7.

Homogeneous and heterogeneous

combustion

7.1.1 Homogeneous combustion and bromine to form the corresponding hydrochloric and
hydrobromic acids. In short, the category of combustion
Overview processes includes all reactions that are very rapid and
Our understanding of the phenomenon of combustion strongly exothermic.
dates from the second half of the Eighteenth century, when The peculiar nature of combustion processes stems from
Antoine-Laurent Lavoisier demonstrated that the process the interaction of an ensemble of physical and chemical
was not due to the release of phlogiston, then considered to phenomena which give rise to particular and diversified
be one of the constituent parts of matter, but instead to the situations. The simplest is that of a gaseous mixture
combining of one component of the air, oxygen, with containing a fuel, methane, for example, oxygen and any
particular gaseous, liquid or solid combustible materials. Its inert gases such as nitrogen which may be present, in which
special characteristic is the manifest emission of light and a very rapid exothermic reaction is triggered that produces a
heat which often takes on the typical aspect of flames. strong heating effect with consequent sharp variations of the
In reality, combustion processes have been the main temperature and concentrations of the reagents in time and
source of energy for mankind ever since prehistoric times, and space. The geometrical configuration and the physical
even today they play a central role in our economy, supplying conditions of the mixture, in particular the temperature, of
about 90% of the energy we consume. In spite of the relentless the mixture create the right conditions for its ignition to take
and, in certain respects, challenging search for alternative place.
sources of energy, this pre-eminent position is bound to If the fluids are in motion, changes in components and
remain unchallenged for a long time to come. The fuels operating conditions cause different identifiable situations to
commonly used as sources of energy are natural gas, which is arise, each of which is characterized by specific problems
principally made up of methane; petroleum products, which for the description and handling of combustion processes.
are composed of mixtures of hydrocarbons; and coal. When The case of a gaseous mixture formed by a fuel and a
these combine with oxygen (the comburent substance) carbon comburent (combustion-support) that moves with laminar
dioxide and water are formed through a series of chemical flow in a given direction at constant rate will be considered
reactions that, as a rule, are very complex because they take first. If a combustion reaction is triggered at a particular
place in various stages and involve, as will be explained position in the direction of the flow, a flame is generated
further on, many intermediate species. Leaving aside these which causes a sharp increase in the temperature of the
details, it is nevertheless easy to formulate the global reactions mixture. The reactive event takes place in a thin layer in
through which combustion products are formed from the fuels which, besides the temperature increase, an abrupt decrease
mentioned. These can be expressed as follows: in the concentration of the reagents also occurs, while the
[1] CH4 ⫹2O2⫺ 2H2O ⫹CO2
䉴
for methane concentration of the reaction products increases at an
equivalent speed and attains its corresponding value in the

冢 冣
m m final mixture. In substance, one can identify a reaction front
[2] CnHm ⫹ n ⫹ 1 O2⫺ nCO2 ⫹ 1 H2O
䉴

4 2 that moves at a given rate of direct propagation in the

for a generic hydrocarbon direction opposite to that of the gaseous flow. If the
numerical values of the two velocities are identical, the
[3] C ⫹O2⫺ CO2 䉴
for coal position of the flame front remains unchanged in time and
These reactions release significant quantities of heat the flame is said to be stationary. This description identifies
energy whose values can be calculated from the enthalpies of the behaviour of what are known as premixed
formation of the various species involved (see below). one-dimensional flames. A different situation occurs in the
Alongside the preceding combustion reactions, there is also, case of diffusion flames, in which the two gaseous reagents
for example, the combination of oxygen with hydrogen to flow into two separate zones and then converge in a third
form water, and the combination of hydrogen with chlorine zone where the combustion takes place.

VOLUME V / INSTRUMENTS 413

COMBUSTION AND DETONATION

In both of the cases described above, however, two coal, or tiny droplets of liquid hydrocarbons, act as fuels. In
further situations stand out, which result from the fact that the first case the reaction occurs through the interaction of
the flow of the currents are, respectively, either of the oxygen molecules with the carbon atoms that constitute the
laminar or of the turbulent type. When the flow is laminar, solid material, and the diffusive processes that take place in
mixing processes occur only at the molecular level, instead the gaseous film adjacent to the surface of the solid
when the flow is turbulent, they are also the result of the significantly influence the kinetics of the combustion
action of turbulent eddies. These characteristics have a process. The combustion of liquid droplets is even more
significant influence on the behaviour of combustion complex, as the heat resulting from the reaction causes the
processes and point out the importance of the fluid-dynamic partial evaporation of the volatile liquid components which
behaviour of the systems concerned. disperse in a gaseous state and react with the oxygen.
The difference between premixed and diffusion flames In conclusion, combustion processes show the
provides us with a criterion for classifying the characteristics articulated variety of situations summed up in the diagram
of a system subject to a combustion process. For example, in reproduced in Fig. 1. After examining the thermodynamic
the carburettor of an internal combustion engine, petrol and and kinetic aspects of combustion processes, the most
air are mixed and then, after undergoing compression and significant typologies listed in the diagram will be analyzed,
ignition, they burn as a premixed flame, with a front that taking into consideration the determination of ignition
propagates inside the combustion chamber. In the case of a conditions, premixed laminar and turbulent flames, and
lighted candle, on the other hand, the heat released by the diffusion flames. Certain aspects that concern the influence
flame causes the candle’s wax to vaporize and mix with the of combustion processes on air pollution, taking into account
air in the area that is close to the wick, where the combustion especially the chemical reactions involved in the formation
process takes place and assumes the typical characteristics of of nitrogen oxides and solid carbonaceous particles, will also
a diffusion flame. Other examples of diffusion flames are be discussed.
present in burner jets or in the reaction engines of jet
aeroplanes. If, however, the burner is very large, it is better Thermodynamic aspects. Flame temperature
to premix the fuel with the air. A generic chemical reaction is used as a reference,
In cases where the chemical reaction is preceded by a formally expressed as follows:
diffusion process, typical concentration-profiles of the
[4] ν A A + ν B B + ⋅⋅⋅ ⫺
⫺➤ ν M M + ν N N + ⋅⋅⋅
䉴

various reagents are formed. For example, if the reagents,

which are made up of gaseous fuel and air, flow respectively where vi is the stoichiometric coefficient of the ith
into an internal cylinder and into the ring of a co-axial component. At constant pressure, the heat released by the
external cylinder, the combustion process that takes place reaction, QR, is equivalent to the variation in enthalpy
when they come together occurs on a flame front associated with the reaction itself (DH), but with the
characterized by cylindrical symmetry, with the consequent opposite sign; if the pressure is 1 bar, this can be computed
formation of the component concentration-profiles having on the basis of the standard molar formation enthalpies
the same symmetry. Df H° of the various species, linearly combined according to
Heterogeneous combustion processes are entirely stoichiometric coefficients (assumed to be positive for
different. In these processes, particles of a solid, for example reaction products and negative for reagents):

COMBUSTION PROCESSES
(rapid, strongly exothermic reactions)

PREMIXED REAGENTS DIFFUSION FLAMES

NON-STATIONARY STATIONARY HOMOGENEOUS HETEROGENEOUS

STATES STATES their characteristics their characteristics
(time-dependent laminar and are determined are determined
processes) turbulent flames; by the flow by the exchange
their velocity of gases and of heat
is determined by by interdiffusive and material
the physico-chemical phenomena with the surface
characteristics of the solid
phenomena of ignition of the mixture
and extinction

Fig. 1. Classificatory scheme of the various different types of combustion.

414 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

In practice, no chemical reaction ever takes place to a

Table 1. Standard formation enthalpies and standard full extent, but tends to reach an equilibrium composition
formation free energies at 298 K for various gaseous that depends on the pressure and the temperature reached by
compounds involved in combustion processes the system in accordance with the equation:
ν ν
p MM pNN ⋅⋅⋅⋅
Compound
⌬f H ° ⌬f G° [9] ν ν
p AA pBB ⋅⋅⋅⋅
( )
Kγ = K T = e r
− ∆ G ° RT

kJ/mol kJ/mol
where pi is the partial pressure of the component i at
CH4 ⫺74.86 ⫺50.85 equilibrium, which is equal to its mole fraction times the
C2H6 total pressure P, where P, DrG° is the variation in the
⫺84.78 ⫺33.00
standard free energy of the reaction and R is the gas
C6H14 ⫺154.74 ⫺0.25 constant. The quantity DrG°, in turn, is calculated on the
basis of the standard free energies of formation of the
C8H18 ⫺208.59 16.41
various components, Df Gi° (see Table 1), by means of an
C6 H6 82.98 129.75 additive equation similar to [5]. Finally,
Kg⫽fMnMfNnN…ⲐfAnAfBnB…, where fM ,… are the coefficients
CH3OH ⫺202.05 ⫺163.45 of fugacity of the various components which measure the
C2H5OH ⫺234.75 ⫺168.18 deviation from ideal behaviour; plainly, if the components
and the mixture behave ideally, then Kg⫽1.
Actually, apart from global reactions such as those
[5] QR = − ∆ r H  = − ∑ ν i ∆ f H
i indicated above, in a combustion process there are also other
i equilibrium reactions involving the reaction products that
Table 1 reports the values of the formation enthalpies of must be taken into consideration, such as:
various species involved in combustion processes. They refer [10] CO2⫺ 䉳
⫺CO ⫹1Ⲑ2 O2
䉴

to standard conditions, which correspond to a pressure of 1

bar and a temperature of 298 K. [11] H2O⫺
⫺H2 ⫹1/2 O2
䉳
䉴

If one knows the molar heat capacities c̃p,i of the various amongst others. Since the composition of the system at
substances, one can calculate the degree to which variations equilibrium depends on the temperature that is reached, [8]
in enthalpy depend on temperature using the equation: must be solved together with equations of type [9] which
T allow the evaluation of a composition as a function of the
[6] ∆ r H T = ∆ r H 298 + ∫ ∑ ν c
i p ,i
dT temperature. By and large, a system of non-linear algebraic
298 i equations is obtained, which can be easily solved by means
Heat capacities, in turn, are usually expressed as a of adequate computational programmes.
function of temperature by means of a polynomial function
of the following type: Chemical and kinetic aspects of the combustion
of hydrocarbons
[7] c
p ,i = ai + biT + ciT + ⋅⋅⋅
2
The rate ri of a generic chemical reaction expresses
whose parameters ai, bi, ci, ... are known as calorimetric the number of moles of a species of reference i that are
coefficients. Their values for some of the compounds transformed per unit of time and per unit of volume. This
commonly involved in combustion processes are set out in depends on the volume and composition of the system:
Table 2. for a combustion reaction, in its simplest form it is
One of the important parameters that characterizes a usually expressed as a function of the primary reagents,
flame is the temperature that is reached after combustion as follows:
(Tf), starting from the value To. If the flame is adiabatic, i.e.
if all the heat released by the reaction contributes only to
heating the reacting mixture, a simple balancing of energies Table 2. Values of the calorimetric coefficients
permits the calculation of the final temperature, also called for the calculation of molar heat capacities (J/mole⭈K)
in this case adiabatic flame temperature. This value is of some gaseous substances involved in combustion
obtained by equating the heat that is absorbed by the mixture
of the reaction products to the heat released by the reaction Temperature
Compound a b⭈103 c⭈106
(at tempercature To), calculated by means of equations [5-7]: interval (K)
Tf

∫ ( A + BT + CT )
CH4 22.36 48.15 – 273-1,200
[8] QR − ∆ r H T° = 2
+ ⋅⋅⋅ dT =
0
T0 CO 27.63 5.02 – 273-2,500

( B
2
) ( C
= A T f − T0 + T f2 − T02 + T f3 − T03 + ⋅⋅⋅
3
) ( ) CO2 26.67 42.29 –14.25 300-1,500

H2 28.81 0.28 1.17 273-2,500

where A⫽兺iniai , B⫽兺inibi , C⫽兺inici …, in which the
summations extend to the products of the reaction. The H2O 34.42 0.63 5.61 300-2,500
adiabatic flame temperature is calculated by solving
equation [8] for the unknown quantity Tf . The degree of N2 26.38 7.62 –1.44 273-2,500
extension of the polynomial depends on the degree of O2 26.21 11.50 3.22 273-5,000
accuracy one wishes to achieve.

VOLUME V / INSTRUMENTS 415

COMBUSTION AND DETONATION

the process occurs following the collision between two

Table 3. Kinetic parameters of certain fuels species A and B, the rate of reaction is expressed as
follows:
E k = AT n e− E RT
Fuel A(one-step) a b [13]
kJ/mol
where n is a number in the order of unity.
CH4 1.3⭈109 202.64 ⫺0.3 1.3
Thanks to the ensemble of methods of investigation
C2H6 1.1⭈1012 125.60 0.1 1.65 mentioned, a considerable range of values of kinetic
parameters is now available for many of the elementary
C6H14 5.7⭈1011 125.60 0.25 1.5 reactions involved in homogeneous combustion processes.
One of the most thoroughly investigated cases has been that
C8H18 7.2⭈1011 167.47 0.25 1.5
of the oxidation of methane at a high temperature, which has
C10H22 3.8⭈1011 125.60 0.25 1.5 led to the general scheme illustrated in Fig. 2. The scheme
shows the most important reactions involving compounds
C2H5OH 1.5⭈1012 125.60 0.15 1.6 that contain one or more carbon atoms. It can be observed
C6H6 2.0⭈1011 125.60 ⫺0.1 1.85 that the methane molecule is attacked by the radical species
⭈H, ⭈O, ⭈OH, highly reactive because of the unpaired electron
present, indicated with a point. ⭈CH3 radicals are thus
( )
a b
[12] r=k T  fuel   oxygen  formed, which react with the oxygen atoms to produce
formaldehyde, CH2O. By removing a hydrogen atom from
The square brackets indicate the molar concentration or this species the formic radical is obtained, which via thermal
molarities of the components, a and b are two empirical decomposition forms carbon monoxide and atomic
parameters known as orders of reaction, while k(T) is the hydrogen. Furthermore, two ⭈CH3 groups can combine
reaction rate constant that depends on the temperature together to form ethane which in turn can be attacked by ⭈H,
through the Arrhenius equation k(T)⫽A exp(⫺EⲐRT), where ⭈O and ⭈OH to form the ethyl radical which will then be
E is the activation energy. The values of these kinetic oxidized. Hydrocarbons with a molecular weight greater
parameters for some hydrocarbons are reported in Table 3. than methane and ethane are also attacked by these radicals,
The preceding equation [12] expresses the rate of the forming unstable species that are subject to rapid demolition
combustion process taken as a whole, in other words, with the formation of smaller molecular fragments. Thus the
without taking into consideration the fact that in reality the problem of the oxidation of the higher hydrocarbons can be
transformation in question takes place through various stages explained by the same oxidation mechanism involving
which involve several intermediate species, consisting for fragments of small size, whose characteristics are known as
the most part of molecular fragments. This lumped approach, a result of the investigations carried out on the oxidation of
while ‘ignoring’ the details of the transformations examined, methane.
nevertheless provides an effective starting point for It is interesting to observe that in the complex sequence
examining some typical aspects of combustion processes of elementary reactions involved in such processes, the ones
such as ignition and the formation of premixed flames. that play an important kinetic role are:
Very accurate investigations have been conducted [14] H⭈⫹O2⫺ ⭈OH ⫹O⭈
䉴

which, through the use of sophisticated diagnostic

techniques, have supplied interesting information on the [15] CO ⫹⭈OH⫺ CO2 ⫹H⭈ 䉴

details of the combustion reactions of hydrocarbons. Apart which, curiously enough, do not depend on the nature of the
from the traditional analytical techniques such as fuel used.
chromatography and mass spectrometry, new techniques As combustion reactions are normally conducted with
based on the use of laser devices have been developed. For air, and therefore in the presence of nitrogen, nitrogenous
example, by means of the diagnostic procedure known as compounds are also formed, and in particular nitrogen
LIF (Laser Induced Fluorescence), molecular species are oxides. The reactions that correspond to their formation are
identified by exciting them energetically with a laser ray integrated with those relating to combustion as they involve
and then analysing the radiation that they emit by some of the above-mentioned radical species. When
fluorescence. In this way, species can be identified that are operating at a high temperatures and for extended contact
present only in very small concentrations as they have a times, the formation of nitrogen monoxide NO may take
very short average life. Moreover, accurate experimental place via the Zeldovich mechanism:
methodologies have been finalized to determine the rate of [16] O⭈⫹N2⫺ NO ⫹N⭈
䉴

combustion reactions. Lastly, it is worth mentioning that

the study of homogeneous combustion processes is being [17] N⭈⫹O2⫺ NO ⫹O⭈
䉴

conducted also using methodologies that belong to In fuel-rich mixtures, the following reaction also comes
quantum chemistry which, thanks to the use of super into play, which tends to replace equation [17]:
computers, have now made it possible to calculate the
[18] N⭈⫹⭈OH⫺ NO ⫹H⭈ 䉴

reaction rate constant of a number of the elementary

processes involved in said combustion processes with a At relatively low temperatures, NO can be formed also
reasonable degree of accuracy. The expressions of the from N2O through the following sequence of reactions:
velocities of these reactions are based on their molecular
[19] O⭈⫹N2 ⫹M⫺ N2O ⫹M 䉴

nature, as they reflect the characteristics of the elementary

act through which the reaction takes place. For example, if [20] O⭈⫹N2O⫺ 2NO 䉴

416 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

⫹H, O, OH ⫹O ⫹H, O, OH ⫹M, O2, H

CH4 CH3 CH2O CHO CO
⫹H
⫹CH3 ⫹CH3
⫹CH3 ⫹H
⫹M, O2 ⫹M, H
⫹H, O, OH ⫹H ⫹O
C2H6 C2H5 C2H4 C2H3 C2H2 CH2
⫹H ⫹H
⫹O ⫹O, OH ⫹OH ⫹O ⫹H ⫹O, O2

CH3CHO CH3,CH2,CHO CH2CO CH CO

⫹O, O2

⫹H, O, OH ⫹OH ⫹O

CH3CH CH2O, CHO CH3

⫹M

CH3

Fig. 2. Schematic representation of the process of oxidation of methane that shows the most important reactions
involving radical species and molecular fragments.

[21] H⭈⫹N2O⫺ NH ⫹NO

䉴
known as the ignition point and underlies all processes of
combustion. The phenomenon arises from the exponential
where M indicates an inert molecule that facilitates the progression of the Arrhenius equation, which establishes
reaction by subtracting energy from the reagents by means of how the reaction rate depends on the temperature. A method
collisions. A second type of mechanism that leads to the for approximating the value of the critical ignition
formation of nitrogen oxides, called prompt NO, is more temperature was devised by Semenov. It is based essentially
common at low temperatures and with short contact times. It on two premises: that the mixture remain well mixed and
is assumed that hydrocyanic acid HCN, formed by the that the rate at which heat is exchanged between the zone
interaction of carbon radicals with nitrogen, plays the role of where the combustion takes place and the external
intermediary, which is formed by the interaction of carbon as environment be proportional to the difference between their
follows: temperatures. Thus, the quantity of heat dispersed per unit of
[22] ⭈CH ⫹N2⫺ HCN ⫹N⭈
䉴 time can be expressed by the following equation:
[23] ⭈CH2 ⫹N2⫺ HCN ⫹⭈NH
䉴 [24] Q − = hS (T − T0 )
Subsequently, the hydrocyanic acid is converted, through where S is the surface that delimits the system, T0 the
a complex pattern of reactions, into nitrogen oxide. Attention external
. temperature and h the coefficient of heat exchange.
to the reactions that lead to the formation of nitrogen oxides Q⫺ depends upon the physical characteristics of the mixture
is justified by the fact that these species are among the most being used, in particular on its thermal conductivity and
highly polluting agents in the atmosphere. degree of homogenization. The heat that is released per unit
of time can instead be expressed as follows:
Ignition point and temperature
[25] Q = VQ r
+
If a mixture of gaseous fuel and oxygen is contained in a R

closed chamber, the components tend to combine with one where V is the system’s volume and r the rate of the reaction.
another and to release heat. If the rate at which the heat is In the treatment that follows, the latter shall be taken to be a
outwardly dispersed is high, the temperature will remain low, kinetic expression of the order of zero, assuming that the
and the reaction will proceed slowly under a particular exponents a and b in equation [12] are also equal to naught.
condition called slow combustion. This approach, although only approximate since it does not
If, however, the temperature exceeds a certain critical take into account the effect of the reagents’ consumption at
limit which, as we shall see, depends on the specific the start of the ignition process, nevertheless yields results
characteristics of the mixture and the container, the rate at that are satisfactory.
which heat is released can give rise to a great increase in the The energy-balance is expressed by equating the rate of
rate of the reaction, to the point of causing an explosion. The increase of the reactive mass to the difference between the
condition corresponding to the start of such a process is heat that is released Q⭈⫹ and that which is dispersed Q⭈⫺:

VOLUME V / INSTRUMENTS 417

COMBUSTION AND DETONATION

[26] c p rV
dT
dt
(
= hS T − T0 − Ae− E ) RT
QRV
intrinsically unstable and ignition takes place. The threshold
between the two effects is illustrated in case II, in which the
straight line representing dispersed heat is tangential to the
where t is time, cp the specific heat and r the density of the curve representing released heat. At this point, the following
reactive mixture. If the following adimensional values are conditions apply:
introduced: dQ + dQ −
[31] Q + = Q − =
[27] ϑ=
E
RT02
(
T − T0 ) dϑ dϑ
conditions that are called critical, and which lead to the
EAQR following result for the critical parameters:
[28] τ= t
rc p RT02 e
E RT0
[32] ac = e , ϑ c = 1
E RT0
RT02 hSe Thus, if a is less than e (base of natural logarithms) there is
[29] a= ignition; if it is greater there is a stationary state. In
EAVQR
accordance with [27], the following equation relating critical
and bearing in mind that, in general, RT0 ⲐE⬍⬍1, by means of ignition temperature Tc to temperature T0 can be derived:
a series of algebraic transformations it can be demonstrated RTc2
that the energy-balance equation [26] will acquire the [33] Tc − T0 =
following form: E
If, for example, we assume T0⫽500 K and the
dϑ
[30] = eϑ − aϑ reasonable value of 10,000 cal/mol for the energy of
dτ activation, we obtain a value of Tc equal to 550 K, which
The preceding differential equation appears simple at shows that the amount of preheating that would precede
first glance, but, actually, it cannot be analytically integrated. explosive self-ignition is relatively small.
Its particular characteristics can, nonetheless, be note by This theory can be improved if the consumption
recording in a diagram the variations of the two terms to the of reagent is taken into account and expresses the
second member, which refer to the amounts of heat that have combustion reaction rate with a kinetics of the order a⬆0
been either released or dispersed, respectively. In this manner with respect to the reagent fuel. Leaving aside the details
the trends reported in Fig. 3 are obtained, where the three of the calculation, it can be demonstrated that the critical
straight lines refer to the three different values of coefficient value of the coefficient of heat exchange is expressed by
a. In the graph, the stationary state Q⭈⫹⫽Q⭈⫺ corresponds to the following equation:
the intersection between the curve that represents the amount  
( )
2
of heat that has been released and the straight line that [34] ac = e 1 − 2.703 a B 3 
represents the amount of heat that has been subtracted. As  
shown in Fig. 3, intersection occurs in case I. On the where B⫽C0 QR EⲐr CpRT02, in which C0 is the reagent
contrary, no intersection occurs in case III, which implies concentration at time zero. When a⫽0 or when values of B
that the amount of released heat is always greater than that of are very high, the preceding equation is the same as [32], a
dispersed heat. Under these conditions, the system is fact that shows how Semenov’s approach allows the value of
ac to be calculated with a reasonable margin of certainty.

Ramified chain reactions

II In addition to thermal factors, chemical factors may also,
in certain cases, play such an important role in oxidation
reactions as to give rise to explosive evolutions. The radical
species that form in the course of certain stages of the
process, by virtue of their high degree of reactivity due to
the presence of unpaired electrons, can generate, by
interaction with stable molecules, new molecular fragments
I
˙

III that provoke an avalanche of reactions typical of chain-

reaction processes. In these processes, four characteristic
˙

stages can be distinguished:

• Initiation, during which molecular fragments are
generated via thermal decomposition:
[35] A⫺ 2X⭈
䉴

where X⭈ stands for a generic radical species.

• Propagation, during which the number of radicals
0 remains unaltered:
[36] X⭈⫹B⫺ Y⭈⫹C 䉴

Fig. 3. Diagram showing the curve corresponding

to the function eÿ and the straight lines aÿ as a function where Y⭈ stands for another radical. An example of this
of ÿ for different values of a. Line I is subcritical, line II is the reaction between the radical ⭈OH and methane,
is critical and line III is supercritical. that yields water and a methyl radical:

418 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

[37] ⭈OH ⫹CH4⫺ H2O ⫹⭈CH3䉴

which occurs in the course of the oxidation of methane.

• Ramification, a special type of propagation in which the e
number of radicals produced is greater than the number
of radicals participating in the reaction: d

[38] X⭈⫹A⫺ 2Y⭈ 䉴

pressure
An example is provided by the reaction: c
[39] H⭈⫹O2⫺ ⭈OH ⫹O⭈
䉴

b
• Termination, a reaction during which radicals are
eliminated, for example as a result of their pairing:
a
[40] X⭈⫹Y⭈⫺ A 䉴

These reactions are often facilitated by the action of a

third body, a stable inert species that subtracts energy in the
course of the exothermic process during which the two
0 time
radicals combine with one another.
The possible explosive degeneration of a chain-reaction Fig. 4. Typical trends in pressure increase as a function
process is the result of the ramification reactions which, if of time during combustion in a static system. Curves a-e
not controlled by the termination reactions, cause a correspond to increasing values of the initial pressure:
catastrophic increase in the number of radicals that the a and b slow combustion, c in the presence of one cold flame,
system generates. One can carry out an approximate analysis d and e in the presence of two cold flames.
of the phenomenon by assuming that the rate at which
radicals are generated is constant and that the velocities of
propagation and termination are both proportional to the hydrocarbons. If the reaction takes place in a closed
concentration of the radicals present, which will be container and changes in pressure are measured over time,
generically indicated by the symbol X⭈. In a system closed to very different trends will be observed depending on the kind
exchanges of matter, the accumulation of radicals over time of fuel that is used and the conditions in which the reaction
is described by the following equation: takes place. Fig. 4 shows the increase over time in the
d  X· pressure of a mixture in relation to different values of its
[41] = I 0 + f  X· − g  X· = I 0 + ϕ  X· pressure at the start of a reaction. In curves a and b, which
dt correspond to the lowest pressures, there is a period of
where I0 stands for the rate at which radicals are produced induction during which a reaction has yet to take place. After
during the initiation reaction, and f and g, respectively, stand
for two functions, one of temperature and the other of
pressure, that reflect the efficiency of the ramification and
termination stages. The term f⫽f⫺g is known as the
blue-flame zone
ramification factor. Bearing in mind that for t⫽0 the
concentration of radicals is nil, the integration of the
preceding equation yields the following expression for the
concentration of radicals present as a function of time:

[42]
I
(
 X· = 0 eϕ t − 1
ϕ
)
ignition
Its development depends on whether f is positive or
temperature

negative: if it is negative, i.e. if f⬍g, the function tends

asymptotically towards the value:
I slow combustion
[43]  X· = 0
g− f one cold
flame
which is the stationary solution that can be derived directly
from [41] by assuming the first member to be equal to zero. two cold
If instead f is positive, the concentration of radicals flames
increases exponentially. In this case the system is subject to a ignition
catastrophic explosive evolution because the production of hump
radicals via the ramification process prevails over their
destruction via termination.

Ignition diagrams
0 pressure
The mechanism that has been described can be employed
to interpret the combustion of various substances, including Fig. 5. Example of an ignition diagram.

VOLUME V / INSTRUMENTS 419

COMBUSTION AND DETONATION

that, the curves rise sharply and reach a peak pressure point. on the fluid, and to assume that motion occurs with a piston-
If the initial pressure is greater, one obtains the trends like flow, on the basis of which the radial rate profile is
illustrated in curves c, d and e, in which there are described as a function with a constant value at every point of
momentary bursts of pressure that last for a span of one the sections perpendicular to the direction in which the fluid
second, during which time the temperature increases by is flowing, except at the walls of the conduit, where it is nil.
about 150°C. These bursts are accompanied by a weak The formation of a flame is characterized by a sudden and
luminescence and are called cold flames. sharp increase in the temperature along the z axis, as shown
The behaviour of every combustible compound may be in Fig. 6, and by an equally fast and sharp decrease in the
summarized in a diagram known as an ignition diagram; by concentration of the reagents that are then transformed into
plotting temperatures as ordinates and pressure values as combustion products. This means that a frontal part of the
abscissas, different zones can be identified that correspond flame that is only a few millimetres thick may be identified
to the different ways in which combustion occurs. An where the temperature peaks to values as high as 2,000-3,500
example of an ignition diagram is given in Fig. 5, where a K. The front propagates at a rate Su in the direction opposite
sinuous curve is shown that separates the zone on the left, to that of the flow of the gas. Under stationary conditions, the
known as slow combustion, from the one on the right where values of the two velocities are the same, and the position of
ignition takes place. Some ‘humps’ are also present that the flame front, therefore, stays the same.
correspond to zones in which bursts of cold flames take In order for the combustion process in the initial mixture
place. The slow combustion zone and the ignition zone are, to start, the mixture must be brought to ignition temperature
in turn, separated by a narrower region, known as the zone of Ti. Only then will the retroactive mechanism take place
the blue flames, which are warmer than cold flames. whereby the heat released in the zone of the reaction, which
Based on the experimental data that has become flows in the direction opposite to that of the flow of the gas,
available, the ignition diagrams of various hydrocarbons preheats the mixture to the temperature required for the
have been drawn. Although they all have the same overall exothermic reaction of combustion to take place
shape, they differ from one another in their detail, in spontaneously. Over a century ago, Mallard and Le Châtelier
particular in the number and shape of the humps gave a tentative description of the process which, though
characterizing the cold flames. In a certain sense, these somewhat approximate, captures the phenomenon’s essential
diagrams describe the morphology of the combustion aspects. If the thickness of the zone in which the reaction
processes that affect distinct substances and constitute useful takes place is indicated by d the equivalence between the
tools for identifying their characteristics. heat flow that moves backwards and the heat absorbed by the
current that feeds the reaction leads to the following
Premixed laminar flames equation:
Let us assume that we have a gaseous mixture at T f − Ti
temperature T0 that contains a fuel (for example methane) [44] kT
δ
(
= Gc p Ti − T0 )
and oxygen, and that it flows with laminar motion at a rate u0
along the z axis of a cylindrical conduit. The fluid’s particles where Tf is the temperature of the flame, kT is the thermal
move in a regular manner parallel to the axis of the cylinder conductivity of the fluid mixture, cp its average specific heat
and with a parabolic rate profile that starts from a value of and G the mass flow per surface unit which, in turn, can be
zero next to the wall of the cylinder and reaches a peak at its expressed as the product of the rate of flow of the gas and its
centre. In a simplified, and idealised, scenario it is preferable density, G⫽ur. The product of the average reaction rate r̄
not to take into account the effects of the walls of the conduit and the average molecular weight of the mixture M ᎐, yields
the reactive mass transformed per unit of time and of
volume. If it is divided by the density r, the inverse of the
average reaction time is obtained which when multiplied by
reactants the thickness d of the zone in which the reaction occurs
temperature, concentration

temperature yields the rate of the flame propagation, Su:

rM δ
Tf [45] Su =
r
q˙ whose value under stationary conditions is the same as that
of the rate of flow of the fluid. If one derives d from the
preceding expression and replaces it in equation [44], the
following equation results:

T0
H2O
intermediates
Su =
(
1 kT T f − Ti rM )
)
[46]
z
r (
c p Ti − T0

Fig. 6. Changes in temperature as a function The fact that Su is proportional to the square root of the
of the coordinate z in a premixed laminar flame. The figure rate of the reaction is the most significant result that has
also shows the curves that indicate the changes been achieved by the theory of premixed flames, even if the
in the concentration of reactants, intermediates and water preceding equation actually produces only approximate
produced by the reaction. Finally, the curve indicated solutions.
. A more refined model to describe the behaviour of
by q represents the amount of heat released by the combustion
reaction per unit of time. premixed flames, known as the ZSF model after the initials

420 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

of the Russian scientists who developed it, Zeldovich, The preceding is a differential equation of the second
Semenov and Frank-Kameneskij, was produced in the first order, non-linear because of the presence of Arrhenius’s
half of the Twentieth century. If one derives the heat balance exponential term, which prevents an analytic expression to
of a reactive mixture over a volume whose cross-section is describe variations in temperature as a function of z from
equal to that of the mixture’s average flow and whose being obtained. It is, therefore, preferable to identify a value
thickness is infinitesimal, with an axis aligned with the for the coordinate z⫽0 somewhere along the curve of the
direction of the flow of the gas, it can be easily demonstrated flame’s temperature, as illustrated in Fig. 7, where the
that, under stationary conditions, the following equation temperature is assumed to reach Ti. Values of zⱕ0 are in the
holds true: preheating zone where the reaction, for all intents and
dq purposes, does not take place, so that equation [49] can be
[47] = QR r simplified by omitting its third term. In this form, the
dz equation can easily be integrated by setting T⫽Ti as limit
where q is the heat flow per unit of time and per surface unit conditions for z⫽0 and that the temperature will tend
perpendicular to the direction of the flow, which can be towards T0 when z⫺⫺⬁. By setting a⫽G cp ⲐkT , the
䉴

expressed as the sum of two factors, i.e. the amount of following is obtained:
transport associated with the convection of the fluid current
and the amount of thermal conductivity:
[50] ( )
T = T0 + Ti − T0 eα z

dT which describes the trend of the temperature in the preheating

[48] q = c pGT − kT zone. Values of zⱖ0 are in the reaction zone where the
dz temperature increases from Ti to the final value Tf and the
A constant average value has been attributed to the amount of heat released is much greater than that associated
mixture’s specific heat in the expression of the first term, with the convective flow. The latter, therefore, can be ignored
while the second contributing factor is provided by the and equation [49] can be approximated to the following:
product of the coefficient of thermal conductivity kT and the
d 2T
temperature gradient, and moves in the direction opposite to [51] kT = QR Ae− E RT
that of the convective flow. By substituting [48] in [47] and dz 2
assuming a kinetic equation with an order of zero as the rate The preceding equation is also impossible to resolve
of the reaction one arrives at the following equation: analytically. It can be observed, however, that the
corresponding temperature behaviour that connects Ti with
d 2T dT Tf depends exclusively on the physical properties of the
[49] kT − c pG − QR Ae− E RT
=0
dz 2 dz mixture kT and QR as well as on the kinetic parameters A and
E, so that, if these are defined, the curve presents a unique
trend, as shown in Fig. 7 B. If z⬍0, on the other hand, there
A exists a whole family of curves, each of which corresponds
T
to a different value of the parameter a. For z⫽0, the curves
Ti Tf corresponding to the preheating zone and to the reaction
zone, in addition to intersecting when the temperature is Ti,
will also necessarily have the same tangent. This condition,
zone II as the Figure illustrates, can be attained only for a particular
value of a which, since cp and kT are given, will determine a
well-defined value of G, i.e. the rate of the gas flow, which
zone I therefore becomes an eigenvalue imposed upon [49] by its
T0 limit conditions. By simplification the following is obtained:
2
0 z 2 kT 1 Ae E RT RT f
[52] Su =
c p r C0 T f − T0 E
B
T Tf By means of the preceding formula the rate of
propagation of a laminar flame can be calculated starting
from the physico-chemical parameters of the combustible
Ti compound. Generally, Su values are higher for the alkenes
than for the alkanes, and higher for the alkines than for the
T0 alkenes, and decrease when the ramifications in the structure
increase.
0 z
Fig. 7. Temperature profile of a premixed flame. General treatment of combustion phenomena
In A two zones are visible: the first corresponds to preheating (I) In the ZSF model, the combustion process is described
and extends to the ignition temperature Ti, while the second by means of a single reaction, whereas in reality it takes
corresponds to the proper reaction (II) and continues place through complex reaction schemes in which a large
from that temperature. As shown in B, when Ti is reached, number of intermediate species participate. As we now have
the curves relative to the two zones must have a common available a variety of data on the chemical characteristics
tangent, whose slope enables the a parameter in equation [50] and kinetic parameters of these reactions, thanks to
to be derived. supercomputers it has become possible to carry out highly

VOLUME V / INSTRUMENTS 421

COMBUSTION AND DETONATION

detailed calculations to simulate the evolution in the In stationary cases the first term of the first member is
concentration space, due to chemical reactions and processes obviously nil. This equation can easily be applied to the total
of diffusion, of the combined species involved. mass, to the various components and to the energy by using
As already seen, in a combustion process the various the flow formulas summarized in Table 4.
transformations at work take place at the same time, and For reasons of simplicity, in the definition of the energy
proceed along complex patterns of successive and parallel balance only the contribution associated with the thermal
reactions, such as the one pertaining to methane illustrated in state has been taken into account, while those associated
Fig. 2. It is, therefore, useful to distinguish each reaction by with the flow of radiation and the work due to viscous stress
means of an index k, identifying with vi,k the stoichiometric have instead been ignored. The first of these latter terms, in
coefficient of the generic component i. In this manner, the fact, is small for non-luminous flames and significant only
rate at which a given component is formed as a result of the for those that are fuel-rich and contain blazing particles of
effect of all the reactions in which it participates can be coal that radiate energy. The second term is instead often
expressed by the equation: negligible if the rate of flow of the gas does not reach very
[53] ri = ∑ rkν i ,k high values. The parameter u*⫽(兺i Ni)P/RT represents the
k molar mean velocity derived by expressing the density by
Once this has been established, the general treatment of means of the equation of state of perfect gasses, while Di,m is
combustion processes in gaseous systems subject to laminar the average diffusion coefficient of the component i in the
motion proceeds through the application of the laws of the mixture, which can be determined starting from the values of
conservation of motion, energy and matter, whose values per the binary diffusion coefficients Di, j:
unit of volume will be generically indicated with the symbol 1 − yi
f. The symbol F, instead, indicates the amount of flow per [55] Di , m =
 y 
surface unit, associated both with the laminar motion of the ∑  D i 
gas and with transport processes which, in this approach, j ≠i i, j

pertain solely to chaotic molecular motion. It should be Actually, the preceding equation provides only approximate
observed that this flow constitutes a vectorial quantity as it values since it was derived assuming that the values of the
depends both on the rate u and on the local values of the various binary coefficients are comparable to one another.
gradients of the intensive quantities that characterise the The balance in the quantity of motion is to be achieved
physical state of the system, such as the molar bearing in mind that its flow is made up of convection
concentrations, Ci of the various components and the associated with the motion of the fluid, of a contribution
temperature. The balance of f is obtained by equating its from pressure forces, and of a dissipative term connected
build-up in a container of a given volume to the difference with the internal resistance resulting from the viscosity of
between incoming and outgoing flows plus any amount of it the fluid itself. Said balance leads to the Navier-Stokes
that may have been generated per unit of time and volume, equation, whose integration, carried out for the most part
.
which is indicated with the symbol s. In the analysis that numerically, provides the profile of the rate of flow of a gas
follows the flow of the generic component i measured in as a function of time.
moles per unit of time and surface is indicated with Ni, The values of typical quantities of molecular transport
which multiplied by the molecular weight Mi yields the flow that recur in the preceding equations, such as viscosity m,
in terms of mass Gi per unit of surface. The composition of thermal conductivity kT and the binary diffusion coefficients
the gas, moreover, will be characterized, respectively, both Di, j for each pair of components i and j, can be accurately
by its molarity Ci and by the mole fractions yi or mass calculated, on the basis of data on the size of molecules and
fractions wi, which are directly correlated to one another on the intermolecular forces at work between them by means
(wi⫽yi Mi Ⲑ兺j yj Mj). Molarities and mole fractions are related of the kinetic theory of molecular transport, that in its most
to each other via the simple formula Ci⫽r̃yi , where r̃ is the advanced form links back to the integration of Boltzmann’s
molar density that, in general, is calculated using the equation. As a first approximation, if the the kinetic theory
equation of state of perfect gasses r̃⫽PⲐRT. That being of gasses is applied, it follows that (kT Ⲑcpr)⬇(mⲐr)⬇D⬇lc,
stated, it is easy to demonstrate that the general balance where l is the mean free path of the molecules and c is their
equation of f will have the following form: mean velocity as a result of thermal motion.
It is worth noting that if one assumes as a first
∂f . approximation 兺iGi cp,i⬇Gc̄p, the following energy balance
[54] + ∇ ⋅Φ = σ
∂t equation is easily derived by applying [54] to the heat flow q:

Table 4. Variables used in balance equation [54]

f F s⭈
Total mass r ru 0

i Component Ci N i = u*Ci − Di ,m∇Ci ∑r νk i ,k

k

Energy (
rc p T − T0 ) ( )
q = ∑ Gi c p ,i T − T0 − kT ∇T ∑(−∆ f )
H
i rkν i ,k
i k

422 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

[56] ( )
∇ ⋅ kT ∇T − G c p ∇T + ∑ − ∆ f H i rkν i ,k = rc p
k
( ) ∂T
∂t
Actually, these equations only determine the relative
flows of the various components and, therefore, belong to the
category of boot-strap equations that are mutually
For a one-dimensional stationary system in which a interdependent without having any explicit connection with
premixed gas is present, if it is assumed that only one the fluid-dynamic behaviour of the system as a whole. They
reaction takes place, the preceding equation becomes equal must, therefore, be associated with a further equation that is
to [49]. compatible with an adequate reference system moving at
If [54] is applied to the total mass, the matter continuity molar mean velocity in the mixture and for which the
equation is obtained. Analogously, this equation may be formula 兺i Ni⫽0 is valid.
applied to the flow of the individual components; however, As a whole, it becomes a system of non-linear
since the average diffusion coefficients obtained from [55] differential equations. Given the initial and the limit
are only approximate, it is preferable to maintain the balance conditions, their integration can only be carried out
of each component in the form: numerically using extremely powerful computers and
∂Ci employing the kinetic parameters of the various elementary
[57] + ∇ ⋅ N i = ri = ∑ rkν i ,k reactions and coefficients of molecular transport.
∂t k In the particular case of a premixed flame, the results of
and to deal with the thorny problem presented by the the calculations have been compared to the experimentally
mutual inter-diffusion of the various species present in verified trends of the concentrations of reagents along the
the mixture by means of the Stefan-Maxwell equation, axis of a flow reactor. As Fig. 8 illustrates, the results of
which links the local gradients of the compounds calculations performed for the case of methane are very
expressed using the mole fractions yi to the flows of the much in agreement with the experimental data evaluating
individual components: both the temperature (Fig. 8 A) and the chemical behaviour

[58] ∇yi = ∑
(y Ni j )
− y j N i  RT 
(Fig. 8 B), a fact which highlights how our understanding of
the phenomena associated with the homogeneous
Di , j  P  combustion of hydrocarbons has reached a satisfactory level.
j
Analogous confirmations based on experimental data have
also been obtained for other hydrocarbons such as ethane,
2,000 propane and so on.

Premixed turbulent flames

1,500 Turbulence is characterized by the irregular random
motion of the particles of a fluid that occurs over and beyond
T (K)

the average uniform motion of the gaseous current. A

description of the phenomenon, advanced by the Nobel
1,000
Prize-winning physicist, Lev Landau, traces its origin to the
presence of broad, stable vortices with anisotropic
characteristics, that iron themselves out, splinter into pieces
500 and give rise to smaller vortices that are both faster and
isotropic; the latter then become smaller and smaller until
their energy dissipates into the kinetic energy of molecular
0 1 2 3 4 motion. The transition from laminar to turbulent motion
A z (mm) takes place abruptly when the dimensional group known as
the Reynolds number, expressed by the equation:
0.20
ulr
0.90 [59] Re =
O2 H2O µ
0.15 where l represents a typical dimension of the system, exceeds
0.85 a certain critical value, which for a fluid flowing in an empty
tubular conduit is in the order of 2,300. Although many of
yO2
yi

0.10 the phenomenological aspects of turbulence are known, there

CH4 0.80 is still no satisfactory general theory on the subject, so that
CO2 transport processes can only be described with the aid of
0.05 adequate models. In the most commonly used of these, the
0.75 field of turbulent velocity is split into a slower component
CO that acts on a large scale, and a faster, fluctuating
component that acts on a small scale. The effect of the
0 1 2 3 4
faster component on the field of slow motion creates turbulent
B z (mm)
viscosity, whose mechanism is similar to that of molecular
Fig. 8. Comparison between the experimental values collisions in laminar motion. Unfortunately, there is no clear
(represented by the dots) and the corresponding calculated criterion according to which the two scales in question may
curves of the temperature (A) and chemical composition (B) be distinguished from each other. To overcome this limitation,
along the z axis of a flow reactor for a mixture of methane Ludwig Prandtl introduced the concept of mixing length,
and oxygen that undergoes combustion. analogous to that of the mean free path of molecules,

VOLUME V / INSTRUMENTS 423

COMBUSTION AND DETONATION

on the basis of which it is assumed that vortices travel a

typical distance before becoming integrated in, and trading high temperature
their specific properties with, the surrounding fluid. (2,200°C)
According to a more modern approach, justified by the
hypothesis that there exists an invariance of scale in
turbulent vortices, the motion of the fluid is described by
means of the usual conservation equations but employing
transport parameters that reflect the motion’s turbulent
characteristics. In particular, in a model denominated e-k,
viscosity is expressed as a function of two parameters that
represent, respectively, the energy’s rate of dissipation and
turbulent kinetic energy. low temperature
The study of combustion in gaseous mixtures subject to (100°C)
a turbulent flux is very important both scientifically
speaking and for its applications. Turbulence increases a
flame’s rate of propagation and intensifies combustion Fig. 10. Geometrical and thermal configuration
processes. Moreover, it facilitates the exchange of heat of a hydrogen-based laminar diffusion flame, simulated
released by the flame front, and as a result also improves the by numerically integrating material and thermal balance
ignition process. equations. The temperature increases from the value
Since a laminar flame presents a flat front that it has in the environment (blue) to the maximum value of about
propagates at a rate Su, it is important to understand in depth 2,200°C at the core of the flame (red).
the factors that render it unstable and create the conditions
necessary for its degeneration towards a turbulent state. The
first physico-chemical analysis of the stability of laminar tackle the problem, because it ignores the effects of
flames was carried out by Landau with the aim of phenomena relating to the transport of matter and heat,
investigating the response of the flat front of the flame to which are the real protagonists of combustion processes. In
small wrinkles of amplitude x that perturb its surface. The an approach that is diametrically opposed to the preceding,
method applied constitutes a classic example of analysis of Zeldovich arbitrarily assumed for the sake of simplicity that
the stability of dynamic systems, in which the temporal in the passage from unburnt to combusted gasses the density
evolution of the wrinkling is expressed by means of an remained the same, though taking explicitly into account
exponential function of the type: processes of exchange of matter and heat. The model’s
[60] ( )
ξ = f x, y eγ t
mathematical developments highlight the fact that a laminar
flame will contain zones of stability because its wrinkling
where f (x,y) is a function of the two coordinates that define locally increases the rate of reaction and depletes the
the flame’s plane, whose specific characteristics are of no reagent. A diffusive retroactive effect thus occurs that tends
concern. An important role is instead played by the sign that to give the flame a flat configuration once again.
pertains to the parameter g: when it is positive, x increases If the characteristics of unstable solutions are examined
exponentially and the flame is unstable; when it is negative, further, it can be seen that the system’s evolution gives rise
x tends asymptotically towards zero and the flame is stable to cellular configurations such as those illustrated in Fig. 9.
since, after having experienced perturbation, it returns to its These configurations were obtained by simulating the
original configuration. In Landau’s analysis, only behaviour of unstable flames with the aid of a computer
fluid-dynamic aspects were taken into account and processes following a model that is more general than the preceding
of exchange of heat and matter were ignored. Flames were model and in which both fluid-dynamic and
described in terms of a discontinuity in the density of the thermo-diffusion phenomena are taken into account. The
gas, which is greater for an unburnt mixture at a low existence of such cellular structures has been experimentally
temperature and, instead, less for a combusted mixture that confirmed in the flames of various hydrocarbons. Their
has reached a high temperature and that moves at rate Su. In formation represents the beginning of an evolution towards
this approach, the g parameter was always positive, which is the chaotic behaviour that leads to turbulence.
in blatant contradiction with experimental data that There does not yet exist an easily employable
irrefutably confirm the existence of laminar flames. This physical-mathematical approach that can describe the
shows that a mere fluid-dynamic analysis is inadequate to behaviour of a mixture subject to combustion in a fluid in
turbulent motion. The simulation of such phenomena
represents an real challenge that presumably will be met
thanks to the use of ever-more powerful computers, for
example through methods of direct simulation, whose
application, however, is currently limited to very simple
situations.
In a simplified semi-empirical approach it is assumed
that the flame front has a wrinkled configuration. It is also
assumed that the front moves at rate ut, which is greater than
rate ul belonging instead to the corresponding laminar flame.
Fig. 9. Characteristic examples of the structure If one indicates with Wf the surface of the wrinkled flame
of cellular flames derived from model-based calculations. front and with W that of the laminar flat flame, following

424 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

Fig. 11. Schematic diagram air

of the currents relative
to the inflow of fuel through
the nozzle and to the lateral r
inflow of air into a burner. u
fuel r0 u0

0
z

air

Damkoehler it can be assumed that: is also assumed that the specific heat of all the components
ut St Ωf is the same and that at every point of the mixture, which in
[61] = = any case is not uniform, transport properties satisfy the
ul Su Ω following equation derived from the kinetic theory of gasses:
Therefore, computing ut is the same as computing Wf . µ cp r
Various criteria for evaluating the relation expressed in the [62] =D=
preceding equation have been suggested; the most general r kT
criteria is related to the description of turbulent motion The velocities of consumption of the fuel rF and of the
through the concept of mixing length. oxygen rO, furthermore, are to be bound by the following
More recently, it has been suggested that Wf can be equation:
calculated by assimilating the turbulent flame front to a rF = γ st rO
[63]
fractal object whose characteristics, when viewed on
different scale, remain unvaried. where gst is the stoichiometric molar relation corresponding
to the combustion reaction. In this approach, it can be proved
Diffusion flames that the conservation equations [54], when written in
A typical diffusion flame is generated by a jet of gaseous cylindrical coordinates and referring to a stationary state,
fuel injected into a current of air that might also be will acquire the same form (where v is the radial component
oxygen-enriched. The peculiar characteristic of these flames of velocity):
is the fact that the rate at which fuel is consumed is
∂ ∂ ∂  ∂f 
determined by the rate at which it is diffused when it is [64]
∂z
(
rurf + )
∂r
( )
rv rf =  µr 
∂r  ∂r 
brought into contact with oxygen. To illustrate the process,
Fig. 10 shows a simulated geometric and thermal where f indicates the specific values, as they referre to the
configuration of a hydrogen flame. The configuration of the unit of mass, of the axial component of the quantity either of
flame front, where the gaseous current acquires an intense motion u or of thermal energy cp(T⫺T0), respectively. As far
luminosity, is characterized by a cylindrical symmetry. as the two reagents are concerned, a single quantity is used,
Injection systems of this type are very common as they are x, where:
employed both in the burners of industrial furnaces and in [65] ξ = YF − γ stYO
reaction engines. Internal combustion engines of the Diesel
type also operate through a series of transitory injections of The limit conditions are defined by the values of the
fuel into air that has previously been compressed in a composition and of the temperature at entry, along with the:
cylinder. Up until a few years ago, the study of diffusion u = 0
flames had elicited less attention than that of premixed 
[66] r → ∞ ξ = 0
flames, especially because of the greater difficulties that are T = T
encountered in carrying out quantitative experimental  ∞
investigations of these systems. Moreover, their description If we indicate by Re0⫽u0d0 mⲐr the value of the Reynolds
runs into the additional difficulty of having to establish a number at the jet’s point of entry, the general solution of [64]
particular rate of propagation, along the same lines as that can be expressed as follows:
which characterizes premixed flames.
A simplified approach is to refer to a stationary laminar ξz 1 T − T∞ 1 0.1875
[67] = =
r0 Re0 T0 − T∞ Re0
( )
2
jet injected into an air current having the configuration 1 + 0.0117 Re02 r f2 z 2
shown in Fig. 11, where u stands for the axial component of
the rate of propagation. It is also assumed that pressure where r0 is the radius of the fuel feeding conduit, z is the
remains uniform and that effects of the forces of gravity and axial coordinate, while rf is the value of the radial coordinate
flotation can be ignored. If a theoretical solution to the that locally defines the position of the flame front.
problem is attempted, it is best to refer to what is known as In an approximate formulation it is taken into account
the SCRS (Simple Chemically Reacting System) model, in that the rate of the combustion reaction is very high, so that
which the fuel and oxygen combine in fixed proportions. It it is assumed that the combination of oxygen and fuel takes

VOLUME V / INSTRUMENTS 425

COMBUSTION AND DETONATION

place instantaneously in those points where the relation The formation of the coal particles occurs via a complex
between the two components coincides with the process that is not yet understood in detail, but which in any
stoichiometric relation required for the combustion reaction, case involves a series of stages that have been identified as
in other words where yF ⲐyO⫽gst. In this approach, the rate at follows: nucleation, growth, oxidation and coagulation.
which the process takes place is determined entirely by the The least understood of these stages is the first, which
fuel’s rate of diffusion in the internal part of the flame and probably occurs through the formation of small molecules of
the oxygen’s rate of diffusion in its external part, both in the unsaturated hydrocarbons such as acetylene, that then
direction of the combustion front. The radius of the flame rf condense to form cyclic molecules. A possible mechanism
for every particular value of z is expressed by the following for the increase of the rings entails a series of reactions, the
equation: first of which leads to the formation of an aromatic radical
1/ 2 Ar⭈, and then to the reaction of a hydrogen atom with a
9.237  
1/ 2
rf Re r  molecule of the arene ArH:
[68] =  0.1875 0 0  − 1
z Re0  zf st   [73] ArH ⫹H⭈⫺Ar⭈⫹H2
䉴

 
where: The subsequent reaction of the Ar⭈ radical with an
(F O)st acetylene molecule leads to the formation of an unsaturated
[69] fst = product which, after further cyclisation, forms a
1 + (F O)st polyaromatic hydrocarbon:
(F/O)st being the stoichiometric relation between the mass of
C
fuel and of oxygen. CH
The length of the flame is determined, obviously, by
assuming that rf⫽0, so that one obtains:
C CH C
zf C2H2
3 Re
[70] =
z0 16 f st
which can be approximately expressed as follows: The evolution of the process gives rise to the formation
Q πr 2u of polyaromatic hydrocarbons of ever-increasing size which
[71] zf ∝ V ≈ 0 0 coagulate into clusters a few nanometres in size that make up
D D the soot. In an oxidizing atmosphere, the particles may be
where QV is the volumetric capacity of the fuel current and further transformed as they tend to react with O2, O⭈ and
D its diffusion coefficient. The preceding equation can be ⭈OH.
applied even if the flame is turbulent, as long as D is The kinetics of the particle coagulation process
replaced by the value of a coefficient of turbulent diffusion can be described using a second-order function, on the
that reflects the mixing process resulting from the basis of which the rate of the formation of particles of a
fluctuations of the fluid aggregates and can be determined, certain size is proportional to the product of the particle
for example, by applying the concept of mixing length. It concentrations that engender it. Through this growth
can thus be shown that it is approximately equal to u0r0, and process the particles attain sizes of the order of several
so it follows that: dozen nanometres.
r02u0 There are experimental data in the literature that refer to
[72] z ft ∝ = r0 the tendency of various hydrocarbons to form such
r0u0 particulate matter. Their correlation has been achieved by
Essentially, one establishes that the height of a turbulent differentiating premixed flames from diffusion flows and
flame is proportional to the radius of the nozzle, but defining for both a numerical index that characterizes the
independent of the rate of flow and volumetric capacity of threshold of the particulate matter formation, known as TSI
the fuel. This is a significant result whose validity has been (Threshold Soot Index).
verified experimentally and which has important applicative
implications.
The variation of the height of a diffusion flame is thus 7.1.2 Heterogeneous combustion
expressed by [71] if the flame is laminar and by [72] if it is
turbulent. Evaporation and combustion of liquid drops
Various devices and plants of applicative importance,
Formation of carbonaceous particles such as industrial kilns, furnaces and Diesel engines, are
It has already been pointed out in this treatment how supplied with liquid hydrocarbon fuels introduced in the
combustion can give rise to the formation of polluting form of a shower of droplets. In such cases the combustion
substances such as nitrogen oxides. Another source of process can be subdivided into two stages, the evaporation of
pollution is to be found in the production of particles of coal the liquid and its ensuing combustion in the gaseous phase.
that have diameters as small as a few dozen nanometres and It is clear, therefore, that the size of the droplets plays a basic
show up in the form of soot. Their formation is facilitated in role in the rate of the global process, as the rate of the
fuel-rich flames which acquire a pronounced luminosity evaporation process is proportional to the surface area of the
precisely because of the presence of these particles at high drop (in particular, therefore, to the square of its diameter if
temperatures. At times the particles will have a high content the drop is assumed to be round).
of condensed aromatic polynuclear hydrocarbons, which are Commonly the liquid is dispersed by means of an
potentially carcinogenic. atomizer, or spray, in which a rapid formation of drops takes

426 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

of dispersions, as illustrated in Fig. 12. In particular, three

10 zones are identified which may be characterized as follows:
zone I, or the Rayleigh zone, in which the droplets are highly
Ohnesorge number

unstable and tend to disintegrate; zone II, in which helicoidal

1 waves are observed before the disintegration of the droplets
starts; and zone III, in which disintegration into droplets
III
II takes place close to the orifices, due essentially to the
0.1
I turbulent motion of the fluid.
The distribution of the dimensions thus obtained is
0.01 mainly described by means of empirical relations, including:
 δ q
[77] CVFi = 1 − exp 1 −  i  
0.001
1 10 102 103 104 105
  δ 0  
Reynolds number in which CVFi is the cumulative volume of all the drops that
are smaller in diameter than di, d0 is a reference diameter
Fig. 12. Atomization zones of the droplets coming corresponding to a value of CVFi equal to 0.632 and q is an
from a nozzle calculated as a function of the Ohnesorge empirical parameter.
and Reynolds numbers. I, unstable droplets; II, helicoidal Calculation of the velocity of evaporation of a dispersion
waves are present; III, disintegration into droplets takes of liquid drops is mostly conducted following the tendencies
place close to the orifices of the nozzle. of each of them. The simplest case is that of a spherical drop
surrounded by a hot gaseous current. An important factor is
the transfer of heat in the liquid; to evaluate this transfer it is
place, and equally rapidly they become remixed with the necessary to analyze the motions that take place within the
surrounding air. There is a wide variety of vaporizers whose drops. Here it will be assumed that these effects are rapid
operation has been studied in relation to the size of the holes enough to be able to attribute a constant mean temperature
and the flow of liquid supplied. However, it is not easy to of Tl to the drop. If ml is the mass of the drop and cpl is its
foresee the characteristics of the dispersion thus obtained, specific heat, the energy balance may be put as:
above all due to the processes of disintegration of the drops
= ha (T − Tl ) + l L
dTl dm
and their coagulation. So, these assessments are mostly [78] ml c pl
conducted exploiting empirical relations and correlating the dt dt
data by means of three parameters or dimensionless groups, where T is the temperature of the gaseous mass, L is the heat
which are usually shown by the letter j, meaning that a jet of of evaporation of the liquid, a is the surface area of the
liquids is referred to: sphere and h is the coefficient of heat transport from the
rl uδ j surface of the drop to the gaseous mass. For a sphere having
[74] Re j = Reynolds number a diameter of d it results, neglecting thermal expansion for
µl the sake of simplicity, that the term of variation in time of
rgu 2δ j the mass of the liquid drop may be expressed as follows:
[75] We j = Weber number
dml rl πδ dδ (t ) rl πδ dδ 2 (t )
2
σ [79] = =
µl dt 2 dt 4 dt
[76] Oh = Ohnesorge number
rlσδ j If evaporation takes place in stationary conditions, the
heat of evaporation absorbed by the evaporating drop and the
where dj is the diameter of the drops before redistribution heat transferred to the drop from the outside are equal, so
among the various diameters has taken place, s is the surface that:
tension, u is the velocity of the gas, while the indices g and l
ha (T − Tl ) = −
dml
associated with the density (r) and viscosity (m) values [80] L
indicate the gaseous and the liquid phase, respectively. dt
On the basis of the values assumed by the aforesaid This shows that, with reasonable approximations, we
dimensionless groups, it is possible to identify various types may put:

Fig. 13. Schematic diagram

of a cylindrical chamber
where the combustion
air
of liquid droplets takes place.
fuel

air

0 z

VOLUME V / INSTRUMENTS 427

COMBUSTION AND DETONATION

dml dT 6 f β LHV
[81] = −πCi M i βδ [90] =− δ
dt dt 1 + f c pδ 03

where: If this equation is integrated, remembering [88] we

[82] β ∼ 2D(Tm ) obtain lastly the law that expresses the course of the gas
temperature as a function of the diameter of the drops:
Tm⫽(T+Tl )/ 2 being a mean temperature and D the diffusion
f LHV   δ  
3
coefficient of the vapour in the gas.
[91] T − T0 = 1 −   
Making [79] equal to [81] it is seen that: 1 + f c p   δ 0  

dδ 2 (t )
[83] = −4 β T0 being the gas temperature corresponding to z⫽0.
dt
which, integrated, supplies the following expression of time Combustion of solids
of evaporation: The combustion of solid particles will now be
considered, paying particular attention to coal and wood,
δ2
[84] t ev = 0
even if the analysis conducted is of general validity and may
4β be applied to any other combustible material. If a particle of
The description of the combustion process that takes coal is exposed to a hot gaseous current containing oxygen,
place in a generic burner supplied by a dispersion of for example air, it undergoes a process in which the
drops of a fuel obviously depends on the fluid-dynamic following three stages can be identified: loss of water by
conditions that develop in the combustion chamber. drying; loss of mass by the release of volatile gases; and true
These aspects fall outside of the present treatment. For combustion.
the sake of simplicity and for purposes of illustration, Concerning the first stage, it should be remembered that
the case considered here will be that of an atomizer that the water present in the particle can be considered as
produces a shower of monodispersed drops, i.e. with a subdivided into two forms: the free water contained in the
uniform diameter, propagated with a piston-type flow in pores of the solid and the water absorbed on the external and
a current of air along a cylindrical chamber, as illustrated internal surfaces of the particle. The heat balance of this first
in Fig. 13. stage can be formulated as:
If n0 indicates the number of drops of diameter d0 per
unit of volume at point z⫽0 of the axial coordinate, it is [92]
d 
(
dt  w p w
)
m c + mdf c p df T  =
 dt
dm
L+q
demonstrated that the fuel/air ratio ( f ) may be expressed as:
no rl πδ 03 / 6 where the subscripts w and df (dry fuel) indicate the water
[85] f= and the dry solid, respectively; q indicates the heat that is
r0 − n0 rl πδ / 6 3
0 transferred to the solid per unit of time and of mass. The
and therefore: value of q depends on the characteristics of the heating
f r0 6 system; if the process is carried out in a furnace, most of the
[86] n0 = heat is exchanged by radiation; q is in any case given by the
1 + f rl πδ 03 sum of two terms, as in addition to radiation there is also a
If it is assumed that the number of drops remains contribution due to convection. Bearing these aspects in
unchanged, although drops are subject to a process of mind, by integrating the preceding equation it is possible to
evaporation that diminishes their diameter, and that the assess the time required to dry the fuel.
density of the gas remains constant, it is reasonable to apply The second stage is associated with a pyrolytic
the following expression: process through which volatile products are liberated and
n0 r burn when they come into contact with the oxygen of the
[87] n= air. The velocity of the process depends naturally on the
r0 temperature, but it is also very sensitive to the nature and
Integratring [83], we obtain the following expression, the characteristics of the fuel. For example, in lignite the
which gives us the diameter of the particles as a function of process starts at 300-400°C, freeing carbon monoxide and
time: dioxide, and vapours of light hydrocarbons and hydrogen.
The velocity of the process, at a first approximation, is
[88] δ = δ 02 − 4 β t described by means of a first-order kinetic equation with

The heat balance of the gaseous current can then be

formulated by making the heat liberated by combustion from Table 5. Kinetic parameters for some solid fuels
the evaporated gas equal to that absorbed by the gaseous
current: A0, pyr Epyr
dT dm s⫺1 kJ/mol
[89] rc p = n l LHV
dt dt Lignite 280 47.3
where LHV (Lower Heating Value) indicates the combustion
heat per unit of mass of liquid fuel. Taking the preceding Bituminous carbon 700 49.4
relations into account, in particular [81] and [87], the Wood 7⭈107 129.7
following is therefore obtained:

428 ENCYCLOPAEDIA OF HYDROCARBONS

 HOMOGENEOUS AND HETEROGENEOUS COMBUSTION

in which kc is the coefficient of transfer of material, referred

Table 6. Kinetic parameters for some carbons to the concentrations. If this is divided by the factor RT, it is
possible to express the concentration of oxygen by means of
A0 E the partial pressures. The theory of the transport processes
gO (cm2⭈s⭈atm O2)–1 kJ/mol supplies the following correlation through which the value of
2
kc can be obtained:
Anthracite 20.4 79.5
kcδ
Volatile bituminous 66 85.2 [99] Sh =
D
(
= 2 + 0.6 Re p1/ 2 Sc1/ 3 ϕ )
Highly volatile bituminous 60 71.8 in which Sh is the Sherwood number, D is the coefficient of
Subbituminous 145 83.6 diffusion of oxygen in the air, Rep is the Reynolds number
evaluated on the basis of the size of the particles and Sc is
equal to m/rD; /, lastly, is a factor between 0.6 and 1 which
constant kpyr, which when integrated supplies the allows for the influence of the gases produced by
following expression: combustion.
If reference is made to situations corresponding to low
 m p − mc − ma  values of the Reynolds number, in which / may be attributed
[93] ln   = − k pyr t a unitary value, it follows that kc⬇2D/d and resolving [98]
 m pi − mc − ma 
with respect to pO(s) we obtain:
2
in which the mass of the volatile substance is expressed as (2 D / δ RT ) pO
the difference between the mass of the particle (mp, initial [100] p(s)
= 2

value mpi), the mass of the carbonaceous fuel (mc, from

O2
(k / M O ) + (2 D / δ RT )
char) and the mass of the residual ash (ma, from ash). The Substituting [100] in [97] and remembering that a⫽pd2
dependence of the constant of the velocity of reaction on the we obtain:
temperature is expressed by means of the Arrhenius law; the
dmc M (2 D / δ RT ) pO2
typical values of the Arrhenius parameters for a number of [101] = − k πδ 2 C
solid fuels are set out in Table 5. dt M O (k / M O ) + (2 D / δ RT )
We now consider the combustion of impoverished dry To integrate this equation and to obtain the time required
coal, i.e. without its less stable substances. The reaction to burn the particle it is necessary to find a link between the
takes place through the interaction of the oxygen molecules mass of the particle and its diameter. The simplest case is
coming from the gaseous phase with the surface carbon one in which it is assumed that the combustion process
atoms, and leads to the formation of carbon monoxide and occurs at the surface of the particle, leaving the density of
dioxide through the reactions: the solid constant. In this case, for any value of the diameter
[94] C ⫹1/2O2⫺ CO 䉴 the relationmc⫽pd3rc Ⲑ6 applies, so that d⫽(6mc Ⲑprc)1Ⲑ3.
If the operating conditions are those in which the
[95] C ⫹O2⫺ CO2 䉴
velocity of the global process is limited by the velocity of
Moreover, if water is present the following reaction takes the surface reaction, so that 2DⲐdRT⬎⬎kⲐMO, we obtain:
2/3
place:  6m 
dmc 12
= −π  c
16 O2
[102] kp
[96] C ⫹H2O⫺ CO ⫹H2 䉴
dt  πrc 
The most important of these reactions is [94], which from which the following expression of the combustion time
conditions the global process. At a first, reasonable is obtained:
approximation its velocity is expressed by means of a first- rcδ 0
order kinetic equation, in which the values of the Arrhenius [103] t=
parameters for a number of types of coal are set out in 1, 5 kpO2
Table 6. Allowing for the dimensions of the constant of the Vice versa, if the velocity of the global process is limited
velocity reaction (k), the balance of a coal particle, assumed by the diffusion, we obtain:
to be round with a diameter d, can be put: 1/ 3
dmc  6m  DpO2
dmc M [104] = −2 π  c  Mc
[97] = − ka C pO( s2) dt  πrc  RT
dt MO from which the following expression of the combustion time
in which a⫽pd2 is the surface of the sphere pO(s), is the is obtained:
2
partial pressure of the oxygen corresponding to the solid rcδ 02 RT
surface and MC /MO is the relation between the atomic [105] t = 0.785
weights of carbon and of oxygen (12/16). DpO2 M c
In stationary conditions the quantity of oxygen
consumed through the effect of the chemical reaction equals
the quantity of oxygen associated with the diffusive flow Bibliography
from the heart of the gaseous phase to the surface of the
particle, so that: Barnard J.A., Bradley J.N. (1984) Flame and combustion, London,
Chapman & Hall.
[98]
k ( s ) kc
p = p − pO( s )
M O O2 RT O2 2
( ) Borman G.L., Ragland K.W. (1998) Combustion engineering, Boston
(MA), McGraw-Hill.

VOLUME V / INSTRUMENTS 429

COMBUSTION AND DETONATION

Glassman I. (1996) Combustion, San Diego (CA), Academic Press. QR heat of reaction
Jones J.C. (1993) Combustion science: principles and practice, Q⭈⫾ heat released or absorbed per unit of time
Newtown, Millennium. R gas constant
Kanury M.A. (1985) Introduction to combustion phenomena, New r rate of reaction (moles transformed per units of
York, Gordon & Breach. volume and time)
Kuo K.K. (1986) Principles of combustion, New York, John Wiley. Su flame propagation rate
Lewis B., Elbe G. von (1987) Combustion flames and explosions of t time
glass, New York, Academic Press. T thermodynamic temperature
Williams F.A. (1985) Combustion theory: the fundamental theory of u axial component of a gas rate of flow
chemically reacting flow systems, Reading (MA), Persuy.
u* molar mean velocity
v radial component of a gas rate of flow
V volume
List of symbols yi molar fraction of component i
Yi mass fraction of component i
Boldface type indicates a vector, while ∇⭈ stands for the
divergence operator. A line drawn above a generic quantity g
3 z axial coordinate
indicates its average value
Greek letters
A pre-exponential factor of the reaction-rate constant Df Hi molar enthalpy of formation
a adimensional heat-exchange coefficient DH° variation in standard enthalpy associated with a
Ci molar concentration of the component i reaction
C (s) molar concentration at surface DrG° variation in standard free energy associated with a
cp specific heat reaction
c̃p molar heat capacity F flow per unit of surface of a generic size
Di, j binary diffusion coefficient between components i m viscosity
and j d drop or particle diameter
Di,m diffusion coefficient of the component i in a d0 initial value
gaseous mixture vi,k stoichiometric coefficient of component i in kth
E activation energy reaction
G mass flow per surface unit r density of a fluid
Gi mass flow of component i r̃ molar density
h heat-exchange coefficient (calories per unit of time ÿ adimensional temperature
and surface) t adimensional time
k reaction-rate constant
kc coefficient of transfer of material Sergio Carrà
kT thermal conductivity Dipartimento di Chimica, Materiali
Mi molecular weight of component i e Ingegneria chimica ‘Giulio Natta’
Ni molar flow of component i per surface unit Politecnico di Milano
pi partial pressure of component i Milano, Italy

430 ENCYCLOPAEDIA OF HYDROCARBONS