and the heat contained by the
same at characteristic temperature T0 .
It should be noted that in the non-stationary processes a additional parame-

ter appears, Strouhal number, which originates from the time derivatives /t of the
equations of motions. Likewise, if the mass forces are present as occurs in the phe-
nomena of free convection, then another dimensionless parameter appears, due to the
gravity terms of the equations, which originates the Froude number.
Finally, when the number of species and chemical reaction is increased, the
number of the Damkohler parameters (of both kinds time and heat) increases as well.
In technical literature we frequently nd other parameter which are combina-

tions of the above mentioned. Por example, occasionally Prandtl number is substituted
by Peclet number, which is a product of the numbers of Reynolds and Prandtl
0 v0 l0 cp0
Pe = Pr Re = . (11.26)
0
Likewise, the Schmidt number is replaced by the Lewis-Semenov number which is
obtained as follows
Pr 0 D0 cp0
L= = , (11.27)
Sc 0
used specially in combustion processes as seen in Chap. 6.
The advantage of utilizing the Peclet and Lewis-Semenov numbers in combus-

tion phenomena is based on the fact that the conductivity and diffusion coefcients
appear explicity in their expressions. Such coefcients are the ones of interest in these
processes in which the inuence of the viscosity coefcient is negligible in many
cases, as said in Chap. 5.
We shall now see some practical applications of this theory.
11.3 Scaling of rockets
The problem of the scaling of rockets has a great practical interest since its solutions
would allow the prediction of the behavior of the rockets after tests made with reduced
scale models.
11.3. SCALING OF ROCKETS 267
Furthermore, both the scaling under steady operation as well as under low and
high frequency oscillations are important in practice. This problem was recently stud-
ied by Penner-Tsien [6], Rose [7], Crocco [8], Barrere [9] and Penner-Fuhs [10].
We have seen in 2 that physical similarity imposses that the seven characteris-
tic parameters be equal, when passing from the model to the rocket under steady state
conditions and some additional ones under non-steady conditions.
We shall now study the scaling rules imposed by these equalities. If we assume
that we are working with the same mixture of gases and at the same temperature, the
equality of , as well as that of the Prandtl and Schmidt numbers and of Da2 , is in-
sured. Consequently we only have to insure the equality of the other three parameters.
Disregarding subscript zero and using subscripts 1 for the model and 2 for the rocket,
we will have:
a) Equality of the Reynolds Number. The equality of temperatures makes 1 = 2 .

Furthermore, may be substituted by p, to which it is proportional. Therefore,
this condition reduces to the following
p1 v1 l1 = p2 v2 l2 . (11.28)
b) Equality of the Mach Number. The equality of temperature insures the equality
of the velocity of sound a1 = a2 . Hence, this condition reduces to the following
v1 = v2 , (11.29)
this is to say that the velocity must be the same in both the model and rocket.
c) Equality of Da1 . This condition gives
l1 l2
= . (11.30)
v1 ch1 v2 ch2
Since l is the length of the combustion chamber and v the gas velocity through it,
relation
l
r = (11.31)
v
is the residence time of the gases in the chamber. ch is a physico-chemical time, char-
acteristic of the process, the so called time lag or delay from the instant at which the
fuel enters the chamber to the instant at which it transforms into products. Therefore,
it accounts for the time needed by the fuel to form droplets, to evaporate, to mix with
the gases and burn. Hence, Da1 can be written in the form
r
Da1 = . (11.32)
ch
Let
l1
K= (11.33)
l2
be the linear relation of sizes. Then from (11.28), (11.29), (11.30) and (11.33) one
derives
p1 1
= , (11.34)
p2 K
ch1
= K. (11.35)
ch2
The rst condition imposses that the variation of pressure be inversely propor-
tional to size. The second one cannot be insured since in a rocket, as we have seen,
ch is the resultant of a complicated physico-chemical process. Crocco, by means of
a detailed analysis [11], has reached the conclusion that ch may be represented, in
many cases, through a decreasing function of pressure in the form
ch pn , (11.36)
where n is an exponent differing slightly from unity.
If n = 1, by taking (11.36) into (11.34) and comparing with (11.35), we see

that the later is satised, thus obtaining a complete physical similarity.
If n is different from unity the complete similarity is not possible. In such case,
if we disregard the condition of equality of the Mach numbers, which is actually not
too important under steady-state conditions, since velocities at the combustion cham-
ber are very small and compressibility effects can be neglected, the following condi-
tions are obtained, instead of (11.29). By eliminating v1 and v2 between Eqs. (11.28)
and (11.26), which remain valid, and taking into account Eq. (11.33), it results
p1 ch1 2
= K . (11.37)
p2 ch2
If expression (11.36) is assumed for the time lag, the following condition is obtained
for the ratio of pressures
2
p1
= K 1+n, (11.38)
p2
whereas, from here and Eqs. (11.28) and (11.33), it results for the ratio at velocities
1n
v1
= K1+n. (11.39)
v2
Penner and Tsien [6] adopt a different view point. They also neglect the equal-
ity of Mach numbers, but assume that both rockets operate at the same pressure
p1 = p2 . (11.40)
11.3. SCALING OF ROCKETS 269
From here and Eq. (11.37), one obtains

ch1
= K 2. (11.41)
ch2
The authors propose that these condition be satised by proper control of the
time lag, which may be reached, for instance, if the mean size of the droplets is con-
veniently varied, since the evaporation time of a droplet depends on its size as will be
shown in chapter 13.
Now let us see the scaling rules for the thrust and injector.
Physical similarity implies that the number of injectors be the same in both
rockets and identically distributed.
Be e the thrust, g the ow rate of fuel, v1 its velocity through the injector and
d its diameter. It is promptly veried that the following system of relations is valid
e g vi d2 vl2 pvl2 , (11.42)
provided the temperature is preserved.
From here, the following system of scaling conditions is derived

e1 g1 vi1 d21 p1 v1 l12
= = 2 = . (11.43)
e2 g2 vi2 d2 p2 v2 l22
Moreover, since the velocity elds must be similar, it results
vi1 v1
= . (11.44)
vi2 v2
When Eqs. (11.28) and (11.33), which in all cases remain valid, are taken into
Eq. (11.43), one obtains for the ratio of thrusts
e1
= K. (11.45)
e2
The combination of this relation with Eqs. (11.43) and (11.44) gives for the ratio of
injector diameters
d1 v2
= K . (11.46)
d2 v1
Hence, the ratio of diameters depends on the scaling law for the velocities at
the chamber, which, as we have seen, depends on the rule applied. For example, for
the Penner-Tsien rule it results, by virtue of Eq. (11.28),
d1
= K, (11.47)
d2
whereas for Croccos rule, from Eq. (11.39) it is obtained
n
d1
= K1+n. (11.48)
d2
11.4 Scaling of rockets for non-steady conditions

The motion on gases through the combustion chamber of a rocket is always strongly
turbulent with oscillations of pressure, velocities, etc., distributed at random in time
and space, which does not cause important disturbances in the normal operation of the
rocket. Under these circumstances the combustion is called steady.
However, there is also the case, frequently observed, where combustion is un-
stable with strong self-excited periodic oscillations of pressure which in some case
destroy the rocket in a few seconds. This phenomenon is at present one of the main
difculties for the development of rockets, specially large ones. It was observed for
the rst time in the United States in 1941 by the research staff of the Jet Propulsion
Laboratory of the California Institute of Technology under the guidance of Profes-
sor von Karman. Thereafter the great effort applied to the study of this problem has
consderably increased the technical literature on the subject.
Ross-Datner [11] and Crocco-Gray-Grey [12] have written two excellent re-
views on this problem describing the kinds of instabilities observed, their causes, ex-
perimental techniques and the results of the measurements performed, in addition to
the fundamentals of the theories available. These reviews include as well an exten-
sive bibliography. The most up to date and complete work on the subject is the one
performed by Crocco and Cheng [13] which was recently published by AGARD, in-
cluding an abundant bibliography.
Initially, due to limitations of the instruments available for observation, it was

only possible to isolate one type of low-frequency oscillations of the order of about
100 cycles per second, which is normally known as chugging. Later on, as the instru-
mentation was improved, it became feasible to isolate other high-frequency oscilla-
tions, over 1000 c.p.s., generally called screaming.
Soon, the origine of the low-frequency oscillations was known and it was pos-
sible to derive practical rules to prevent them which resulted efcient when applied to
practice. Summereld [14] stated the fundamentals of the theory and the preventive
rules born from it. In low-frequency oscillations, the combustion chamber behaves
like a resonating cavity whose oscillations of pressure act on the feeding system,
changing the rate of fuel injected. The inuence of this variation reects on the cham-
ber with a certain delay which is due partially to the relaxation times of the chamber
and of the feeding system, and partially to the physico-chemical time lag described
in the preceding paragraph. If this delay is the adequate one, the oscillations result
self-excited. Consequently, the unit combustion chamber-feeding system behaves as a
dynamic system with a characteristic time lag, able of producing unstable oscillations.
11.4. SCALING OF ROCKETS FOR NON-STEADY CONDITIONS 271
0 Chamber pressure t
2
Injection pressure drop t
4
i
6 Injection rate t
ch
8
Burning rate t
10
c
12 Effect of burning rate t

Half period
14
Figure 11.1: Time delays leading to self-exciting oscillationes.
Figure 11.1, taken from Ref. [13], clearly illustrates the mechanisn of the pro-
cess capable of producing self-excited oscillations. In this gure, it is assumed that the
feeding system is of the constant pressure type. Therefore, the pressure drop through
the injector, and with it the rate of fuel, decreases as the pressure of the chamber is
increased. However, the variation in rate of fuel follows the drop in pressure with a
certain delay i , determined by the dynamic characteristics of the feeding system. The
fuel injected turns into combustion products with a delay ch , which is the time lag,
and such products act in turn on the pressure of the chamber with a delay c , which
depends on the dynamic characteristics of the same. If the sum i + ch + c is equal
to half the period of the pressure oscillations, then there is a phase agreement between
cause an effect and the oscillations amplify with the only limitation of damping ef-
fects. The above analysis shows that in order to eliminate oscillations it is necessary
to reduce i , ch and c . The measures taken to this effect conrm theoretical predic-
tion. Two efcient ways, for example, are to reduce the relation between pressure drop
through the injector and the pressure in the chamber, or to reduce ch by eliminating
the propellants having an excessive time lag [15].
Aside from the aforegoin causes, Crocco [15] has pointed out an intrinsic cause
for instability, so called because it depends only on the conditions in the combustion
chamber and not on the interaction between it and the feeding system, indispensable,
as it was established, for the previously analyzed case. The intrinsic instability of
Crocco shows the possible existence of low-frequency oscillations in a rocket with a
constant volume feeding system, which is impossible with the preceding model.
We have said in the preceding paragraph of this chapter, that Crocco has proven
that time lag ch is not constant but a function of the conditions in the chamber. In
his study he schematizes this by making ch depending on the pressure, as shown in
Eq. (11.36). In intrinsic instability the variation of ch with p plays the same part than
the consumption of fuel in Figure 11.1, since the element acting on the pressure of
the chamber is not the fuel rate injected into it, but the rate of fuel transformed into
products, which may vary from one instant to another, even if the former is constant,
when ch varies with pressure. Under such conditions, if a coupling of phases is
reached we will have a self-excited oscillation as in the preceding case.
Let us now see the rules obtained for the scaling of rockets, to maintain physical
similarity, so that both rockets be equaly stable with respect to the low-frequency
oscillations.
In such case, aside from the conditions derived in the preceding paragraph, it
is necessary that some additional ones be satised relative to feeding system, since its
inuence is important.
In particular, we must keep the equality of the ratio of the pressure drop p
through the injector to the pressure in the chamber, that is
p1 p2
= . (11.49)
p1 p2
From here, one obtains the following scaling law for the pressure drop through
the injector
p1 p1
= . (11.50)
p2 p2
However, we must keep in mind that the pressure drop through the injector
determines velocity vi of the fuel when passing through the same, and that the ratio
of such velocities has been determined before by the condition of physical similarity
between velocity elds in Eq. (11.44). Consequently we cannot insure the simultane-
ous satisfaction of conditions (11.50) and (11.44), except for certain particular cases
or by adopting special precautions. For instance, if the injectors have similar friction
characteristics, then the following ratio would exist between p and vi
p vi2 , (11.51)
and we have ( )2
p1 vi1
= , (11.52)
p2 vi2
which is not compatible with Eq. (11.50), except for n = 2, as it can readily be
veried.
11.4. SCALING OF ROCKETS FOR NON-STEADY CONDITIONS 273
The similarity conditions that still need to be satised refer to the dynamic
characteristics of the feeding system which determine the value for i in Fig. 11.1.
For their study, we refer the reader to the papers by Crocco [8] and Penner-Fuhs [10]
previously mentioned.
Low-frequency oscillations may be analyzed by adopting the assumption that

the state of the gases in the chamber is constant, because their period is very long com-
pared to the propagation time of a wave through the chamber. To the contrary, when
both the period of the oscillations and the propagation time of the wave are of the
same order of magnitude, it is necessary to consider the differences in state between
different points of the chamber. Such is the situation in the case of high-frequency
oscillations, where, as shown by experimentation, the chamber oscillates like an or-
gan pipe with longitudinal and transversal oscillations. The existence of a time lag
makes possible the maintenance of these oscillations in a similar way as it happens
for low-frequency oscillations. High-frequency oscillations are particularly danger-
ous because, in addition to pressure oscillations being very large, the transmision of
heat to the walls or injectors increases drastically and they are destroyed within a few
seconds. The study of these oscillations is less advanced than for the low-frequency
ones which are easier to control.
As for the scaling law, it appears evident that the residence time r should be
substituted by propagation time p of a pressure wave through the chamber. Therefore,
in order to mantain the same level of stability for high-frequency oscillation when
passing from the model to the rocket, the following condition must be satised
ch1 p1
= . (11.53)
ch2 p2
However, since the velocity of sound is the same for both rockets, between p1 and
p2 the following ratio exists
p1
= K. (11.54)
p2
Consequently we obtain
ch1
= K. (11.55)
ch2
It happens also as for low-frequency oscillations, that this condition can not always be
satised. In particular, it can be veried that for n = 1 and applying Croccos rule the
above condition may be attained.
11.5 Flame stabilization
In the preceding chapter we have studied the aerodynamic eld of a ow with a ame
stabilized by a holder. We veried that the expansion of the burnt gases may originate
a chocking effect which imposses a limitation to the maximum velocity of the ow
entering the chamber.
Another limitation to this velocity, bearing a great practical interest, is that im-
possed by the need to x the ame to the holder in order to maintain combustion.
In fact, it has been experimentally observed that when the velocity of the incoming
ow exceeds a certain value, the ame blows out. The blowing velocity of the ame
is determined by the composition and state of the combustible mixture and by the
characteristics of the holder. The problem arising from this circunstance holds a great
practical interest in the new propulsion systems, specially in ramjets and after-burners,
in which their limits of practical application are often determined by this effect. Con-
sequently, this phenomenon has been the subject of a particular interest during the last
years and abundant experimental data have been gathered on a great variety of types of
holder and on the inuence of the parameters of the mixture on the blowing velocity
of the ame. An extensive bibliography on the matter can be found in the proceedings
of the congresses and colloquia on combustion celebrated in recent years.
The initial study on this problem was carried out by Scurlock [16], who pointed
out the signicance of the recirculation zone that forms in the wake of the holder. Fig-
ure 11.2 taken from Ref. [17] illustrates the phenomenon. Behind the holder a recir-
culation zone forms in which the ame stabilizes. The boundary of this zone consists
in a mixing zone formed by a free boundary layer which separates the unburnt gases
from the combustion products. Within this mixing zone is where the combustible
gases ignite through a process of heat and mass transport, whose characteristics are
not yet well known.
Since the presence of a recirculation zone is essential for the existence of the
ame, the holders used for this purpose are given a shape which produces a large
aerodynamic wake.
In many of the tests conducted, attempts were made to establish a correlation

between the blowing velocity of the ame, in a mixture of given composition and
temperature, and the size of the holder, characterized by a linear dimension of the
same, for example by its diameter in the case of a holder of circular section as the one
shown in Fig. 11.2. A summary of the works performed from this stand-point, up to
1953, and of the state or knowledge at the time may be found in Ref. [18].
11.5. FLAME STABILIZATION 275
Flame front End of recirculation zone

Recirculation zone
Flame holder Mixing zone Propagating zone
Figure 11.2: Zones of a stabilized ame.
The theoretical study of this problem is very difcult. Several theories are
available whose development may be found in the works listed under Refs. [19]
through [21], but none of them has been experimentally conrmed. On the other
hand, the correlation between experimental results as presented by Longwell is very
difcult. At this stage it appears justied to perform the study by means of a di-
mensionless analysis, searching for the signicant variables and the relation existing
between them.
Inasmuch as we are far from the actual regime in which the compressibility
effect would appear, the dimensionless variable characteristic of the motion is the
Reynolds number referred to the velocity and state of the gases before the holder and
to a linear dimension of the same. The composition of the mixture is characterized by
the relationship between the fuel fraction and that corresponding to the stoichiometric
one. Finally, the variable we are trying to analyze is the blowing velocity or, if written
in dimensionless form although it is not generally necesary, this velocity referred to
the propagation velocity of the ame.
Zukoski and Marble [22] have performed a very interesting systematic analysis
on the experimental data available. A summary of this analysis appears in Figs. 11.3
and 11.4
Figure 11.3, corresponding to a circular ame holder as the one shown in

Fig. 11.2, gives the following: a) the inuence of the Reynolds number on the maxi-
mum blowing velocity of the ame, and b) the composition of the mixture for which
such velocity is reached, for a given Reynolds number.
Figure 11.4 shows the law of variation of the maximum blowing velocity as a
function of the Reynolds number for a set of experiments performed by other authors
with different types of ame holders.
2.0
1.5
m
1.0
600
Maximum blowoff
velocity (ft/s)
400
200
100 5
103 2 4 6 8 10 4 2 4 6 8 10
Reynolds number
Figure 11.3: Maximum blowoff velocity and corresponding equivalence ratio for a cylin-
drical ame holder.
Maximum blowoff velocity (ft/s)
1000
800
600
400
300
200
150
100 4 5
2.5 4 6 8 10 1.5 2 3 4 6 8 10 1.5
Reynolds number
Figure 11.4: Maximum blowoff velocity for different types of ame holders (see [22] for
more details).
Both gures show clearly the existence of a transition Reynolds number, ap-
proximately equal to 104 , which separates two different regions. For Re < 104 , the
maximum blowing velocity corresponds to a composition different from the stoichio-
metric one, which furthermore varies with the Reynolds number, whilst if Re exceeds
the transition number the composition remains constant. Likewise, for Re > 104
the blowing velocity varies with the square root of the Reynolds number, while for
Re < 104 it follows a different law.
Zukoski and Marble, Ref. [17], have demonstrated that such a change is due to
the fact that the free boundary layer of the mixing zone changes from laminar to turbu-
lent. This transition is important because the transfer of heat and chemical species be-
tween unburnt gases and combustion products, within the mixing zone, occurs through
laminar diffusion under subcritical conditions and through turbulence in the opposite
case.
Flame holder D i
geometry
D 1/8 inch 3.09 104 s

3/16 2.85
1/4 2.80
D
1/4 3.00
mesh screen
1/4 2.38
D
3/8 2.70
1/2 2.65
3/4 2.58
1/4 3.46
D
3/8 3.12
1/2 3.05
3/4 3.03
D
3/4 3.05
D
3/4 2.70
Table 11.1: Ignition time for different types and dimensions of ame holders.
Since the existence of the ame is governed by the possibility of igniting the
mixture, while it travels along the mixing zone, the residence time r of the mixture
within this zone must be an important factor of the phenomenon. The residence time
may be expressed by the ratio
L
r = (11.56)
V
of length L of the mixing zone to the velocity V of the gases when travelling along
it, which is approximately the velocity of the free ow. Consequently, if this inter-
pretation of the phenomenon is correct, the blowing velocity may be reached when r
reaches a limiting value i , below which there is no time for ignition. Furthermore,
i must depend exclusively on the composition and state of the mixture and on the
conditions at the free boundary layer, but not on the shape of the holder or any other
mechanical factor.
In order to check the validity of this assumption, Marble and Zukoski computed
the value of i for a set of quite different cases. Results are summarized in Table
11.1 in which the constancy of the ignition time may be clearly seen. Hence, the
characteristic dimensionless group determining the blowing velocity of the ame is
Damkohlers rst parameter
L/V
, Da1 = (11.57)
i
and blowing velocity is reached for Da1 = 1.
The advantage of this study lies on the fact that it separates two variables. One,
being mechanical, depends only on the conditions of motion, whereas the other, being
physico-chemical, depends exclusively on the composition and state of the mixture.
Fig. 11.5, from Ref. [17], shows the law of variation for i as a function of the mixture
composition for a typical case.
1.8
1.6
Characteristic ignition time (ms)
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
, Equivalence ratio
Figure 11.5: Ignition time as a function of the composition of the mixture (see [17] for more
details).
With reference to the mechanical time, the problem reduces to a study of the
relationship between the length L of the wake and the characteristics of the holder.
For a great number of holders it has been proved that, in the supercritical regime, this
relation is of the form
L D1/2 , (11.58)
which explains the law of variation of V with D observed by experimentation.
There exist, however, other cases for which the following condition is satised
L D, (11.59)
which is the reason for the special behavior of the holders observed by Longwell [18].
In this formulae, D is a linear dimension of the holder.
The problem which remains to be solved is the process in the mixing zone
which determines the value for i .
References
[1] Schultz-Grunow, F.: Similarity Laws of Deagration. Fourth Symposium (Inter-

national) on Combustion, Williams and Wilkins Co., Baltimore, 1953, pp. 439-
443.
[2] Damkohler, G.: Einusse der Stromung, Diffusion and des Warmeliberganges
auf die Leistung von Reaktionsofen. Z. Elektrochem., Vol. 42, 1936, pp. 846-
862.
[3] Penner, S. S.: Similarity Analysis for Chemical Reactors and the Scaling of
Liquid Fuel Rocket Engines. Combustion Researches and Reviews, AGARD,
1955.
[4] von Karman, Th.: Dimensionslose Grosen in Grenzgebieten der Aerodynamik.
Z. F. W., Jan-Febr. 1956, pp. 3-5.
[5] Weller, A. E.: Similarities in Combustion, a Review. Selected Combustion Prob-
lems, Vol. II, AGARD, 1956, pp. 371-383.
[6] Penner, S. S.: Models in Aerothermochemistry. International Symposium on
Models in Engineering, Venice, Italy, 1955.
[7] Ross, Ch. C.: Scaling of Liquid Fuel Rocket Combustion Chambers. Selected
Combustion Problems, Vol. II, AGARD, 1956, pp. 444-456.
[8] Crooco, L.: Considerations on the Problem of Scaling Rocket Motors. Selected
Combustion Problems, Vol. II, AGARD, 1956, pp. 457-468.
[9] Barrere, M.: Similarity of Liquid Fuel Rocket Combustion Chambers. ONERA,
Paris, 1956.
[10] Penner, S. S. and Fuhs, A. E.: On Generalized Scaling Procedures for Liquid-
Fuel Rocket Engines. Combustion and Flame, Vol. I, June 1957, pp. 229-240.
[11] Ross, Ch. C. and Datner, P. P.: Combustion Instability in Liquid Propellant
Rocket Motors. A Survey. Selected Combustion Problems, Vol I, AGARD,
1954, pp. 352-380.
[12] Crocco, L. and Grey, J.: Combustion Instability in Liquid Propellant Rocket
Motors. Proceedings of the Gas Dynamic Symposium on Aerothermochem-
istry, Northwestern Univ., Evanston Illinois, 1956.
[13] Crocco, L. and Cheng, S. I.: Theory of Combustion Instability in Liquid Propel-
lant Rocket Motors. Butterworths Scientic Publications, 1956.
[14] Summereld, M.: A Theory of Unstable Combustion in Liquid Propellant Rocket
Systems. Journal of the American Rocket Society, Sept. 1951, pp. 108-114.
[15] Crocco, L.: Aspects of Combustion Stability in Liquid Propellant Rooket Mo-
tors. Journal of the American Rocket Society, Nov. 1951, pp. 163-178.
[16] Scurlock, A. C.: Flame Stabilization and Propagation in High Velocity Gas
Streams. Meteor Report No. 19, Mass. Inst. Tech., May 1948.
[17] Zukoski, E. E. and Marble, F. E.: Experiments Concerning the Mechanism of
Flame Blowoff from Bluff Bodies. Proceedings of the Gas Dynamic Symposium
on Aerothermochemistry, Northwestern Univ., Evanston Illinois, 1956.
[18] Longwell, J. P.: Flame Stabilization by Bluff Bodies and Turbulent Flame in
Ducts. Fourth Symposium (International) on Combustion, Williams and Wilkins
Co., Baltimore, 1953, pp. 90-97.
[19] Longwell, J. P., Frost, E.E. and Weis, M.A.: Flame Stability in Bluff Body Re-
circulation Zones. Industrial and Engineering Chemistry, Vol. 45, August 1953,
pp. 1629-1633.
[20] Spalding, D. B.: Theoretical Aspects of Flame Stabilization. Aircarft Engineer-
ing, Sep. 1953, pp. 264-268,276.
[21] Lees, L.: Fluid Mechanical Aspects of Plane Stabilization. Jet Propulsion, July-
August 1954, pp. 234-236.
[22] Zukoski, E. E. and Marble, F. E.: The Role of Wake Transition in the Process
of Flame Stabilization on Bluff Bodies. Combustion Researches and Reviews,
AGARD, 1955, pp. 167-180.
Chapter 12
Diffusion ames
12.1 Introduction
So far we have considered combustion problems dealing with premixed reactant gases.
Yet, it can also happen that species are initially separated, and mixing and combustion
take place simultaneously. Thus, a new type of ame is obtained which Burke and
Schumann [1] called diffusion ame. In these ames the reaction zone separates the
two reacting species which diffuse through inert gases and combustion products from
each side towards the ame. A ame of this type is obtained, for example, when a jet
of combustible gas discharges from a burner into the open atmosphere or into an air
stream parallel to the jet burning at the outlet of the burner, see Fig. 12.1. Such ames,
are largely used in industrial applications. The ame length needed to burn a given
Flame
Air Air
r
o
Fuel
Figure 12.1: An open diffusion ame.
281
282 CHAPTER 12. DIFFUSION FLAME
quantity of fuel per unit time under given conditions is highly interesting in these ap-
plications, since the dimensions of burners and furnaces depend on it. The reacting
species burn very rapidly as they reach the reaction zone. Thereby, the combustion
velocity is generally conditioned by the accessibility of the species to the ame. In
other words, by their facility to diffuse across the inert gases and combustion prod-
ucts. Diffusion can be either laminar or turbulent depending on the conditions of the
phenomenon, originating the ames accordingly.
The actual state of knowledge on diffusion ames is by far less advanced than
on premixed ames The fundamental work by Burke and Schumann [1] has been the
starting point for practically all the later studies. Even though considerably ahead the
publication of this work a correct qualitative idea on diffusion ames was held,1 Burke
and Schumann were the rst to achieve a quantitative study of the phenomenon. For
the purpose they used a highly simplied model that enabled an analytical solution
of the problem. This model retains the essential factors of the process and can be
summarized in the following two assumptions:
1) The thickness of the reaction zone is zero.

2) The reacting species burn instantaneously upon reach the ame.
This simple model eliminates chemical kinetics from the process thus making
available its computation. This simplication is justied when the reaction rate is very
large compared to the diffusion velocities of the species towards the ame. Thereby it
is not applicable, for example, to the study of ames rareed or very diluted by inert
gases where the time of reaction can appreciably inuence the process. The model
disregards the structure of the reaction zone as was done in Chap. 6 for premixed
ames. The difference between both cases lies upon the fact that while in the case
of premixed ames the combustion front is a discontinuity surface for velocity, tem-
perature, composition, etc., in diffusion ames such magnitudes change continuously
across the front.
The rst assumption reduces the reaction zone to a surface called the ame
front, which acts as a sink for reacting species and as a source for combustion products.
The second assumption implies the following consequences:
a) Concentrations of reactants at the ame must be zero.

b) Reactants must diffuse towards the ame in the stoichiometric ratio correspond-
ing to the produced reactions.
1 See, i.e., Barrs review paper [2] where data and bibliography on the subject can be found.
In fact, otherwise, some of the species would diffuse across the ame and react with
those on the other side.
In order to judge the validity of this model Hottel and Hawthorne [3] performed
some experiments to determine the composition of the gases in a diffusion ame of
hydrogen burning in the open atmosphere, taking samples of the gas at different points
of the ame by means of a special sound. Details of these experiments can be found
in the said work. Figure 12.2, taken from it, shows the results of the sounding made
in three cross sections of the ame. The results of such measurements prove the
correctness of Burke and Schumanns model.
100
N2
75
flame front
50 distance from port= 30.5 cm

H2
25
O2
0
Percentage in dry sample
100
N2
75
H2
flame front
25
0
O2
axis of flame
100
H2 N2
75
flame front
25
O2
0
0 10 20 30 40
Radial distance (mm)
Figure 12.2: Gas composition in a hydrogen diffusion ame
Burke and Schumann have successfully applied their model to the calculation
of the shape and height of laminar diffusion ames for the case where the fuel jet
discharges within a tube where an air stream moves with the same velocity than the
fuel jet, Fig. 12.3. This device eliminates the difculties originating from momentum
transfer between the fuel and the surrounding air when their velocities are not equal.
Furthermore, for this reason it is easier to obtain a stable ame. By introducing several
drastic simplications, to be considered further on, Burke and Schumann were able
to calculate the shape of the ame. Some of their results are outlined in Fig. 12.3.
Fig. 12.3 (a) corresponds to an overventilated ame, which closes at the axis of the
air fuel air air fuel air
(a) overventilated (b) underventilated
Figure 12.3: Conned laminar diffusion ames.
tube, and Fig. 12.3 (b) to an underventilated ame, which ends at the wall of the
exterior tube. The fuel-air ratio can be changed by varying the diameters ratio of both
tubes. Both types of ames were experimentally observed by Burke and Schumann.
The ame length is proportional to the amount of fuel burnt per second and does not
depend on the diameter of the burner.
Burke and Schumanns analysis on conned ames has been extended by J.

Barr, [4] and [5], to the case where air and fuel velocities are different. He has ex-
perimentally analyzed the appearance of the ames that form when the streams of air
and fuel change within limits far more extended than those covered by any previous
investigation. As a result he concludes that the length of a laminar diffusion ame is
proportional to the fuel consumption not only for the case of short ames, as stated
by Lewis and von Elbe [6], but also for ames of a length as much as a hundred times
the diameter of the burner. The results of his work are summarized in Fig. 12.4, taken
from Ref. [5], which shows the different types of ames observed when the ow rates
of air and fuel are changed. In the same gure, regions 1 and 2 correspond to ames
with excess of air and fuel respectively, of the type studied by Burke and Schumann.
In between these regions a zone of carbon formation exists. Region 3 corresponds to
meniscus ames for which diffusion in the axial direction is important. A meniscus
ame blows-out when the fuel ow rate is reduced. Region 4 corresponds to the so-
called convective or Lambent ames. Within this region, where the fuel and air ratios
are small, buoyancy forces are important and oscillating ames are obtained. Region
5, where fuel ow rate is very large, corresponds to tilted ames preceding blow-off.
Region 6, where the fuel and air streams are very strong, corresponds to lifted ames
which preceded blow-off. Region 7 corresponds to vortex ames. Finally, under given
conditions, ames similar to the manometric and singing ames were also observed
by Barr. This incomplete enumeration of the types of ames observed gives idea on
the complexity of the phenomenon. Further data can be found in Ref. [4].
100
Extinction
2
6
10
Smoke
Lifted flames
Extinction
Butane flow (cm 3 /s)
1
1 int
po Laminar
e
ok diffusion
Sm
flames
5
7
4
0.1
Meniscus flames Vortex flames

3
Convective flames Extinction
0.01
1 10 100 1000 10000
3
Air flow (cm /s)
Figure 12.4: Flow limits for formation of enclosed diffusion ames.
The case of an open diffusion ame where the fuel jet discharges and burns in
an atmosphere at rest has been experimentally analyzed, among others, by Rembert
and Haslam [7] , Cuthbertson [8] , Gaunce [9], Hottel and Hawthorne [3], Wohl,
Gazley and Kapp [10], Barr and Mullins [11], Parker and Wolfhard [12] and Yagi and
Saji [13]. The fundamental results of their experiments are as follows:
1) When the Reynolds number of the jet at the mouth of a burner is small, a laminar
ame establishes.
2) The length of the ame increases when the ow rate of the jet is increased.
3) When Reynolds number is larger than a given value a turbulent region initiates
at the ame tip. When Reynolds number of the jet is increased this region prop-
agates towards the ame base.
4) When the tip of the ame becomes turbulent its length decreases as Reynolds
number is increased until turbulence reaches its base. There on it remains con-
stant independently from the discharge velocity of the jet.
5) Within the laminar region the length of the ame is independent from the burners
diameter and for a given fuel it depends only on its ow rate. Such result agrees
with that obtained by Burke and Schumann for conned ames.
6) Within the turbulent region the length of the ame is proportional to the diameter
of the tube and independent from the fuel velocity.
Diffusion flames Transition region Fully developed turbulent flames
Envelope of flame heights

Height
Increasing nozzle velocity Envelope of

break points
Figure 12.5: Change in ame type with increasing nozzle velocity.
Fig. 12.5 taken from the work by Hottel and Hawthorne summarizes the experiments
performed by these authors. This gure shows the law of variation of the height of
the ame with the velocity of the jet, and the extension of the turbulent zone. The
following states are observed:
a) Laminar. When the velocity of the jet is small. Here the length of the ame
increases, at rst proportionally to the jet velocity and then slows down up to the
maximum height that limits the laminar zone.
b) Transition. The turbulent zone of the ame initiates close to the tip and when
the velocity of the jet increases it propagates towards its base. The total length
of the ame decreases when the extension of the turbulent zone increases. This
is such because the diffusivity of the reactant gases is much larger in turbulent
than in laminar state.
c) Turbulent. The whole ame is turbulent throughout its extension except within a
small zone close to the mouth of the burner. In this state the length of the ame
remains appreciably constant as the velocity of the jet increases. Fig. 12.6 gives
some of the results of the measurements done by Gaunce [9] with city gas in a 3
mm diameter burner. The three aforementioned zones are clearly shown in this
gure.
24
20
Distance from nozzle (inch)
Total length, onport flames
Flame separation begins here
Upper part of flame blowsoff

16
12
Break point length

4
0
0 50 100 150 200 250 300
Nozzle velocity (ft/s)
Figure 12.6: Effect of nozzle velocity on ame length.
Hottel and Hawthorne [3], Wohl, Gazley and Kapp [10], Yagi and Saji [13]
and Barr [5] have attempted an extension of Burke and Schumanns method to the
prediction of the length of open ames both laminar and turbulent. Through rudi-
mentary approximations they obtain an expression for the ame length containing an
unknown function which they determine empirically from the results of their experi-
ments. Fig. 12.7 taken from Ref. [5] shows the theoretical and experimental lengths
of some laminar open ames as functions of the fuel ow rate. This gure gives an
idea on the agreement to be expected between experimental measurements and those
predicted by semi-empirical formulae. The extrapolation to values not included in
this gure is risky. A summary on the state of knowledge regarding this matter can
be found in the work by Hottel listed in Ref. [14] where additional bibliography is
included.
Lately, J. A. Fay [15] has calculated the shape and characteristics of the laminar
diffusion ame obtained when a fuel jet discharges into the open atmosphere for the
two-dimensional case and for the case with axial symmetry. Fay has taken into account
the inuence of the variation of the velocity in cross direction to the jet, computing
the cross distributions of the velocities, concentrations and temperatures. The model
120
A
Butane tube
B
100
Flame length (cm)
80
D
City gas nozzle
C
60
E
B
City gas tube
40
Theory Experiments
A & E: Wohl et al. Butane, Wohl et al.
20
B, D & F: Barr City gas, Wohl et al.
C: Hottel et al. City gas, Rembert & Haslem
Gaunce
0
0 50 100 150 200 250
3
Fuel flow (cm /s)
Figure 12.7: Length of different open ames.
proposed by Burke and Schumann is also used by Fay by reducing the reaction zone
to a surface. This work represents the rst serious attempt to analyze the velocity eld
induced by a diffusion ame.
So far no attempt has been made for the study of the inuence of free convec-
tion on the phenomenon which can be very important specially when the discharge
velocity of the jet is small.
The structure of the reaction zone of a diffusion ame has experimentally been
analyzed by Wolfhard and Parker [12] by means of a two-dimensional laminar dif-
fusion ame obtained with two parallel jets, combustible and oxidizer respectively.
This type of ame is specially suitable for spectroscopic studies. Wolfhard and Parker
have mainly experimented with ammonia-oxygen and ethylene-oxygen ames. Their
studies conrm Burke and Schumanns stand-point. The reaction zone is in thermal
and chemical equilibrium and therein temperature is practically constant. The concen-
trations of reacting species are very small within this zone and, except when dealing
under special conditions that reduce the reaction rate, the process is governed by the
diffusivity of the species. The reacting species generally dissociate before reaching
the reaction zone due to temperature. In the case of hydrocarbons such cracking leads
to carbon formation. Temperatures observed agree fairly with those predicted by the-
ory.2 Zeldovich [16] has taken into consideration the nite thickness of the reaction
2 See further on 3.
12.2. GENERAL EQUATIONS FOR LAMINAR DIFFUSION FLAMES 289
zone to explain the blowing-off phenomenon. A similar study has been performed
by Spalding to explain the extinction phenomenon in the combustion of fuel droplets
with convection.3
The following paragraphs are devoted to the deduction of the general equations
for diffusion ames as well as to the simplied equations on which their study is based.
12.2 General equations for laminar diffusion ames
The general equations given in chapter 3 will be applied here to the study of laminar
diffusion ames by assuming that Burke and Schumanns conditions stated in 1 are
satised. The study limits to the case of steady motions. In the deduction that follows
the notation in chapter 3 will apply.
Continuity equations
1) For the mixture.

Since the motion is steady, equation (3.6) reduces to
(v) = 0. (12.1)
2) For the species.

As per Burke and Schumanns rst condition, the reaction rate outside the ames
surface f is zero. Therefore, system (3.7) reduces to
(v )Yi + (Yi vdi ) = 0, i = 1, 2, . . . , l. (12.2)
Equation of motion
Equation of motion (3.18) takes the form
(v )v = p + ev + F . (12.3)
The following cases are specially interesting:
1) Pressure is constant and the inuence of mass forces negligible. Then Eq. (12.3)
takes the form
(v )v = ev . (12.4)
3 See 13 of Chap. 13.
Moreover, as done in the theory of free jets, here only some terms of the viscosity
forces need to be preserved.4
2) The inuence of mass forces is important. Hydrostatic equilibrium in the uid
undisturbed by the ame determines the pressure eld. If 0 is the density of the
undisturbed uid
p + 0 F = 0, (12.5)
which gives for Eq. (12.3)
(v )v = ev + ( 0 )F . (12.6)
Furthermore, in each case only the signicant terms of ev will be retained. So

far a solution that includes the inuence of the last term of Eq. (12.6) has not been
obtained.
Energy equation
Kinetic energy of motion, energy dissipated by viscosity and work done by mass
forces are negligible. Hence, equation (3.38) reduces to
( )

(v )h (T ) + Yi hi vdi = 0. (12.7)
i
Diffusion equations
Such equations are given by system (3.8) which takes the form
Yj ( vdi vdj Yi Yj
)
+ = 0, (i = 1, 2, . . . , l)
j
Mj Dij Yi Yj
(12.8)
Yj vdj = 0,
j
where pressure and thermal diffusion are not included since they are negligible.
State equation
This is equation (3.39)

p
= Rm T. (12.9)

4 See Fay, Ref. [15].
12.3. BOUNDARY CONDITIONS ON THE FLAME 291
The previous system together with the adequate boundary conditions determine
the values for p, , T , v and for the mass fractions Yi at each point as well as the shape
of the ame which so far is unknown.
12.3 Boundary conditions on the ame
Chemical species Ai forming the mixture can be classied into three different groups:
a) Reacting species Ari , which diffuses towards the ame where they burn. These
species are unable to cross f .
b) Reaction products Api , which produce at the ame and diffuse from it towards
both sides.
c) Inert species Adi , which dilute reacting species. These species can cross f .
According to Burke and Schumanns model mass fractions Yir of the reacting
species must be zero on the ame surface, that is
Y1r = Y2r = = Ylrr = 0 (12.10)
on f , where lr is the number of reacting species.5 These conditions determine the

position of the ame surface.
Let mri be the mass of species Ari that reaches f per unit surface and per unit
time. mri is given by
mri = Yi vi ni , (12.11)
where ni is the unit vector normal to f at Ari side. mri is consumed by the chemical
reactions taking place at f . Let

0 00 p
ij Ari ij Ai , (j = 1, 2, . . . , r), (12.12)
i i
be one of these reactions, which transforms reacting species Ari into reaction products
Api . Since the thickness of the reaction zone is assumed to be zero, it becomes neces-
sary to dene a surface reaction rate for each reaction. Such rate gives the masses of
the species consumed or produced by the reaction per unit surface and per unit time.
Be rj the value of such surface reaction rate for reaction j. The fraction mrij of mri
consumed by this reaction is
0
mrij = Mir ij rj , (j = 1, 2, . . . , r), (12.13)
where Mir is the molar mass of species Ari .

5 Actually, these mass fractions are very small but not strictly zero. Their values are practically those of
the equilibrium of species under the prevailing conditions at the ame.

Therefore mri is given by

0
mri = Mir ij rj , (i = 1, 2, . . . , lr ). (12.14)
j
Likewise if mpi is the mass of species Api produced per unit surface and per
unit time one has

00
mpi = Mip ij rj , (i = lr + 1, lr + 2, . . . , la ), (12.15)
j
where la = lr + lp is the number of active species (reactants plus products) and lp is

the number of products.
Reaction rates rj are unknown a priori. Elimination of the same between

equations (12.14) and (12.15) gives la r relations between Yir and Yip which enables
to express all mass fractions of active species as functions of r selected from them.
Thus the variables of the problem reduce in la r.
So far no solutions have been obtained for this general system of equations.
In order to make computation available, additional assumptions must be introduced
to simplify considerably the problem. Such simplications will be performed in the
following paragraph.
12.4 Simplied equations
In this approximate study the number of species is assumed to be three. Namely, fuel
A1 , oxidizer A3 and products A2 . Products include not only species resulting from
reactions, but also diluents of fuel and oxidizer initially mixed with them.6 Flame sur-
face divides space into two regions: an interior region at the fuel side and an exterior
one at the oxidizer side. Only A1 and A2 species exist within the interior region. In
the exterior one only A2 and A3 . Therefore, the composition of the mixture within the
interior region is determined by the value of Y1 which must satisfy equation (12.2)
(v )Y1 + (Y1 vd1 ) = 0. (12.16)
Similarly in the exterior region one has for Y3
(v )Y3 + (Y3 vd3 ) = 0. (12.17)

6 A larger number of species could easily be included by distinguishing, for example , between diluents
and products. Such is done, among others, by Zeldovich [17] and Fay [15]. However, this does not represent
any fundamental advantage and computations become more elaborate. A differentiation between products
and diluent can be done once the problem is solved.
12.4. SIMPLIFIED EQUATIONS 293
As for Y2 , its value is given by expressions
Interior region: Y2 = 1 Y1 . (12.18)

Exterior region: Y2 = 1 Y3 . (12.19)
Diffusion velocities vd1 and vd3 are given by Ficks law7
Y1 vd1 = D12 Y1 , (12.20)

Y3 vd3 = D23 Y3 . (12.21)
On surface f
Y1 = Y3 = 0; Y2 = 1. (12.22)
Furthermore, since Y1 and Y3 must diffuse towards the ame in the stoichiometric
ratio from (12.11), (12.20) and (12.21) we have on f
Y1 Y3
D12 = D23 . (12.23)
ni ne
Here is the ratio between the masses of oxidizer and fuel needed for complete com-
bustion. ne and ni are normals to f towards the exterior and interior regions respec-
tively.
System of equations (12.16) and (12.17) can be substituted by a single equation

for a new variable Y dened as follows

Y1 in the interior region,
Y = (12.24)
1
Y3 in the exterior region.

Y must satisfy the following differential equation
(v )Y (DY ) = 0, (12.25)
where D takes the value

D in the interior region,
12
D= (12.26)
D in the exterior region.
23
Furthermore, Y is zero on the ame and takes values of opposite sign at both sides of
it. The derivatives of Y at both sides of the ame in normal direction to it must satisfy
condition
Y Y
D12 = D23 . (12.27)
ni ne
7 See Chap. 2.
In particular, if condition
D12 = D23 = D (12.28)
is satised, Y as well as its derivatives are continuous even at the ame. Continuity
Eq. (12.1) for the mixture and Eq. (12.3) for motion still hold unchanged.
As for the energy equation we shall assume that the specic heats at constant
pressure of the three species are constant and equal
cp1 = cp2 = cp3 = cp . (12.29)
In such case, one can immediately check that Eq. (12.7) reduces to the following
cp v T (T ) = 0, (12.30)
which hold throughout space.
Let us assume, that besides condition (12.28) the following is satised

= 1, (12.31)
Dcp
in accordance with the result of the elementary Kinetic Theory of Gases. Then, com-
parison between (12.25) and (12.30) suggests the existence of solutions of the form
T = aY + b, (12.32)
where a and b are constant but can be different for the interior and exterior regions.
These solutions are available for the study of diffusion ames if boundary conditions
can be satisfy through them. In particular, since on the ame Y = 0, temperature Tb
of the ame will be constant for the cases where Eq. (12.32) is valid. The study of
diffusion ames reduces, hence, to a computation of the values for , T , v and Y . For
this purpose, system of Eqs. (12.1) and (12.3) is available which must be completed
with the boundary conditions adequate for each case. These refer in general to the
conditions at the discharge sections of fuel and oxidizer and at great distance from the
ame. Let us assume, for example, that in the fuels discharge section
Y = Y10 , T = T0 , (12.33)
and in the oxidizers discharge section
Y = Y30 , T = T0 , (12.34)
Then Eq. (12.32) holds and one obtains in the interior region,
Y Y1
T = Tb (Tb T0 ) = Tb (Tb T0 ) , (12.35)
Y10 Y10
12.4. SIMPLIFIED EQUATIONS 295
while in the exterior region

Y Y3
T = Tb + (Tb T0 ) = Tb (Tb T0 ) . (12.36)
Y30 Y30
In these expressions Tb is still undetermined. Its value can be obtained by considering
an element of the ame as the one in Fig. 12.8 and applying to it the principle of
conservation of energy. One immediately obtains
( )
T T
m2 h2 (m1 h1 + m3 h3 ) = + . (12.37)
ni ne

Here, m1 and m3 are the masses of fuel and oxidizer that reach f per unit surface
m 2,i + m 2,e = m 2 =(1+ ) m 1
m 2,i m 2,e
m1 m3= m1
ni ne
T n i Tn e
Flame
Figure 12.8: Schematic diagram of an element of a diffusion ame.
and per unit time and m2 is the mass of products that emerges from it. But
m3 = m1 , m2 = (1 + )m1 . (12.38)
Furthermore, from Eqs. (12.11), (12.20) and (12.28)

Y1
m1 = D . (12.39)
ni
From Eqs. (12.35) and (12.36)
T 1 Y1
= (Tb T0 ) , (12.40)
ni Y10 ni
T 1 Y3
= (Tb T0 ) . (12.41)
ne Y30 ne
When expressions for m1 , m2 , m3 , T / ni and T / ne are taken into Eq. (12.37)
keeping in mind Eqs. (12.23), (12.29) and (12.31), one obtains for Tb
qr Y10 Y30
Tb T0 = , (12.42)
cp Y30 + Y10
where
qr = h1 + h3 (1 + )h2 (12.43)
is the heat of reaction per unit mass of fuel.
Formula (12.42) shows that Tb is the temperature that would correspond to a

premixed ame where fuel and its diluent were mixed with oxidizer and its diluent in
the stoichiometric ratio. In fact, in such mixture, mass fraction of fuel is
Y10 Y30
Y30 + Y10
and the corresponding temperature of the ame is given by (12.42).
12.5 Solutions of the simplied system
Some approximate solutions have been obtained for the case of two-dimensional and
axis symmetrical ames either conned or open. A detailed study of such ames is
available in the works listed in Refs. [1], [3], [6], [10], [15] and [17]. The follow-
ing chapter studies in detail the diffusion ame that forms surrounding a fuel droplet
which evaporates and burns in an oxidizing atmosphere. The problem has great im-
portance in jet propulsion.
The following simplifying assumptions have generally been adopted:
1) Stream velocity is throughout space parallel to the ame axis and uniform through-
out space or at least at each cross-section of the same.
2) All transport coefcients are independent from the composition and temperature
of the mixture.
3) Diffusion produces only in cross direction to the ame axis.
These assumptions reduce the problem to the integration of the one-dimensional

diffusion equation for the computation of the fuel, oxidizer and temperature distribu-
tions. For example, in the case of an axis symmetrical ame, Fig. 12.1, the preceding
assumptions reduce Eq. (12.25) to
( )
Y 1 Y
v =D r . (12.44)
x r r r
Here, D and v are constant or at least independent from r. In the rst case
the integration of Eq. (12.44) is straightforward. This also occurs in the second case
through the previous change of variable

v
x= d, (12.45)
0 D
which to be determined would require knowing v/D as a function of distance x to

the ame base, which has been introduced by Hottel [3], Wohl [10] and Barr [5] as
12.5. SOLUTIONS OF THE SIMPLIFIED SYSTEM 297
an unknown function empirically determined. The aforementioned simplifying as-

sumptions only represent a poor approximation to reality. In fact, with respect to the
velocity eld it can easily be veried that it is only constant in open ames where fuel
and oxidizer move at equal velocity within the discharge section and, furthermore,
where pressure is uniform throughout space. In such case the gas expansion produces
entirely in cross direction to the ame. As for transport coefcients they can change
in the ratio ten to one between ame surface and unburnt gases. While the assumption
that diffusion produces only in cross direction is well justied, except for special cases
like that of meniscus ames which so far have not been theoretically analyzed.
Before ending this brief review on diffusion ames we shall develop a simple
argument by W. Jost [19] which enables an easy establishment of the inuence on the
length of a diffusion ame of some of the fundamental variables of the process. In
view of the rudimentary approximations needed in Burke-Schumanns calculations in
order to obtain usable expressions, Jost assumes the same validity for his argument.
R
z
L
vt
z
v v
fuel oxigen
burner wall
burner axis
Figure 12.9: Schematic diagram of a overventilated diffusion showing the elements of the
Josts criterion.
Let us consider, for example, a conned ame with an excess of air or an

open ame, Fig. 12.9. According to Jost, the length of the ame is determined by the
condition that the oxygen of the atmosphere surrounding the jet can reach the jets axis
through diffusion. Let D be the oxygen diffusion coefcient and z its mean quadratic
displacement due to diffusion at instant t. Diffusion theory gives for z as function of t
z 2 = 2Dt. (12.46)
But, at the point where the ame closes
z = R, (12.47)
where R is the burners radius, and

L
t= , (12.48)
v
where v is the mean velocity of the jet between the base and the tip of the ame and L
the length of the same. Through elimination of t and z between (12.46), (12.47) and
(12.48) one obtains
R2 v
L . (12.49)
2D
In laminar ames D depends only on the properties of the gases. Furthermore
2
R v is proportional to the fuel ow rate G. Consequently, we have
G
L , (12.50)
2D
in accordance with the result obtained by Burke-Schumann.
In a turbulent ame D is the turbulent diffusivity which has the form
D vR. (12.51)
When this value is substituted into (12.49) it results
L R, (12.52)
which as previously seen also agrees with experimental results.
References
[1] Burke, S. F. and Schumann, T. E. W.: Diffusion Flames. Industrial and Engi-
neering Chemistry, Vol. 20, October 1928, pp. 998-1004.
[2] Barr, J.: Diffusion Flames in the Laboratory. AGARD Memorandum No.
AG11/M7, 1954.
[3] Hottel, H. C. and Hawthorne, W.R.: Diffusion in Laminar Flame Jets. Third
more, 1949, pp. 254-266.
[4] Barr, J.: Diffusion Flames. Fourth Symposium (International) on Combustion,
Williams and Wilkins Co., Baltimore, 1953, pp. 765-771.
[5] Barr, J.: Length of Cylindrical Laminar Diffusion Flames. Fuel, Jan. 1954,
pp. 51-59.
Academic Press Inc. Publishers., New York, 1951.
12.5. SOLUTIONS OF THE SIMPLIFIED SYSTEM 299
[7] Rembert, E. W. and Haslam, R. T.: Factors Inuencing the Length of a Gas
Flame Burning in Secondary Air. Industrial and Engineering Chemistry, Dec.
1925, pp. 1236-1238.
[8] Cuthbertson, J.: Journal Society of Chemical Industries, Transactions, Vol. 50,
1931, pp. 451-457.
[9] Gaunce, H.: Unpublished Research on Flames, M.I.T., 1937.
[10] Whol, K., Gazley, C. and Kapp, N. M.: Diffusion Flames. Third Symposium
pp. 288-301.
[11] Barr, J. and Mullins, B. P.: Combustion in Vitiated Atmospheres. Fuel, Vol. 28,
1949, pp. 131, 200, 205, 225.
[12] Parker, W. G. and Wolfhard, H. G.: Journal of the Chemical Society, 1950,
p. 2038.
[13] Yagi, S. and Saji, K.: Problems of Turbulent Diffusion and Flame Jets. Fourth
more, 1953, pp. 771-781.
[14] Hottel, H. C.: Burning in Laminar and Turbulent Fuel Jets. Fourth Symposium
pp. 97-113.
[15] Fay, J. A.: The Distributions of Concentration and Temperature in a Laminar Jet
Diffusion Flame. Journal of the Aeronautical Sciences, October 1954, pp. 581-
589.
[16] Gaydon, A. G. and Wolfhard, H. G.: Flames, their Structure, Radiation and
Temperature. Chapman and Hall Ltd., London, 1953, p. 135.
[17] Zeldovich, Y. B.: On the Theory of Combustion of Initially Unmixed Gases.
N.A.C.A. Tech. Memo. No. 1296, June 1951.
[18] Spalding, D. B. : The Combustion of Liquid-Fuels. Fourth Symposium (Inter-
864.
[19] Jost, W.: Explosion and Combustion Processes in Gases. McGraw-Hill Book
Comp., New York, 1946, p. 210.

Chapter 13
Combustion of liquid fuels
13.1 Introduction
The present chapter is devoted to the study of some of the problems of the combustion
of liquid fuels. In certain cases surface reactions can take place in the liquid phase, for
example, in the combustion of hypergoles. On the contrary, when a fuel burns in an
oxidizing atmosphere the reaction takes place in the gaseous phase. Thereby, the fuel
evaporation must preceded combustion. In such case, if the mixing of the fuel vapours
and the oxidizing gas takes place before the chemical reaction, a ame produces of
the type studied in chapter 6. On the other hand, if mixing and combustion take place
simultaneously, a diffusion ame is obtained, similar to those studied in chapter 12.
This is the type of combustion considered in the present chapter. Khudiakov [1], [2]
has studied the characteristics of the diffusion ame that forms on the free surface of
a fuel burning in the atmosphere. Spalding [3] has also studied this problem taking
into account the inuence of free and forced convection. He made this study as an
application of his uniform method [4] for the study of all the mass transport processes
(absorption, evaporation, condensation and combustion).
Of further technical interest is the case where fuel forms droplets of small
diameter. Fuel mists formed by droplets of very small diameter (smaller than a few
hundredths of a millimeter) burn forming a ame similar to the premixed gas ames
[5], [6]. In particular, in the said mists, like in the premixed gas ames, the propagation
velocity of the ame, its ammability limits, etc., are well dened magnitudes. Their
values are close, but not equal, to those corresponding to premixed gas ames [7].
The difference is probably due to the droplets evaporation effect. The diameter of the
droplets is so small that the evaporation produces within the heating zone of the ame,
so that the fuel reaches the reaction zone in gaseous phase.
301
302 CHAPTER 13. COMBUSTION OF LIQUID FUELS
If the diameter of the droplet is larger than, for example, a few tenths of a
millimeter the phenomenon becomes more complicated. Such size is too large to
allow evaporation across the heating zone of a ame. Therefore, in this case individual
diffusion ames surrounding the droplets are produced.
For technical applications the fuel is normally introduced in the combustion

chamber by means of an atomizer producing droplets of different sizes. These droplets
spray and evaporate in contact with the atmosphere of the chamber, which is strongly
turbulent. Furthermore, in the combustors of jet engines, the gases ow at high ve-
locity through the combustion chamber. Generally, in such cases not enough space
is available for the complete evaporation of the fuel before it reaches the combustion
zone. Consequently when the fuel reaches the said zone it is only partially evaporated
and the mixing with the oxidizing atmosphere is incomplete. The entire process starts
with the entrance of the fuel in the combustion chamber and ends when the combustion
products are formed. The complete study of this process includes the following stages:
atomizing, spraying, mixing, evaporation and combustion of the fuel jet. Due to the
complexity of the process a study in detail can not be attempted with any probabilities
of success. Therefore, we must resort to empiricism and to the study of simplied
physical models which emphasize some of the features of the phenomenon.
Hereinafter we shall rst briey refer to the atomization, spraying and mixing
of fuel jets and then, more closely, to the combustion of isolated droplets. Finally the
evaporation of droplets with no combustion will be considered.
13.2 Atomization
A great effort has been made both theoretically and experimentally for the study of
the atomization of fuel jets [8], [9]. The process depends on:
1) The geometrical characteristics of the atomizer.

2) The physical properties of the liquid and the atmosphere in which it discharges.
3) The working conditions.
Depending on the discharge velocity of the liquid and on the conditions of the sur-
rounding atmosphere, several states of atomization can be observed [9]. In the dis-
charge at high velocity, normally produced in burners, the atomization starts at the
nozzle of the atomizer. Atomization is produced by the turbulence of the liquid, whose
radial velocities tend to brake the jet, and by the action of the atmosphere at the outlet.
Surface tension and viscosity of the liquid oppose to the jet atomization. The prob-
13.2. ATOMIZATION 303
lem has been reviewed by Heinze [10], who studied in detail the action of its various
factors.
Atomization is characterized by the average size of the droplets and by their

distribution in sizes. The average size of the spray droplets is characterized by the
Sauter diameter ds [11]. This is the diameter that would be obtained in an ideal at-
omization where all the droplets were of equal size and where the surface and total
volume of the fuel were equal to those of actual atomization. The value ds depends
on the variables that characterize the particular conditions of atomization, according
to rules empirically determined (see Ref. [9], p. 117 and f.).
The size distribution in characterized by the mass fraction R (d/ds ) of the fuel
corresponding to droplets of a diameter larger than d. Therefore, the size distribution
dR
function is . Several empirical formulas have been proposed for R (d/ds ) or
d (d/ds )
dR
, [12]. The two formulas generally used are the Rosin-Rammler formula [13]
d (d/ds )
and the Nukiyama-Tanasawa formula [14]. The results given by these two formulas
are in fair agreement with the distributions experimentally observed [15], [16]. These
formulas are as follows.
Rosin-Rammler:
( )1
dR d
= e(d/dm ) , (13.1)
d (d/dm ) dm
or else

R = 1 e(d/dm ) , (13.2)
In these formulas dm is an average diameter, characteristic of the size of atomization,

which is related to the Sauter diameter through formula
( )
dm 1
= 1 , (13.3)
ds
where is the factorial function,1 and is a characteristic parameter of size unifor-

mity. When increases the distribution becomes more uniform.
Nukiyama-Tanasawa:
( )5
dR d
= e(d/dm ) (13.4)
d (d/dm ) (6/) dm
R
1 Also called gamma function, dened as (a) = 0 ta1 et dt, Ed.
or else ( )

6/, (d/dm )
R= (13.5)
(6/)
Here the parameters have the same meaning that in the RosinRammler formula and
is the incomplete gamma function.2 The Sauter diameter is related with dm by means
of ( )
ds (6/)
= . (13.6)
dm (5/)
13.3 Mixing
Once the fuel is atomized it mixes with the surrounding atmosphere due to the action
of turbulence. Lately, Longwell and Weiss [17] have studied the problem for the case
where the fuel atomizes in a turbulent gas stream owing with uniform velocity. Let
f be the ratio of the fuel mass to the air mass. Assuming that:
1) Turbulent diffusion produces only transversely to the main ow.

2) Mean motion is stationary.
3) The process has axial symmetry.
The following approximate equation for f is obtained

( )
f E 1 f 2f
= + 2 . (13.7)
x v r r r
Here v is the motion velocity, E is the coefcient of turbulent diffusion of the

fuel and x and r are the cylindrical coordinates of the system. For the derivation of this
equation E is assumed to be constant. Equation (13.7) is identical to the molecular
diffusion equation under similar conditions.
If the fuel is vaporized, E is the coefcient of turbulent diffusion dened by

Taylor [18]. In this case, E equals the product of intensity by scale of turbulence.
If the fuel is in the liquid state, E is appreciably smaller due to the inertia of the
droplets which are unable to follow the air uctuations. In a typical case calculated by
Longwell and Weiss, E was only 35% of the value corresponding to the diffusion of
the vapour.
If the boundary conditions in the atomizing section are known, equation (13.7)
can be integrated. For example, if the mixing starts from the origin of coordinates,
which acts as a source of fuel of strength Gc , and if the duct radius is large, the
Rx
2 Dened as (a, x) = 0 ta1 et dt, Ed.
13.4. COMBUSTION 305
solution corresponding to Eq. (13.7) is
vr2
Gc v
f= e 4Ex , (13.8)
Ga 4E
where Ga is the air mass ux in the duct. This solution corresponds, for example, to
the case of a cylindrical atomizer discharging downstream of the gas ow. Formula
(13.8) can be used to obtained the value for E.
From this fundamental solution, the solutions corresponding to different distri-

butions of fuel sources in the atomizing section can be found by superposition. Such
as discs, rings, etc. These solutions represent with fair approximation the actual mix-
ing processes for several practical cases, such as the upstream atomization or the case
of a swirl atomizer, etc. Some of these solutions can be found in the work by Longwell
and Weiss.
13.4 Combustion
The study under given conditions of the combustion of a fuel spray, for example, under
the prevailing conditions in the combustion chamber of a jet engine, is very arduous.
At present it can only be attempted empirically.3 Therefore it is justied as a prelimi-
nary phase to analyze the combustion of isolate droplets under laboratory conditions.
This problem has lately been studied successfully both theoretically and experimen-
tally by several investigators. Such problem is the subject of this and the following
paragraphs where it will he considered in detail. Further on a complementary study
will be made on the evaporation of the droplet when combustion is absent.
Experimental evidence shows that combustion takes place within a region of

small thickness surrounding the droplet. From each side towards this region diffuse
fuel vapours and oxygen. Burnt gases diffuse from this region towards the exterior.
Therefore we have a diffusion ame of the type considered in chapter 12. Thus it
will be assumed that the ame thickness is zero and that the mass fractions of fuel
and oxidizer at the ame are also zero. Moreover, the reacting species must diffuse
towards the ame in the stoichiometric ratio.
The ame must supply the necessary heat for the previous evaporation of the
fuel. This heat is transferred to the droplet surface partially through conduction and
partially through radiation. The energy transmitted through radiation is only a small
fraction of the heat absorbed by the droplet through conduction. Therefore, in the
3 See 14 of the present chapter.
present approximate study its inuence is neglected. Otherwise, the analysis of the
problem becomes far more complicated. By doing so the results obtained do not
substantially varies.
The heat received by the droplet is partially used for its heating and partially for
the evaporation of the fuel. The rst fraction increases the temperature of the droplet
up to the boiling point Ts of the fuel. Thereon it remains constant and equal to Ts . All
the heat received is used up in evaporating the fuel. Hereinafter, the transient initial
period will be neglected by assuming that the droplet temperature is initially Ts and
that no thermal gradients exist therein.
Strictly the process is non-stationary since the size of the droplet decreases as
combustion progresses. However, the recession velocity of the droplet surface is very
small compared to the velocity of the burnt gases. Thereby, an excellent approxima-
tion is obtained by assuming that the phenomenon is stationary. That is to say by
neglecting the local variations with time of temperature and mass fractions produced
by the droplet reduction in size. This is an important simplication of the problem
since it eliminates an independent variable from the equations.
The existence of free and forced convection introduces a privileged direction

destroying the spherical symmetry of the phenomenon. Even in the case of free con-
vection with approximately round droplets experimental results show that the shape
of the ame differs appreciably from the spherical within the upper region. A study of
the problem, taking into account these convection effects is very arduous and has not
yet been achieved. Nevertheless, when assuming that both the droplet and the ame
front are spherical, that is when neglecting these effects, the results obtained show a
good agreement with the experimental results as for the overall magnitudes such as the
droplet burning velocity and its law of variation with size. Thereby, neglect of convec-
tion effects is justied. With this assumption and the assumption of quasi-stationary
state previously stated, there is only one independent variable, this is the distance r to
the center of the droplet.
The composition of the fuel vapours, oxidizing atmosphere and burnt gases is
very complex. However, as done in the study of diffusion ames, we shall assume for
simplicity that only three different chemical species exist, namely: fuel, oxidizer and
inert gases, which include these in the atmosphere surrounding the droplet as well as
the combustion products.
Summarizing, the following assumptions are adopted:
1) The phenomenon has spherical symmetry.

13.5. NOTATION 307
2) The phenomenon is quasi-stationary.

3) Combustion takes place at constant pressure.
4) The droplet temperature is uniform and equal to the boiling temperature of the
fuel at ambient pressure.
5) The chemical reaction takes place upon a spherical surface, named the ame
front. Fuel vapours and oxygen diffuse in the stoichiometric ratio towards this
ame front from which the burnt gases ow. The gases that reach the ame front
react instantaneously. Thereby, on the ame front both the mass fractions of fuel
vapours and of oxygen are zero.
6) Only three chemical species exist, namely: fuel, oxygen and inert gases.
These assumptions allow a simple analysis of the problem and lead to results
which have been experimentally conrmed. Such a treatment of the problem has been
applied by Godsave [19] and Spalding [3] in England, and by Penner and Goldsmith
[20] in the U.S.A. Aside from the aforementioned assumptions Godsave and Spalding
also assume that transport coefcients are independent from temperature by adopting
a mean value for these coefcients. This rather arbitrary assumption reduces apprecia-
bly the suitability of the method. In fact, for the existing range of temperatures, these
coefcients can vary in a ratio of ten to one. Goldsmith and Penner take into account
this variation as well as that of heat capacities. The following study is mainly based
on the work of these two authors but it assumes that heat capacities are independent
from temperature since their variation has no substantial inuence on the results.
13.5 Notation
In the present study the notation used in the preceding chapter will be completed with
the following (see Fig. 13.1):
M = Droplet mass
m = Mass ow across a closed surface surrounding the droplet.
ql = Latent heat of evaporation per unit mass of fuel.
r = Radial distance to the center of the droplet.
v = Radial velocity of the mixture.
vj = Radial velocity of species Aj .
vjd = Radial diffusion velocity of species Aj .
Yj = Mass fraction of species Aj .
= Stoichiometric ratio of oxygen mass to fuel mass.
Subscripts:
1 = Fuel
2 = Inert gases
3 = Oxygen
e = Exterior to the ame front
i = Interior to the ame front
l = On the ame
s = On the droplet surface
= At great distance from the droplet
Y Y2
Y1
Y2 oo
Y3 Y3 oo
rl
Ts rl r
rs
Too
T
T Tl
Figure 13.1: Schematic diagram of the droplet combustion problem.
13.6 Continuity equations
Since the process is stationary, the uid mass that crosses any spherical surface con-
centric to the droplet must be constant. Due to the spherical symmetry of the problem
such condition can be expressed in the form
4r2 v = const. (13.9)
The constant can be determined by particularizing this equation on the droplet surface
r = rs . This surface acts as a fuel source. The strong of this source is the mass m of
vapour produced per unit time. Therefore, we have
4r2 v = m. (13.10)
This equation is valid throughout the uid space surrounding the droplet.
13.6. CONTINUITY EQUATIONS 309
Chemical reactions take place only on the ame surface r = rl , which acts as
a sink for fuel and oxidizer and as a source for the combustion products. Therefore
one obtains for each different chemical species an equation similar to Eq. (13.10). In
this equation must be substituted by the partial density i = Yi corresponding to
species Ai , v by velocity vi = v + vdi , where vdi is the radial diffusion velocity of the
species, and m must be substituted by the partial ow mi of the species. Thus
4r2 Yi vi = mi , (i = 1, 2, 3). (13.11)
Moreover, the following evident conditions must be satised

Yi vi = v, (13.12)
i

mi = m. (13.13)
i
Equation (13.11) is valid for each species Ai , as long as r does not cross the ame.
Let us now apply these equations separately to the interior and exterior regions
of the ame.
Interior region rs r rl
In this region only fuel vapours and inert gases exist, since the incoming oxygen is
entirely consumed as it reaches the ame without crossing it.
Equation (13.11) applied to the fuel vapours A1 gives
4r2 Y1 v1 = m1 = m, (13.14)
since on the droplet surface the mass ow m is the mass ow m1 of the fuel produced
by evaporation.
By comparing (13.9) and (13.14)
Y1 v1 = v. (13.15)
Equation (13.12) gives

Y1 v1 + Y2 v2 = v. (13.16)
From Eqs. (13.15) and (13.16) results
v2 = 0, (13.17)
that is to say within the interior region the inert gases are at rest and the fuel vapours
diffuse through them towards the ame.
Exterior region r rl
In this region only oxygen and inert gases exist. Equation (13.11) applied to each of
them gives
Inert gases: 4r2 Y2 v2 =m2 . (13.18)

2
Oxygen: 4r Y3 v3 =m3 . (13.19)
Similarly, Eqs. (13.12) and (13.14) give
Y2 v2 + Y3 v3 = v, (13.20)
m2 + m3 = m. (13.21)
Let be the stoichiometric ratio of oxygen mass to fuel mass. Since, the fuel
mass reaching the ame per unit time is m, the oxygen mass that reaches the ame
per unit time must be m. And since the oxygen moves towards the ame, v3 must be
negative, that is
m3 = m. (13.22)
From here and (13.21) there results for m2
m2 = (1 + )m. (13.23)
Equations (13.22) and (13.23) determine the constants for the continuity equa-
tions (13.18) and (13.19) as functions of m, thus obtaining
4r2 Y2 v2 = (1 + )m, (13.24)

4r Y3 v3 = m.
2
(13.25)
Therefore, within the exterior region the ame acts as a sink of strength m for the
oxygen and as a source of strength (1 + )m for the combustion products.
13.7 Energy equation
The process is stationary and it takes place at constant pressure. Moreover the kinetic
energy of the motion and the work of the viscous forces can be neglected. Therefore
the summation of the heat and enthalpy uxes through any spherical surface concentric
to the droplet must be constant. Due to the spherical symmetry, this condition can be
expressed in the form
dT
mi hi 4r2 = const. (13.26)
i
dr
The value for the constant can be obtained by applying this equation to the droplet
surface r = rs . Since, here m1 = m and m2 = m3 = 0, we have
( )
dT
mh1s 4rs
2
= const., (13.27)
dr s
( )
dT
where 4rs2 is the heat received by the droplet surface per unit time. Since
dr s
the droplet temperature is assumed to be constant and equal to the boiling temperature
Ts of the fuel, this heat must be used in evaporating the combustible. Furthermore
this heat is the only source of energy for the evaporation, since the energy received
from the ame by heat radiation has been neglected.4 Let ql be the latent heat of
evaporation per unit mass at temperature Ts . Since the evaporated mass per second is
m, the following condition is obtained
( )
2 dT
4rs = mql . (13.28)
dr s
Consequently, the value for the constant is m(h1s ql ), which taken into
Eq. (13.26) gives
dT
mi hi 4r2 = m(h1s ql ). (13.29)
i
dr
The specic enthalpy hi of species Ai can be expressed in the form

T
hi = h0i + cpi dT. (13.30)
T0
When this expression is substituted into (13.29) we obtain

T ( )

dT
4r2 mi cpi dT = m(ql h1s ) + mi h0i . (13.31)
dr T0 i i
The form taken by this expression within the interior and exterior regions will
be studied separately.
As aforesaid in this region only fuel vapours ant inert gases exist. Furthermore due
to Eq. (13.17) the ow of inert gases throughout the spherical control surface is zero.
4 This matter has been considered by G.A. Godsave who concludes that the radiation energy received
by the droplet per gram of evaporated fuel is approximately proportional to the droplet radius. For large
droplets (rs ' 1 mm) it can become a 20% of the energy needed to produce evaporation.
Therefore, there only remains the enthalpy ux of the fuel vapours and Eq. (13.31)
reduces to
T ( Ts )
2 dT
4r m cp1 dT = m ql cp1 dT , (13.32)
dr T0 T0
which determines the distribution of temperature between the droplet surface and the
ame front.
The integration constant of Eq. (13.32) is determined by expressing that the

value of the temperature on the droplet surface must be Ts , that is
r = rs : T = Ts . (13.33)
Here, only oxygen and inert gases exist and their ows are given by Eqs. (13.22) and
(13.23) respectively. Therefore, the energy equation (13.31) takes the form
T
2 dT
( )
4r m (1 + )cp2 cp3 dT =
dr T0
( Ts )
( )
= m ql h01 + h03 (1 + )h02 cp1 dT (13.34)
T0
Let qr be the heat of reaction per gram of fuel in gas state at the temperature
T0 . This heat is
qr = h01 + h02 (1 + )h03 . (13.35)
Substituting Eq. (13.35) into (13.34), the latter takes the form
T ( Ts )
2 dT
4r m cp dT = m ql qr cp1 dT , (13.36)
dr T0 T0
where, in short
cp = (1 + )cp2 cp3 . (13.37)
The constant resulting from the integration of this equation is determined by

expressing that the temperature at great distance from the droplet has the value T
r r : T T . (13.38)
The continuity of the temperature at the ame imposes an additional condition
T (rl ) = T (rl+ ), (13.39)
which will be used in determining the position of the ame.

13.8. DIFFUSION EQUATIONS 313
13.8 Diffusion equations
To compute the distribution of the mass fractions of fuel vapours and oxygen, the
equations (13.14) and (13.25) must be integrated. For this v1 and v3 must be expressed
as functions of the mixture velocity v and of the mass fractions Y1 and Y3 by using the
diffusion equations. Both the interior and exterior region will be studied separately.
The continuity equation 13.14) for the fuel vapour can be written in the form
4r2 Y1 (v + vd1 ) = m, (13.40)
where vd1 is the diffusion velocity of the fuel vapours through the atmosphere of inert
gases. Ficks law,5 gives for this velocity
1 dY1
vd1 = D12 , (13.41)
Y1 dr
where D12 is the diffusion coefcient for the fuel vapours and the inert gases.
When Eq. (13.41) is substituted into Eq. (13.40) and Eq. (13.10) is taken into
account, one obtains
dY1
4r2 D12 = m(1 Y2 ) (13.42)
dr
for the determination of Y1 . The integration constant for this equation is determined
by expressing that the mass fraction of the fuel vapours at the ame is zero
r = rl : Y1 = 0, (13.43)
This procedure applied to Eq. (13.25) gives the following equation for the distribution
of oxygen in the exterior region of the ame
dY3
4r2 D23 = m( + Y3 ). (13.44)
dr
The solution to this equation must satisfy the condition that on the ame front mass
fraction of oxygen must be zero,
r = rl : Y3 = 0, (13.45)
which determines the value for the corresponding integration constant.

5 See chapter 2.
The solution of Eq. (13.44) must satisfy another additional condition. In fact,
the mass fraction of oxygen must tend to the value Y3 corresponding to the compo-
sition of the atmosphere surrounding the droplet at great distance from the same
r: Y3 Y3 . (13.46)
This equation will be used in determining the burning velocity m of the droplet.
13.9 Combustion velocity of the droplet. Temperature

and position of the ame
From the preceding paragraph it results that the solution of the combustion problem
of a droplet reduces to the following:
1) The integration of the system of differential equations

T ( Ts )
2 dT
4r m cp1 dT =m ql cp1 dT , (13.32)
dr T0 T0
dY1
4r2 D12 = m(1 Y1 ), (13.42)
dr
for the determination of temperature T and mass fraction Y1 of the fuel vapours
within the interior region rs r rl , with the two boundary conditions
r = rs : T = Ts , (13.33)
r = rl : Y1 = 0. (13.43)
2) The integration of the differential equations

T ( Ts )
2 dT
4r m cp dT =m ql qr cp1 dT , (13.36)
dr T0 T0
dY3
4r2 D23 =m( + Y3 ), (13.44)
dr
with boundary conditions
r : T T , (13.38)
r =rl+ : Y3 =0, (13.45)
for the determination of temperature and mass fraction Y3 of oxygen within the
exterior region r rl .
13.9. COMBUSTION VELOCITY OF THE DROPLET 315
When Y1 and Y3 are known, the distribution of inert gases Y2 is determined by

the following equations
rs r rl : Y2 = 1 Y1 ,
(13.47)
r r1 : Y2 = 1 Y3 .
The solution of the system must also satisfy conditions
r = rl : T (rl ) = T (rl+ ), (13.39)

r: Y3 Y3 . (13.46)
These two additional conditions determine, as aforesaid, the burnt burning velocity m
of the droplet and the position rl of the ame front.
The ame temperature Tl is the value for T given by the solution of Eq. (13.32),
or (13.36) since both values are equal for r = rl .
For the integration of these systems the specic enthalpies of the various species
must be known. Therefore, the laws of variation with temperature of the heat capac-
ities at constant pressure for the said species must be known. Furthermore one must
also know the coefcients of thermal conductivity and diffusion as functions of the
temperature and composition of the mixture. As well as the heat of reaction qr and the
latent heat of evaporation ql of the fuel. Hereinafter it will be assumed that the heat
capacity at constant pressure does not vary with temperature, within the range under
consideration.6 In such case Eq. (13.32) takes the form
hi = h0i + cpi (T T0 ), (13.48)
and Eqs. (13.32) and (13.36) reduce respectively to

dT
4r2 mcp1 T = m(ql cp1 Ts ) (13.49)
dr
and
dT ( )
4r2 mcp T = m ql qr cp T0 cp1 (Ts T0 ) . (13.50)
dr
Such is the form in which these equations are used in the following calculations.
As for the transport coefcients we shall only consider the case where is
independent from the mixture composition but varies proportionally to the absolute
temperature. That is

= const., (13.51)
T
6 Goldsmith and Penner in [20] take into consideration the variation of the heat capacity with tempera-
ture.
where, in general, the value for the constant will vary when passing from the interior
to the exterior region.
Moreover, it is assumed that the following condition is satised

= i = const., (13.52)
Dij cpi
in accordance with the results of they elementary theory of diffusion. Experimental

evidence shows that this condition is approximately satised.
13.10 Integration of the equations
Under the assumptions stated in 9 the system of equations (13.49) and (13.42), cor-
responding to the interior region, takes the form
T dT
4r2 s mcp1 T = m(ql cp1 Ts ), (13.53)
Ts dr
T dYl
4r2 (D12 )s = m(1 Y1 ), (13.54)
Ts dr
where the values of and D12 are expressed as
T
= s , (13.55)
Ts
T
D12 = (D12 )s , (13.56)
Ts
as functions of the corresponding values on the droplet surface and of the ratio of ab-
solute temperatures T /Ts . The integration of Eq. (13.53) is straightforward. Making
use of condition (13.33) one obtains
( ( ) [ ( )])
1 1 4s T ql cp1 Ts T
= 1+ 1 ln 1 + 1 . (13.57)
rs r mcp1 Ts cp1 Ts ql Ts
The integration of (13.54) is simplied by previously eliminating r between

(13.53) and (13.54). Thus one obtains for as a function of T
( )1
cp1 (T Ts ) + ql
Y1 = 1 , (13.58)
cp1 (Tl Ts ) + ql
where conditions (13.43) has been used and according to (13.44) and 1 is given by

1 = . (13.59)
D12 cp1
13.10. INTEGRATION OF THE EQUATIONS 317
Similarly, the system of Eqs. (13.50) and (13.44) corresponding to the exterior
region takes the form
T dT ( )
4r2 mcp T = m ql qr cp T0 cp1 (Ts T0 ) , (13.60)
T dr
2 T dY3
4r (D23 ) = m( + Y3 ), (13.61)
T dr
where and D23 are expressed as functions of ratio T /T and of their values at
innity.
Proceeding as for the interior region and with boundary conditions (13.38) and
(13.45), the following solution is obtained for this system
( )
1 4 T q q cp T
= 1 + ln , (13.62)
r mcp T cp T q cp T
(( ) )
q cp T
Y3 = 1 , (13.63)
q cp Tl
where the following abbreviations
q = qr ql + (cp cp1 )T0 + cp1 Ts , (13.64)

cp3 3
= = = cp2 , (13.65)
D23 cp D23 cp3 cp (1 + )
cp3
are used for simplicity.
Taking Eq. (13.46) into (13.63) one obtains

(( ) )
q c p T
Y3 = 1 . (13.66)
q cp Tl
This expression gives for Tl

( )1/ ( ( )1/ )
Y3 q Y3
Tl = T 1 + + 1 1+ . (13.67)
cp
Consequently, the temperature of the ame is independent from its position

and from the size of the droplet.
When condition (13.39) is expressed making use of Eqs. (13.57) and (13.62)
one obtains
( ( ) [ ( )])
1 4s Tl ql cp1 Ts Tl
= 1+ 1 ln 1 + 1
rs mcp1 Ts cp1 Ts ql Ts
( ) (13.68)
4 Tl q q cp T
+ 1 + ln .
mcp T cp T q cp Tl
From which results for m

( )
m 4 Tl q q cp T
= 1 + ln
rs cp T c p T q cp Tl
( ( ) [ ( )]) (13.69)
4s Tl ql cp1 Ts Tl
+ 1+ 1 ln 1 + 1 .
cp1 Ts cp1 Ts ql Ts
Since the right-hand side of this equation depends only on the physicochemical char-
acteristics of the process, it results that the burning velocity of the droplet is directly
proportional to its radius. There is a simple explanation to this fact. The evaporated
mass is proportional to the droplet surface, s , and to the heat Q received per unit
surface and per unit time, m s Q. But, s rs2 . Furthermore, Q 1/rs , since
Q is proportional to the temperature gradient and this gradient is inversely propor-
tional to the distance from the ame to the droplet surface, which is proportional to
rs . Therefore it results m rs2 (1/rs ) = rs .
Let c be the density of the fuel. The droplet mass M is

4 3
M= r e (13.70)
3 s
and its variation with time is
dM drs
= 4rs2 e . (13.71)
dt dt
When this expression is combined with Eq. (13.69) the following law is ob-
tained for the time variation of the droplet radius
drs
2rs = k, (13.72)
dt
where
( )
m 2 Tl q q c p T
k= = 1 + ln
2e rs e cp T cp T q cp Tl
( ( ) [ ( )]) (13.73)
2s Tl ql cp1 Ts Tl
+ 1+ 1 ln 1 + 1
e cp1 Ts cp1 Ts ql Ts
is a constant of the process named evaporation constant.7
When (13.72) is integrated the following expression is obtained for the law of
variation of the droplet radius as a function of time
rs2 = rsi
2
kt, (13.74)
7 The evaporation constant used by Godsave is equal to 4k.
13.11. NUMERICAL APPLICATION 319
where rsi is the initial radius. This formula is the fundamental result of the present
theory.
The combustion time tc of a droplet of radius rsi is, obviously,

2
rsi
tc = . (13.75)
k
The position rl /rs of the ame front is obtained by writing T = Tl in Eq. (13.62).
Furthermore if m/rs is eliminated from the result by means of Eq. (13.73), one obtains
ke cp
rl 2
= . (13.76)
rs Tl q q cp T
1 + ln
T cp T q cp Tl
Therefore, the radius of the ame and the droplet are proportional.
Formula (13.73) enables the study of the inuence of the physical constants of
the fuel and the state of the surrounding atmosphere on the burning rate of the droplet.8
In particular, the said formula shows that the heat transmitted from the ame to the
droplet surface is the determining factor for its burning velocity. This is due to the
fact that the burning rate of the droplet is determined by the heat received by the same.
Therefore, the essential factor in the combustion process is not the volatility of the fuel
but its heat of evaporation. Experiments results conrm this conclusion. In fuels at the
high temperature reached when combustion occurs, the evaporation mechanism differ
essentially from the evaporation when the temperature of the surrounding atmosphere
is normal. In the last case the evaporation is maintained by the diffusion of the vapors
from the droplet towards the exterior. The evaporation velocity is determined by the
difference between the vapour pressures on the droplet surface and at innity.9
13.11 Numerical application
The above formulas are applied here to the study of the combustion of a gasoline
droplet in the air. The following typical values are adopted
T = 300 K, Ts = 355 K, c = 0.85 gr/cm3 ,

= s = 55 106 cal/cm s K, ql = 95 cal/gr, q = 10 000 cal/gr,
cp1 = 0.60 cal/gr K, cp2 = 0.35 cal/gr K, cp3 = 0.26 cal/gr K,
Y3 = 0.23, = 3, 1 = 2 = 0.9.
8 Information on the subject can be found in Godsaves works [19].
9 See 15, Eq. (13.103).
Taking these values into Eq. (13.67) for the temperature of the ame, it is
obtained
Tl = 3112 K. (13.77)
The evaporation constant is obtained from Eq. (13.73)
k = 2.26 103 cm2 /s. (13.78)
The position of the ame is given by (13.76)

rl
= 9.48. (13.79)
rs
The distributions of temperature and mass fractions are given by Eqs. (13.57) and
(13.58) for the interior region and by Eqs. (13.62) and (13.63) for the exterior region
Interior region
r 12.42
= ( ), (13.80)
rs T T
13.60 0.66 ln 1.89 1.24
T T
( ( ))0.9
1 T
Y1 = 1 0.56 . (13.81)
9.72 T
Exterior region
r 10.84
= , (13.82)
rs T 52.76
1 + 53.76 ln
T T
53.76
T
(
( ))0.38
1 T
Y3 = 3 53.76 1 . (13.83)
43.38 T
The distribution of Y2 is given in the interior region by
Y2 = 1 Y1 , (13.84)
and in the exterior region by

Y2 = 1 Y3 . (13.85)
The distribution of T /T , Y1 , Y2 and Y3 have been taken into Figs. 13.2 and 13.3.
13.11. NUMERICAL APPLICATION 321
T /T
l 10
6

T/T
2
Ts/T
0
0 5 10 15 20 25 30
rl/rs=9.48 r/r
s
Figure 13.2: Temperature prole given by Eqs. (13.80) and (13.82).
1.0
Y1s
Y2 Y
2
0.8
Y2
3
Y ,Y ,Y
0.6
2
1
0.4
Y1
Y3
0.2
Y3
Y2s
0.0
0 5 10 15 20 25 30
rl/rs=9.48 r/r
s
Figure 13.3: Mass fraction proles given by Eqs. (13.81) and (13.83)-(13.85).
The combustion time for a droplet of a radius rsi millimeters is given by

Eq. (13.75). We obtain
2
tc = 4.42 rsi s. (13.86)
2
In Fig. 13.4 the variation law of tc as a function of rsi is represented.
3
t (s)
c
0
0 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000
r (mm)
si
Figure 13.4: Droplet combustion time as a function of the initial size.
13.12 Comparison with experimental results and limi-

tations of the theory
Godsave [19], [21], Topps [22], Hall and Diederichsen [23], Spalding [24], Hottel
et al. [25], Wise et al [26], Kobayasi [27], Nishiwaki [28] and others have pub-
lished experimental results on the combustion of droplets. Such results have been
obtained from different technical procedures and they conrm the main conclusions
of the present theory. For example, Fig. 13.5 obtained in the I.N.T.A Combustion
Laboratory at Madrid10 conrms the linear law (13.74) of variation of the square of
the radius with time. The following table 13.1, taken from the work by Goldsmith and
Penner [20], compares theoretical values of k with these experimentally obtained by
Godsave [21] for different fuels. As seen the agreement is generally excellent.
Fuels k 103 cm2 /s

Theoretical Experimental
Benzene 2.5 2.43
Ethyl alcohol 1.98 2.03
Ethyl Benzene 2.13 2.15
N-heptane 2.15 2.43
Table 13.1: Comparison between theoretical and experimental results for k.
10 The results have been obtained by applying Godsaves technique, that is to say, by suspension of a
droplet from a quartz lament and photographing the variation of radius as a function of time.
13.12. COMPARISON WITH EXPERIMENTAL RESULTS AND LIMITATIONS OF THE THEORY 323
0.5
0.4
rs 10 (cm )
2
0.3
2
0.2
2
0.1
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
t (s)
Figure 13.5: Experimental results showing the linear dependence between rs2 and time, ob-
tained for isooctane in air at ambient pressure.
On the other hand a comparison of the theoretical and experimental radius

of the ame [20] shows that the former is two or three times larger than the lat-
ter. Although formula (13.76) shows that ratio rl /rs is practically independent from
pressure, Wise, Lovell and Wood [26] have experimentally observed that this ratio
increases as pressure decreases. The authors explain theoretically this effects as a
consequence of free convection. Photographic evidence shows that combustion is ac-
companied by strong free convection ows due to which the spherical ame appears
only in the lower half of the droplet whilst in the upper half it takes a considerably
elongate shape. The intense formation of carbon observed makes difcult the optical
study of this region. This convection effect may account for the difference between
the theoretical and experimental values of the ame radius making the values of k
predicted by theory agree with the experimental values as previously seen.
Formula (13.73) shows that the burning velocity of a droplet is practically in-
dependent from pressure. In fact, its only inuence shows small reduction of latent
heat of evaporation that accompanies and increase in pressure. However, Hall and
Diederichsen [23] have experimentally checked that the burning velocity of a droplet
is approximately proportional to the 4th root of pressure. It has been suggested [20]
that the following factors could account for this effect, aside from the aforementioned
decreases in evaporation heat:
1) Augmentation of the energy transmitted by radiation from the ame to the droplet
which increases very rapidly with pressure.
2) Augmentation of the rate of chemical reactions which could be very signicant
if the burning velocity of the droplet would depend, even if only slightly, on the
said reaction rate, contrary to what is postulated in the present theory.
3) Augmentation of the free convection effects due to the increase of the Grashof
number with pressure.
The theoretical study of such effects is very arduous as it implies taking into
account the inuence of radiation, nite thickness of the ame and free convection.
So far this study has not been satisfactorily accomplished.
13.13 Inuence of convection
The present theory excludes convection effects. Experiments with suspended droplets
show that the inuence of free convection is not signicant. Less experimental in-
formation is available for the case of forced convection. Spalding [24] has performed
some measurements for very large Reynolds numbers (from 400 to 4 000). His experi-
ments show that for this range the burning velocity of droplets with forced convection
can be computed by applying the Frossling formula11 for the evaporation with no
combustion. Spaldings experiments bring forth the fact that, at least in the analyzed
range, two different combustion states can exist. For convection velocities lower than
a given critical value, a semi-spherical ame surrounds the front part of the droplet,
whilst in the opposite side a long wake forms with a strong formation of carbon. If the
convection velocity is larger than the critical value the front ame extinguishes and
the wake ame remains. It is questionable whether this second state also produces in
the case of very small droplets.
Spalding explains this extinction as follows. Even in the present theory the
thickness of the ame is assumed to be zero, actually it is nite. Now, the convection
activates evaporation and increases the ame thickness in order to burnt the largest
quantity of fuel that must be consumed per second. But such an increase in thickness
is accompanied by a decrease in the maximum temperature of the ame. Since reac-
tion rate changes very rapidly with temperature if the said decrease is large enough
the reaction is incomplete and combustion extinguishes. The same occurs in diffusion
ames.12 When assuming a reaction rate of the Arrhenius type, Spaldings calcula-
11 See Eq. (13.106)
12 See chapter 12.
13.14. COMBUSTION OF FUEL SPRAYS 325
tions demonstrate that the extinction velocity is proportional to the droplet diameter.
Experimental results conrm such prediction. Spalding also arrives to the conclusion
that, even when no convection exists, the ame extinguishes if the diameter of the
droplet is very small.
13.14 Combustion of fuel sprays
The present theory enables the calculation with good approximation of the burning
velocity of an isolated droplet under laboratory conditions. In particular, this theory
allows the study of the inuence of the physical characteristics of the fuel and the
state of the surrounding atmosphere on the burning velocity of the droplet. Now these
results must be extended to the study of the combustion of fuel sprays of the type
existing in the combustion chambers of jet engines [29]. Such an extension has not
yet been achieved. In the actual state of knowledge this seems impossible due to the
interaction of the many factors of the process. In fact, for this extension to become
possible it is necessary to know in advance the distribution of droplets in the spray
and the characteristics of the surrounding atmosphere, in a system in which the fuel
is partially in the liquid phase and partially in the gaseous phase as it enters the com-
bustion zone. It is also necessary to take into account the interaction of droplets and
the inuence of the highly turbulent motion of the gas. All this makes the theoretical
study of the problem extremely difcult. For this reason the studies made up to date
are of a highly empirical character. These studies limit themselves either to an analy-
sis of the inuence of some fundamental parameters on the combustion efciency of a
burner [30], or to obtain general conclusions as for the way in which several variables
can inuence the process [24].
13.15 Droplet evaporation
When combustion is absent the evaporation process of a fuel spray is also a very
complicated phenomenon. This problem has only been studied empirically for some
typical cases [16]. Bahr [30], for example, has given an empirical formula to express
under given conditions the evaporated fraction as a function of the distance to the
atomizer. The Bahr formula represents a good approximation to the results obtained
from his own experiments and those performed by Ingebo [16].
As a rst step towards the solution of the problem, the evaporation of isolated
droplets can be studied by the same procedure applied in the preceding paragraphs
to the study of combustion. The extrapolation of the obtained values to the case of a
spray is difcult. In fact, it is necessary to know the size distribution of the droplets
and the composition of the surrounding atmosphere. Furthermore, in order to account
for the convection effects, which are very important, the motion of the droplets relative
to the atmosphere must also be known. The available information is not sufcient to
show the way in which this extrapolation can be performed [34].
The evaporation velocity of a droplet when convection is absent can be easily

obtained from the formulas deduced in the preceding paragraphs. In fact, assuming
that the droplet is isothermal, it is enough to make use of the equations corresponding
to the interior region, enlarged to innity. Therein the following boundary conditions
must be satised which determine the integration constants
r: T T , Y1 Y1 , (13.87)
where Y1 is the mass fraction of the fuel vapours at great distance from the droplet.
Moreover, if one keeps the assumptions previously established with respect to the law
of variation of the values of the transport coefcients as functions of temperature, the
two equations for T and Y1 are Eqs. (13.53) and (13.54), that is
T dT
4r2 mcp1 T = m(ql cp1 Ts ), (13.88)
T dr
T dY1
4r2 (D12 ) = m(1 Y1 ). (13.89)
T dr
The solution of this system that satises conditions (13.87) is

( )
1 4 T ql cp1 Ts cp1 (T Ts ) + ql
= 1 + ln , (13.90)
r mcp1 T cp1 T cp1 (T Ts ) + ql
( )
1 Y1 cp1 (T Ts ) + ql
= . (13.91)
1 Y1 cp1 (T Ts ) + ql
The last equation when applied to the droplet surface r = rs gives the follow-
ing relation
( )
1 Y1s ql
= . (13.92)
1 Y1 cp1 (T Ts ) + ql
Let p1s be the partial pressure of the fuel vapour on the droplet surface. Thermody-
namics shows13 that p1s is determined by the temperature Ts on the droplet surface.
13 See Prigogine, I. and Defay, R.: Chemical Thermodynamics. Longmans Green & Co., 1954, pp. 332
and f.
13.15. DROPLET EVAPORATION 327
4
0.25
0.20 nheptane nheptane

noctane
P1 (kg/cm2) 0.15
nexane
0.10
3 0.05
noctane
0.00
250 275 300 325 350
P (kg/cm )
2
T (K)
2
1
nexane
1
ndecane
ndodecane
0
250 300 350 400 450 500
T (K)
Figure 13.6: Partial pressure of fuel vapour as a function of temperature.
Therefore a relation exists between p1s and Ts of the form
f (p1s , Ts ) = 0. (13.93)
Figure 13.6 gives Eq. (13.93) for some typical fuels.
Furthermore14
p1 Ma Y1
= ( ) , (13.94)
p Mc Ma
1+ 1 Y1
Mc
where Mc and Ma are, respectively, the molar masses of the fuel vapour and of the
gas through which it diffuses.
When (13.94) is particularized on the droplet surface, taking the result into
(13.93), the following relation between Y1s and Ts is obtained
F (Y1s , Ts ) = 0. (13.95)
This relation and (13.92) determine Y1s and Ts .
Once Ts is known the evaporation velocity m of the droplet is deduced from

(13.90) by making r = rs and T = Ts . Thus obtaining
( )
4 rs Ts ql cp1 Ts ql
m= 1 + ln . (13.96)
cp1 T cp1 T cp1 (T Ts ) + ql
14 See chapter 1.
The parenthesis on the right-hand side of this equation is independent from

rs . Therefore, evaporation velocity, like combustion velocity, is proportional to the
droplet radius.
Making use of Eq. (13.71) a relation similar to (13.74) is obtained for the vari-
ation of rs
rs2 = rsi
2
kv t, (13.97)
where kv is the evaporation constant, which from (13.96) is given by the expression
( )
2 Ts ql cp1 Ts ql
kv = 1 + ln . (13.98)
c cp1 T cp1 T cp1 (T Ts ) + ql
If the following condition is satised
cp1 (Ts T )
1, (13.99)
ql
a linearization of (13.96) gives for m
4 rs
m= (T Ts ). (13.100)
ql
A similar linearization of (13.92) gives
cp1 (T Ts ) 1 Y1s Y1
= . (13.101)
ql 1 Y1
By taking this expression into (13.100), it results for m
Y1s Y1
m = 4rs D12 , (13.102)
1 Y1
or else, as a function of p1s and p1 by virtue of (13.94),
4rs D12 p1s p1
m= p, (13.103)
Rc T p p1
where Rc = R/Mc is the gas constant for the fuel vapour. Eq. (13.103) is Langmuirs
evaporation formula valid for small evaporation velocities.
Let
cp1 (T Ts )
x= . (13.104)
ql
If this value is taken into (13.96) and the result divided by (13.100), the following
relation between m and m0 is obtained
[ ( ) ]
m ql cp1 T ln(1 + x)
= 1 1 +x (13.105)
m0 cp1 T ql x
where m is the actual evaporation velocity and m0 is the velocity that would be ob-
tained if Langmuirs formula would apply to all cases.
13.15. DROPLET EVAPORATION 329
0.9
0.8
0.7
0.6
0
m/m
0.5
0.4
0.3
0.2
0.1
cp1T/ql=2
4 6 8 10
0
0 1 2 3 4 5 6 7 8 9 10
x
Figure 13.7: The ratio m/m0 as a function of x = cp1 (T Ts )/ql for some values of
cp1 T /ql .
Figure 13.7 gives m/m0 as a function of x, for some values of cp1 T /q1 .
When the evaporation velocity is high the fast decrease of the evaporation constant
can be seen due to the transport of enthalpy done by the vapour motion.
The previous formulae do not take into account the inuence of convection.
Such effect has been studied by Frossling [32], who obtained the following relation
between the evaporation velocity mc with convection and the evaporation velocity at
rest
mc ( )
= 1 + 0.276 Sc1/3 Re1/2 . (13.106)
m
Here Sc = /D12 and Re = vds / are, respectively, the Schmidt num-

ber and the Reynolds number of the motion, is the gas viscosity coefcient, ds the
droplet diameter and v the motion velocity. Ranz and Marshall [33] have experimen-
tally veried this formula. Fig. 13.8 gives mc /m as a function of Sc2/3 Re.
Formula (13.105) was obtained from experiments at ambient temperature. In-

gebo [34] gives the following empirical formula which represents with good approxi-
mation the experimental results obtained by him from 30 C to 500 C
mc ( )
= 1 + 0.151(Sc Re)0.6 . (13.107)
m
24
Ingebo (Sc=0.65)
20
16
Frssling
m /m
12
c
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
2/3
S Re 103
c
Figure 13.8: Experimental relations of Frossling and Ingebo, giving mc /m as a function of
Sc2/3 Re.
This formula is represented by the dash line in Fig. 13.8. It is seen that the Frossling
formula underestimates the evaporation velocity for the large Reynolds numbers.
Miesse [35] has lately calculated the motion of an evaporating droplet by as-
suming:
1) The coefcient of the aerodynamic drag of the droplet is inversely proportional

to the Reynolds number (laminar ow).
2) The evaporation velocity of the droplet follows the Frossling law.
Miesse has obtained solutions for the case where the air velocity changes linearly with
distance. Such solutions allow the calculation of the distance covered by the droplet
as a function of the reduction of its diameter and the distance needed for complete
evaporation, etc. His work contains interesting practical applications. However, when
trying to extend his conclusions to the case of sprays the remarks made at the begin-
ning of this paragraph should be taken into account. On the other hand the possible
interaction of evaporation and aerodynamic drag of the droplet is neglected.
Penner [36] has calculated the evaporation time of a propellant droplet within
the combustion chamber of a rocket by assuming that the droplet is isothermal but its
temperature changes with time. He has estimated the possible inuence of the radia-
tion energy on the evaporation process reaching the conclusion that such inuence is
of no signicance.
13.16. APPENDIX: APPLICATION OF PROBERTS METHOD 331
13.16 Appendix: Application of Proberts method for

the combustion of fuel sprays
Probert15 has developed a theoretical method for the study of evaporation or combus-
tion of fuel sprays, under the assumption that evaporation constant is the same for all
droplets. The justication of this assumption and the value that should assigned to the
constant if it is valid, depend on the results the experimental measurements.
After chapter 13 was written, some theoretical works on the application of

Proberts method have been performed at the I.N.T.A.16 corresponding to steady burn-
ing as well as to transition from ignition to steady burning and periodic combustion.
In these computations the size distribution function of Mugele-Evans was used with
preference to those of Rosin-Rammler and Nukiyama-Tanasawa, since it allows the
prediction of the sizes with a very good approximation taking into account the maxi-
mum size of the droplets. Let F be the mass fraction of fuel corresponding to droplets
with a diameter smaller than d. Mugele-Evans formula gives for F the following
expression ( ( ))
1 d
F = 1 + erf ln . (13.108)
2 dmax d
Here erf is the error integral, dmax is the droplets maximum diameter and and
are two parameters characteristic of the distribution. Fig. 13.9 shows some of the
distributions corresponding to typical values for these parameters. It is seen that the
increment of increases the uniformity of the spray, whilst when increases the mean
diameter of the droplets reduces.
Let G be the volume of fuel, injected to the burner per unit time, and g the
volume of the droplets existing in the burner. It can be veried that g is expressed as a
function of G through formula
I
g = G tv , (13.109)

where I is given by expression
( p )2
1 1x ln
x2 + y
p
e 1 x2 + y
I= 4
x dx ( ) dy, (13.110)
0 0 (x2 + y)5/2 1 x2 + y
and tv is the life time of the largest droplets of the spray.

15 Probert. R.P.: The Inuence of Spray Particle Size and Distribution in the Combustion of Oil Droplets.
Philosophical Magazine, February 1946.

16 Millan, G., and Sanz, S.: Analysis of the Combustion Processes in Gas Turbines. Fourth International
Congress of Combustion Engines, Zurich, 1957.

1.0
0.9
(1.5,1.0)
(1.0,1.0)
0.8
0.7
(1.5,1.5)
(1.0,1.5) (1.0,0.5)
0.6
(0.5,0.5)
F
0.5
0.4
(0.5,1.0)
(0.5,1.5)
0.3
(1.5,0.5)
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
d/d
max
Figure 13.9: Droplet size distribution without combustion according to Mugele and Evans
(fraction F of droplets with a diameter smaller than d/dmax ), for several val-
ues of the parameters ( , ).
Magnitude (13.109) is interesting since the combustion intensity of the burner

should be considered inversely proportional to it.
Table 13.2 gives the values for g/G tv for three typical cases corresponding to
= 1.
0.5 1 1.5
g
0.129 0.110 0.105
G tv
( )
ds
0.269 0.438 0.895
dmax spray
( )
ds
0.517 0.454 0.401
dmax ame
Table 13.2: Values of g/Gtv for = 1 and = 0.5, 1, 1.5.
It is seen that when increases, that is when the uniformity of the spray in-
creases, the mass fraction of fuel in the burner decreases. Consequently, it is advan-
tageous to work with spray as uniform as possible in order to decrease the volume of
the primary zone of the burner.
Combustion changes the droplet size distribution, preserving, obviously the

maximum diameter. Figure 13.10 shows the distribution functions at the ame corre-
sponding to the three cases studied. For comparison this gure includes the distribu-
tions corresponding to the spray. Table 13.2 also gives Sauters diameters for both the
spray and the ame.
1.0
0.9
0.8
0.7
= 1.5 = 1.0
0.6
F
0.5
= 0.5
0.4
0.3 Spray
Combustion
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
d/d
max
Figure 13.10: Effect of combustion on the droplet size distribution for some values of .
References
[1] Khudyakov, G. N.: Combustion of Liquids from Free Surfaces. Isv. Acd. Sci.
USER, Div. Chem. Sci., Nos. 10-11, 1945.
[2] Khudyakov, G. N.: Distribution of Temperature within a Liquid Burning from
a Free Surface, and Description of the Flame Formed. R.A.E. Translation no.
422, 1953.
[3] Spalding, B. D.: The Combustion of Liquid Fuels. Fourth Symposium (Inter-
864.
[4] Spalding B. D.: The Calculation of Mass Transfer Rates in Absorption, Vapor-
ization, Condensation and Combustion Processes. Proceedings of the Institu-
tion of Mechanical Engineers, Vol. 168, No. 19, 1954.
[5] Haber, F. and Wolff, H.: Uber neboloxplosionen. Zeitschrift fur Angewante
Chemik, July 1923, pp. 373-377.
[6] Jones, G. W. et al.: Research on the Flammability Characteristics of Aircraft
Fuels. WADC Tech Rept., June 1952, pp. 52-35.
[7] Browning, J. A. and Krall, W. G.: Effect of Fuel Droplets on Flame Stability,
Flame Velocity and Inammability Limits. Fifth Symposium (International) on
Combustion, Reinhold Publishing Corp., New York, 1955, pp.159-163.
[8] The Penn State Bibliography on Sprays. The Texas Company, 1953.
[9] Griffen, E. and Muraszew, A.: The Atomization of Liquid Fuels. Chapman and
Hall Ltd., 1953.
[10] Hinze, J. C.: On the Mechanism of Desintegration of High Speed Liquid Jets.
Proceedings of the Sixth International Congress of Applied Mechanics, 1946.
[11] Sauter, J.: Die Grossenbestimmung der in Gemischnebel von Verbrenungskrait-
maschinen verhandenen Brennstoffteilchon. Forschung Gebiet des Ingenieur-
swesens, No. 279, 1926.
[12] Mugele, R. H. and Evans, H. D.: Droplet Size Distribution in Sprays. Industrial
and Engineering Chemistry, June 1951, pp. 1317-1324.
[13] Rosin, P. and Rammler, E.: The Laws Governing the Fineness of Powdered
Coal. Journal of the Institute of Fuel, 1933, pp. 29-36.
[14] Nukiyama, S. and Tanasawa, Y.: An Experiment on the Atomization of Liquid
by Moans of Air Stream. Transactions of the Society of Mechanical Engineers
of Japan, Feb. 1938, p. 86; May 1938, p. 1389; Feb. 1939, pp. 63 and 68.
[15] Bevans, R. S.: Mathematical Expressions for Drop Size Distribution in Sprays.
Conference on Fuel Sprays, Univ. of Michigan, March 1949.
[16] Graves, Ch. C. and Gerstein, M.: Some Aspects of Combustion of Liquid Fuel.
AGARD Memorandum AG 16/M, May 1954.
[17] Longwell, J. P. and Weiss, M. A.: Mixing and Distribution of Liquids in High-
Velocity Air Streams. Industrial and Engineering Chemistry, March 1953, pp. 667-
677.
[18] Taylor, G. I.: Diffusion by Continuous Movements. Proceedings of the London
Mathematical Society, 1921, p. 196.
[19] Godsave, G. L.: Studies of the Combustion of Drops in a Fuel Spray. The Burn-
ing of Single Drops of Fuel. Fourth Symposium (International) on Combustion,
Williams and Wilkins Co., Baltimore, 1953, pp. 818-830.
[20] Goldsmith, M. and Penner, S. S.: On the Burning of Single Drops of fuel in on
Oxidizing Atmosphere. Jet Propulsion, July-August 1954, pp. 245-251.
[21] Godsave, G. A.: The Burning of Single Drops of Fuel. Part. I, Temperature
Distribution and Heat Transfer in the Preame Region. NGTE Report no. R.66,
1950. Part II, Experimental Results. NGTE Report no. 87, 1951. Part III,
Comparison of Theoretical and Experimental Burning Rates and Discussion of
the Mechanism of the Combustion Process. NGTE Report no. R.38, 1952.
[22] Topps, J. E. C.: An Experimental Study of the Evaporation and Combustion of
Falling Droplets. Journal of the Institute of Petroleum, 1951, pp. 535-537.
[23] Hall, L. R. and Diederichsen, J.: An Experimental Study of the Burning of Sin-
gle Drops of Fuel in Air at Pressures up to Twenty Atmospheres. Fourth Sym-
1953, pp. 837-846.
[24] Spalding, D. B. Combustion of Single Droplet and of a Fuel Spray. Selected
Combustion Problems, Vol. I, AGARD, 1954, pp. 340-351.
[25] Hottel, H. C., Williams, G. C. and Simpson, H. C.: The Combustion of Droplets
of Heavy Liquid Fuels. Fifth Symposium (International) on Combustion, Rein-
hold Publishing Corp., New York, 1955, pp. 101-129.
[26] Wise, H., Lowell, J, and Wood, B. J.: The Effects of Chemical and Physical
Parameters on the Burning Rate of a Liquid Droplet. Fifth Symposium (Inter-
national) on Combustion, Reinhold Publishing Corp., New York, 1955, pp. 132-
141.
[27] Kobayasi, K.: An Experimental Study on the Combustion of a Fuel Droplet.
Fifth Symposium (International) on Combustion, Reinhold Publishing Corp.,
New York, 1955, pp. 141-148.
[28] Nischiwali, N.: Kinetics of Liquid Combustion Processes: I. Evaporation and
Ignition Lag of Fuel Droplets. Fifth Symposium (International) on Combustion,
Reinhold Publishing Corp., New York, 1955, pp. 148-158.
[29] Surugue, J.: Combustion Problems in Turbojets. AGARD AG 5/P2, December
1952, pp. 54-61.
[30] Bahr, D.: Evaporation and Spreading of Isooctane Sprays in High-Velocity Air
Streams. NACA RM E511 E53114, 1953.
[31] Saks, W.: The Rate of Evaporation of a Kerosene Spray. National Aeronautical
Institute of Canada, Note No. 7, 1951.
[32] Frossling, N.: On the Evaporation of Falling Drops. Gerlands Beitrage zur
Geophysik, 1938, pp. 170-216.
[33] Ranz, W. E. and Marshall, W. R.: Evaporation from Drops. Chemical Engineer-
ing Progress, 1952, pp. 141-46 and 173-180.
[34] Ingebo, R. D.: Vaporization Rates and Heat-Transfer Coefcients for Pure Liq-
uid Drops. NACA Tech. Note No. 2368, 1951.
[35] Miesse C. C.: Ballistics of an Evaporating Droplet. Jet Propulsion, July-August
1954, pp. 237-244.
[36] Penner, S. S.: On Maximum Evaporation Rates of Liquid Droplet in Rocket
Motors. Journal of the American Rocket Society, March-April 1953, pp. 85-88.

Aerotermoquimica

Cargado por

Copyright:

Formatos disponibles

Aerotermoquimica

Cargado por

Información del documento

Descripción original:

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Aerotermoquimica

Cargado por

Copyright:

Formatos disponibles

aerothermochemistry 2009/4/2 14:20 page 1 #1

Gregorio Millan Barbany

Reedicion conmemorativa del 50 aniversario de la edicion original

Escuela Tecnica Superior de Ingenieros Aeronauticos

Asociacion de Ingenieros Aeronauticos de Espana

Aerothermochemistry, Gregorio Millan Barbany

Editores cientcos: Manuel Rodrguez Fernandez, Carlos Vazquez Esp

c 2009 Escuela Tecnica Superior de Ingenieros Aeronauticos

Es un gran placer para m presentar esta re-edicion literal de la monografa

Cuando Gregorio Millan cursaba sus estudios de Ingeniera Aeronautica, que

En la formacion de Gregorio Millan haba tenido un papel crucial el inusual

aerorreactores, siendo el creador del Jet Propulsion Laboratory. Comprendiendo que

La Aerothermochemistry se basa en el ciclo de conferencias que Teodoro von

El ARDC facilito la difusion internacional, a traves de laboratorios universi-

que no cambiaron los resultados, y se han incluido algunas anotaciones signicativas.

En la Aerothermochemistry aparece por primera vez el marco multidiscipli-

La importancia de la Aerothermochemistry para la Escuela Espanola de Mecani-

Gregorio Millan agradece en el prologo la valiosa colaboracion de Carlos San-

Los tres fueron determinantes en el desarrollo de la Escuela Espanola de Mecanica de

Las contribuciones cientcas de Millan, que estan recogidas en parte en este

Espero que esta re-edicion de la Aerothermochemistry sirva de modesto home-

INSTITUTO NACIONAL DE TECNICA AERONAUTICA

I have great pleasure in introducing the report of Prof. Gregorio Millan on

Special acknowledgment is given to the contribution of the European Ofce

Finally, I wish to acknowledge the valuable collaboration of Carlos Sanchez

Tarifa, Manuel de Sendagorta and Ignacio Da Riva, aeronautical engineers of the

Madrid, January 1958

2 Transport phenomena in gas mixtures 37

3.4 Energy equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.5 General equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

3.6 Entropy variation . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3.7 One-dimensional motions . . . . . . . . . . . . . . . . . . . . . . . 72

3.8 Stationary, one-dimensional motions . . . . . . . . . . . . . . . . . 75

3.9 The case of only two chemical species . . . . . . . . . . . . . . . . 77

3.10 Stationary, one-dimensional motion of ideal gases with heat addition 79

3.11 Appendix: Notation and repertoire of vectorial and tensorial formulae 84

4.1 Detonation and deagration . . . . . . . . . . . . . . . . . . . . . . 89

4.2 Kinds of detonations and deagrations . . . . . . . . . . . . . . . . 91

4.3 Velocity of the burnt gases . . . . . . . . . . . . . . . . . . . . . . . 94

4.4 Propagation velocity . . . . . . . . . . . . . . . . . . . . . . . . . . 98

4.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

4.6 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

4.7 Indeterminacy of the solution . . . . . . . . . . . . . . . . . . . . . 103

5 Structure of the combustion waves 105

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

5.2 Wave equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

5.3 Characteristic times . . . . . . . . . . . . . . . . . . . . . . . . . . 110

5.4 Limiting form of the wave equations . . . . . . . . . . . . . . . . . 111

5.5 Detonations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5.6 Deagrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

5.7 Transition from deagration to detonation . . . . . . . . . . . . . . 126

6 Laminar ames 131

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

6.3 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 135

6.4 Modication of the conditions at the cold boundary . . . . . . . . 137

6.5 Propagation velocity of the ame . . . . . . . . . . . . . . . . . . . 140

6.6 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

6.7 Reaction velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

6.8 Flame equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147