Energy and Combustion Science

CHIGIER

Introduction

POLLUTION FORMATION AND DESTRUCTION IN FLAMES—INTRODUCTION

NORMAN A. CHIGIER,     Department of Chemical Engineering and Fuel Technology, University of Sheffield

Publisher Summary

The energy crisis of 1974 followed by the fivefold increase in the price of oil emphasized the need for fuel conservation; therefore, the solutions proposed for the reduction of emission of pollutants must also result in high combustion efficiency. This chapter discusses the formation and destruction of pollutants in flames. Most forms of combustion involve the burning of the principal fossil fuels, oil, gas, and coal. High temperatures are associated with the combustion process; therefore, both major products of combustion and pollutant species are formed. The redesign of combustion systems can lead to avoidance of emission of pollutants even for low grade fuels. Most combustion systems involve the reaction of hydrocarbons in the fuel and oxygen as a constituent of air. Certain ways have been suggested in which the efficiency of burning fossil fuels can be radically improved. Principally, these involve burning under premixed lean conditions so as to increase reaction rates without increasing temperature.

THE major cause of air pollution is the emission of pollutants from combustion systems burning fossil fuels. These arise in the combustion of oil, coal and gas in the furnaces of electricity generating power stations and industrial plant as well as in automobiles, trucks and aircraft. In large industrial complexes and urban areas, deterioration of the environment due to pollution has reached such high levels that the largest cities in the world have been threatened with suffocation. Damage to property and landscape has reached such high proportions that the public has demanded a reduction in emission of pollutants. Under pressure from environmentalist lobbies, legislation has been introduced in the United States and other industrialized countries restricting the quantity of pollutants emitted from engines and industrial plant. The restrictions have been imposed with increasing severity and huge financial resources are being expended on changing engineering design, as well as monitoring and controlling the emission of pollutants.

The environment in many major cities in England has been radically improved simply due to the control of domestic coal fire burning. Fog and smog levels have been reduced, sunshine has been increased, bird and plant life have reappeared and the quality of life has improved. In the largest Californian cities, where automobile exhaust fumes are the main cause of pollution, the rates of increase of levels have been reduced and can be expected to decrease in the near future. Some of the initial attempts at reducing pollutants from automobiles resulted in a decrease in combustion efficiency. The energy crisis of 1974 followed by the fivefold increase in the price of oil emphasized the need for fuel conservation so that solutions proposed for the reduction of emission of pollutants must also result in high combustion efficiency.

Volume 1 of Progress in Energy and Combustion Science is devoted to the formation and destruction of pollutants in flames. This Introduction serves the dual purpose of giving a general introduction to the subject as a whole, as well as introducing the individual contributions which follow the Introduction. The aim is to set the scene for the reader of the complete volume, and at the same time provide an article for the general reader who does not wish to be concerned with a great deal of mathematical and scientific detail. Reference is only made to authors in this volume. Comprehensive sets of references and diagrams are contained in each individual contribution.

The contributions commence with a very broad discussion of the overall problem by Weinberg, who examines the general philosophy of the efficient management of combustion—set in an historical context. This is followed by a series of fundamental papers dealing with the kinetics, physical processes and mathematical modelling associated with chemical reaction, fluid flow and heat transfer in flames. Bowman covers the chemical kinetic aspects, Caretto introduces and discusses the equations used in mathematical modelling. Pratt analyses mixing and reaction concepts as used in chemical reactors and applies them to combustion systems, while Bilger looks specifically at turbulent jet diffusion flames which play such an important role in practical combustion systems. The more practical and applied problems are introduced in the second half of Volume 1. Gas turbine engine design is critically examined by Mellor, and Heywood discusses in detail the problems facing design engineers in the automobile industry who are required to satisfy both the requirements of improved fuel efficiency and reduction in emission of pollutants. The special problems associated with diesel engines are examined by Henein.

Weinberg shows that pollution has always been associated with burning ever since man discovered the use of fire for heating, cooking and subsequently for industrial use. In our modern technological society, combustion plays such an important role that it is found in almost all aspects of society, both industrial and domestic. Most forms of combustion involve the burning of the principal fossil fuels, oil, gas and coal. Because of the high temperatures associated with the combustion process both major products of combustion and pollutant species are formed. Re-design of combustion systems can lead to avoidance of emission of pollutants even for low grade fuels. Most combustion systems involve the reaction of hydrocarbons in the fuel and oxygen as a constituent of air. The air is usually forced into the combustion chamber so that jet and wake flows occur and the mixing between fuel and oxidant gases plays an important role in controlling the flame characteristics. Weinberg has made novel and pioneering contributions to the effects of electric fields on flames and he shows how these can lead to a reduction in carbon formation. He concludes that there is no forseeable likelihood of doing without burning fuels during the next half-million years. He also emphasizes and suggests ways in which the efficiency of burning fossil fuels can be radically improved. Principally these involve burning under premixed lean conditions so as to increase reaction rates without increasing temperature.

PRINCIPAL POLLUTANTS

Emissions from engine exhausts and chimney stacks are mainly in the gaseous phase with small quantities of solid particulate matter held in suspension. The gas constituents are made up of inert gases which have been passed through the combustion chamber unchanged, products of combustion and unburned fuel and oxidant. The solid particulate matter arises from the fuel and consists of metallic compounds and other materials which cannot burn and hydrocarbons which were not completely burned in the flame. Under very poor combustion conditions, liquid fuel may be emitted when full throttle is used during rapid accelerations. Since pollution is defined as the contamination of man’s environment, all matter emitted from combustion chambers that alters or disturbs the natural equilibrium of the environment must be considered as a pollutant. The major concern is to prevent damage to human, animal and plant life and generally to create no artificial disturbance to the natural ecosystem.

Damage to the ecosystem by pollutants can be due to changes in the concentration of gas components in the atmosphere and the deposition of particulates and chemically reactive species. This damage can be direct, e.g. the change to life by toxic gases, the suffocation of plant life by particulate matter and the corrosion of metal and contamination of drinking water produced by chemical reactants. In addition to these direct consequences of pollution, there are many equally severe indirect effects, e.g. the formation of smog due to photochemical reaction and the increase in fog formation and rainfall due to particulate matter. When heat is released in sufficient quantities, thermal pollution leads to changes in the local temperature gradients causing inversion layers in cities while relatively small changes in temperature of water and air can be harmful to temperature-sensitive life systems.

Pollutants can arise both as products of combustion and also as impurities in the fuel which are not burnt during the process of combustion. Unburned hydrocarbons arise from the incomplete combustion of fuel due to the fuel rich operation or inadequate mixing of fuel and air, as does the formation of carbon during the combustion process. Oxides of nitrogen are formed during the combustion process, the source of nitrogen either being from nitrogen contained within the fuel or fixation of nitrogen from the air supply for the combustion process.

The four principal chemical classes of pollutant species emitted from combustion sources are nitrogen oxides, carbon monoxide, organic compounds (unburned and partially burned hydrocarbons), and sulphur oxides. As part of the combustion process, pollutant species are formed by chemical reaction and subsequent chemical reaction can also lead to their destruction or elimination. The formation of carbon monoxide, soot and other pollutant species, may be controlled to constitute intermediates within the combustion system, which can be destroyed by chemical reaction prior to emission from the system. At the temperatures achieved in most flames, the presence of fuel and oxidant leads inevitably to the formation of some pollutants. These high concentration levels may be reduced by further chemical reaction as the temperature is reduced, but quenching leads to the freezing of certain species, resulting in high emission rates. Pollution formation, destruction and control is thus intimately connected with the combustion process. The general aims of maximizing combustion efficiency and minimizing pollutants are conflicting, since combustion efficiencies are maximized at, or close to, stoichiometric mixing conditions where the highest temperatures are achieved. These high temperatures lead to formation of high levels of pollutants. Optimization can only be achieved by very careful control of air-fuel ratio and temperature levels throughout the system.

Particulates

The particulate matter emitted from combustion chambers has three possible sources: (i) matter which was not combustible; (ii) matter which was capable of being burned but was not burned; and (iii) matter formed during the process of combustion. Temperature conditions in most combustion chambers are sufficiently high for vaporization of liquids so that, except for conditions of very rich low temperature burning, all emitted particulate matter will be in the solid phase. Particulate matter can be deposited on surfaces within the combustion chamber or may be emitted with the exhaust gases. Most pollution problems arise from particulates which are sufficiently small that they are held in suspension as they are transported by the exhaust gases. The particulate matter may be clearly visible as smoke or water vapour, but small concentrations of particles can be significant sources of pollution without being clearly visible.

Almost all hydrocarbon fuels contain traces of metals and other solid matter, which normally do not burn and pass through the combustion chamber as an element or, more generally, as a metal oxide. Most of these metals are capable of burning and are even used as fuels in rockets and other high temperature devices. Carbon has a melting point of 3813 K, yet solid carbon burns easily in air, due to its low thermal conductivity. Most of the particulate matter emitted from practical combustion devices is solid carbon. Solid carbon particles formed in flames are known as soot. Soot can be formed from purely gaseous fuels and is more commonly formed when liquid fuels are used. Carbon particles are the main cause of luminosity of flames and the emission from the yellow region of flames is due to carbon. For flames that rely on radiative heat transfer, attempts are thus made to promote the formation of soot particles. In automobile engines and other systems with water-cooled surfaces, coking, due to the deposition of hard carbon deposits, can readily arise when fuel is allowed to come into direct contact with the cooled surfaces. In gas turbine combustion chambers, formation of solid particles is generally considered to be undesirable because radiative heat transfer from the flame to the combustor-can walls would require additional cooling and also the presence of particles causes damage to gas turbine blades, either by direct impingement of the particles, causing pitting of the blades, or by deposition on the blades, leading to subsequent corrosion especially in marine applications.

In industrial furnaces, flames are separated into two main regions—a luminous region, where particles are present directly as fuel particles, or have been formed by the combustion process and, subsequently, a region where sufficient oxygen is supplied to the particles in order to complete their combustion before leaving the combustion chamber. In addition to the carbonaceous particles responsible for the main continuous radiation, some contribution from banded emission is made by large carbon and hydrocarbon molecules in regions close to the reaction zone. The mechanisms for the formation of soot involve the dehydrogenation of organic compounds and polymerization leading to formation of large carbonaceous particles. There are distinct differences in the formation mechanisms of diffusion flames and premixed flames. For very small diffusion flames on circular burners, luminous regions appear as the mass flow is increased and further increase in mass flow leads to the formation of soot at the top. For paraffins, the tendency to smoke increases with molecular weight but the reverse is true for the olefine, di-olefine, benzene and naphthalene series. Primary alcohols form more soot as the molecular weight is increased and secondary alcohols produce more soot than primary alcohols. The C/H ratio is the most important parameter, but molecular structure also plays an important role. For the same C/H ratio, branched chain paraffins produce more smoke than the corresponding normal isomers. Thus iso-octane produces flames of higher luminosity than n-octane. Flame temperature affects soot formation in two ways; higher flame temperatures provide greater temperature gradients, which favour the formation of soot particles but also the higher flame temperatures lead to faster burning. The low luminosity found in benzene and toluene flames has been explained as being due to the reduction in flame temperature as a result of soot liberation, whereas in flames with oxygen, the fuel is decomposed in the preheating zone and there is no direct contact between undecomposed fuel and oxygen because they are separated by combustion products. Sufficient residence time is also required for carbon formation. For methane, carbon is formed in regions of relatively high temperature, where the particles are consumed rapidly without formation of soot. Acetylene decomposes at a lower temperature and carbon particles form in cooler regions and are more likely to form soot because of the lower concentration of oxygen-containing substances. Turbulence leads to an increase in the formation of carbon and also to a more rapid burning of the carbon particles. Control of the mixing distribution in a turbulent flame allows control of soot formation and soot burning.

In premixed flames where there is sufficient oxygen to react with carbon to form carbon monoxide, no carbon will be formed provided the temperature is sufficiently high and the residence time is sufficiently long. When there is insufficient oxygen to react with all the carbon to form CO, then carbon will be found at the end of the flame. The formation and burning of carbon is so dependent upon local temperature and mixture ratio conditions that, unless these are clearly known or specified, it is not possible to determine the local concentrations of carbon particles.

Carbon luminosity occurs at mixtures which are much less rich than those required to liberate free carbon under equilibrium conditions. On an air/fuel ratio basis, the tendency to form carbon increases in the order: aldehydes; ketones; ethers; alcohols; acetylene; light aromatic compounds; olefins; isoparaffins; paraffins; heavier monocyclic aromatic compounds; and then naphthalene derivatives. For methane, ethane, propane and butane, carbon formation has been observed well before the rich limit, although at equilibrium, carbon should not be formed, even at the rich limit. Premixed flames have been separated into two groups—an acetylene type, in which carbon forms throughout the whole outer cone, and a benzene type, in which it forms at the tip of the inner cone. The effect of increasing pressure is to increase soot formation. It has also been demonstrated by experiment that, for diffusion flames, there is a decrease in carbon formation with decreasing pressure and that soot becomes more dense with increasing pressure, as in premixed flames.

A change of flame temperature without concurrent change in fuel i.e. via preheat, has a complex effect on soot formation. Higher temperatures can lead to suppression of soot formation, whereas, for very rich flames, increases in the amount of soot deposit have been found. Preheating the gases of a premixed flame results in a shift in the threshold for carbon formation to slightly richer mixtures; increasing the flow rate in premixed ethylene/air flames leads to an increase in flame temperature and the shifting of the threshold value to a higher carbon/oxygen ratio. Near the threshold, higher temperature decreases soot formation because it has a greater effect on the competing oxidation process than on the soot formation process. In very rich mixtures, however, the competing oxidation is less important and the high temperature accelerates the dominant pyrolysis reactions.

The addition of additives to fuel has, in some cases, produced striking changes to carbon formation in flames. The addition of 0.1% of SO3 to a town’s gas flame causes the flame to show strong oxidizing action of SO3, leading to an increase in the O/C ratio. 0.2 mole percent of SO3 added to isobutane/air diffusion flames increases carbon by 40%; whereas both SO2 and H2S tend to decrease carbon formation in premixed flames, these additives lead to decreased soot concentration in diffusion flames. Addition of CO or H2 to premixed flames results in increased carbon formation and nitrogen dilution will slightly increase carbon formation in flames of benzene or kerosene with air.

The majority of substances, when added to premixed flames in small quantities, have little effect, whereas, in diffusion flames, there is a general tendency for inactive diluents to reduce carbon formation if added in sufficient quantity. An addition of 45% CO2 can stop carbon formation in methane diffusion flames. Carbon formation has been found to be reduced in diffusion flames by addition of nitrogen and by recirculating part of the flame gases. Some metal salts, such as nickel, alkaline, earth and barium salts, reduce carbon formation and the amount of smoke emitted from engines and furnaces.

The formation of carbon particles in flames is due to the thermal decomposition of the hydrocarbons and whilst at high temperatures these decompose largely into carbon and methane in a very short time, some pyrolysis processes occur in the preheating zones of flames. Small hydrocarbons or organic molecules act as nuclei for the formation of soot which, after growth, lead to the formation of relatively large particles containing many thousands of atoms and a much higher carbon/hydrogen ratio. Both dehydrogenation and growth or condensation occur. Dehydrogenation to atomic carbon or C2 radicals, followed by condensation to solid carbon, has been found, as well as polymerization to a very large hydrocarbon molecule, which then loses hydrogen and forms graphite. Both polymerization and oxidation proceed by chain reactions, usually involving free radicals. Soot is formed from products of some of these chain reactions. Acetylene is the commonest stable intermediate, observed in rich premixed flames of other hydrocarbons. Acetylene may form carbon through the poly-acetylene route or by surface decomposition on existing carbon nuclei. Polymerization may also occur after formation of aromatics and polycyclic hydrocarbons. During the burning of oil sprays, large droplets may crack to carbon, sometimes forming cenospheres. Condensation, in the form of mist formation, and graphitization have also been observed. Small particles can also coagulate to form larger ones and electron micrographs show the linking of particles together in clusters or chains. C2 molecules, or carbon atoms, may also be important as nuclei and lead to liberation of carbon from a gas at relatively low temperature. Weinberg has shown the strong effects of electric fields in controlling soot deposition and in modifying the size of particles. Electrical effects have also been shown to play some role in nucleation, growth and coagulation in flames. Oxygen, in sufficient quantity, suppresses soot formation and carbon luminosity. In excess, oxygen may reduce carbon formation by interfering with the polymerization process, or lead to direct burning of carbon particles as they are formed.

The formation and growth processes of solid carbon in flames are not fully understood but there are at least five major routes for the process. For acetylene itself, the poly-acetylene route is important. For diffusion flames of other fuels, polymerization appears, followed either by condensation to droplets or ring-closure to aromatics, leading to graphitization. In premixed flames, competition with oxidation may prevent the build-up of large polymers and then growth of nuclei may occur.

Nitrogen Oxides

Oxides of nitrogen are formed during the combustion process, mainly as a result of chemical reactions of atmospheric oxygen and nitrogen. The oxides of nitrogen are referred to as NOx and the two major oxides of nitrogen emitted from combustion systems are nitric oxide, NO, and nitrogen dioxide, NO2. Interplay between the concentrations of these two nitrogen oxides in the presence of hydrocarbons and sunlight results in the formation of brown smog. In Los Angeles the principal source of NOx is the NO emitted from automobile exhausts. Concentrations of nitrogen dioxide are generally considerably lower than those of NO but significant concentrations of NO2 have been measured in the exhaust from gas turbines. Pollution arising from NO leads to physical discomfort, eye smarting and feelings of suffocation in locations with high smog concentration. In Tokyo, oxygen masks are issued to police on traffic control and in the United States’ cities, school classes are cancelled for children when smog levels exceed a threshold value. NO2 is itself one of the most toxic of commonly encountered gases and can be a serious health hazard.

Nitric oxide can be formed in one of the following three ways: (i) at the high temperature found in flames N2 reacts with oxygen to form thermal NO; (ii) when the fuel has nitrogen-containing compounds the nitrogen is released at comparatively low temperatures to form fuel NO; (iii) lastly NO formed in flame fronts other than that from atmospheric N2 and O2 is referred to as prompt NO. In most combustion devices, thermal NO is the dominant source of oxides of nitrogen. Crude oil and coal often contain significant amounts of organic nitrogen compounds and fuel nitrogen can be an important source of fuel NO. Under relatively low temperature conditions, approximately 1000°C, of the combustion of coal in fluidized beds, fuel NO is the dominant source. Prompt NO is formed in turbulent diffusion gaseous flames where maximum temperature levels may be as low as 1300°C. Prompt NO arises principally from reactions of fuel derived radicals with N2 which ultimately lead to NO formation when the species so formed react with O-containing species. Each individual mechanism will be discussed in turn.

The principal reactions governing the formation of thermal NO from molecular nitrogen during the combustion of lean and near stoichiometric fuel–air mixtures are given by the Zeldovich equations,

(1)

(2)

(3)

Reaction (3) becomes important only in near stoichiometric and rich flames that are held at high temperature long enough to produce significant amounts of nitric oxide. The atomic species arise from the decomposition of O2 and N2 during the chain reactions in which concentrations of atomic species can reach several times the equilibrium values, referred to as atom overshoot.

The thermal NO-formation rate is much slower than the combustion rate and in general most of the NO is formed after completion of combustion in systems where emission levels are significant. By designating a post-flame region which is downstream of the main reaction zone, the NO formation process is decoupled from the combustion process and NO-formation rates can be calculated assuming equilibrium of the combustion reactions. The calculation of NO-formation rates is, therefore, greatly simplified by introducing the equilibrium values of temperature and concentrations of O2, N2, O and OH in the NO formation rate equation.

From a survey of experimental studies Bowman shows that, in the post-combustion zone of some laboratory flames, measurements of NO are in good agreement with predictions. When, however, measurements have been made in the combustion zone, NO-formation rates are found to be significantly larger than predicted. NO-formation rates increase as the fuel/air equivalence ratio increases and the largest discrepancies between measured NO-formation rates and rates predicted have been observed in the combustion of fuel-rich hydrocarbon-air mixtures. For the majority of gaseous flames, and liquid fuel flames with negligible concentrations of fuel-bound nitrogen, combustion will take place under lean or near stoichiometric fuel–air conditions. For these flames, NO is formed principally in the post-flame region after completion of combustion. Since we are generally interested in predicting the final emissions level based upon integration of rates of formation throughout the system, predictions can be made using steady-state equilibrium approximations. These partial equilibrium models improve in accuracy as temperature or pressure increase, due to the increase in the rates of reaction leading to NO formation in the post flame region.

The simplest approximation of the kinetics of nitric oxide formation is based on C—H—O equilibrium. It is assumed that the time required to achieve equilibrium nitric oxide concentrations is much greater than the time required to attain equilibrium for the other compounds in the combustion products. The concentration of all species and the temperature can be computed for equilibrium conditions with NO excluded. The C—H—O equilibrium assumption has been used extensively in models of NO formation and is found to be generally more valid at higher pressures and in leaner mixtures.

For fuels with chemically bound nitrogen, fuel NO is formed. The nitrogen content of fossil fuels varies considerably. The nitrogen content of distillate fuels is highest in the asphaltenes fraction, 2.30 wt% and 1.40 wt% for heavy distillates. The average nitrogen content of crude oil is 0.65 wt%. The nitrogen content of most coals ranges from 1 to 2% by weight. Organic nitrogen compounds undergo thermal decomposition in the preheating zone.

The oxidation of low molecular weight nitrogen-containing compounds (NH3, HCN, CN) is rapid, occurring on a time-scale comparable to that of the combustion reaction. NO concentrations exceed calculated equilibrium values in the combustion zone and in the post-flame zone the NO concentration decreases, relatively slowly for fuel-lean mixtures and more rapidly for fuel-rich mixtures. The amount of fuel-nitrogen converted to NO is referred to as the NO yield. High NO yields are obtained for lean and stoichiometric mixtures, whereas rich mixtures provide relatively low yields. NO yields are only slightly dependent on temperature. Under cool flame conditions, as occur in fluidized bed combustion of coal, fuel NO is the dominant source and, as temperatures rise, thermal NO concentrations increase until, at high temperatures, thermal NO will generally be the dominant source. Reactions of fuel nitrogen produce reactive radical species which contain nitrogen and these species react rapidly with oxygen-containing species. Primary fuel nitrogen compounds react to form intermediate nitrogen compounds by pyrolysis or reaction with the fuel. The nitrogen-intermediates react via two competitive reactions paths—reaction with oxygen-containing species to form NO or reaction with NO to form N2. Since fuel NO formation and combustion processes occur on a similar time-scale, the reactions involving nitrogen-containing species (with the exception of N2) can be assumed to be sufficiently rapid so that the concentrations of these species are in equilibrium relative to one another from early in the reaction. NO yields are calculated by coupling the pool of partially-equilibrated nitrogen-species to the combustion process and to a kinetic scheme for formation and removal of NO. The simplest calculation procedure assumes that the combustion reactions are equilibrated at the adiabatic combustion temperature, that NO is included in the pool of partially-equilibrated nitrogen species and that decay of the nitrogen pool, and hence NO, is kinetically controlled by the single reaction N + NO → N2 + O. Some similarities can be expected between the fuel NO and prompt NO formation mechanisms on the basis of experimental observations that nitrogen intermediates found during conversion of fuel-nitrogen to NO have also been found during NO formation in fuel-rich hydrocarbon–air flames. Several investigators have shown experimentally that substantial concentrations of NO can be formed which cannot be ascribed either to thermal NO or to fuel NO. The term prompt NO was initially used because of the rapid and appreciable formation of NO in the flame. Prompt NO in its more general sense refers to NO formed in flames by mechanisms other than the Zeldovich mechanism and excluding formation of fuel NO. Bowman reviews a number of the more recent experiments in which prompt NO has been measured.

In turbulent diffusion flames, conditions favourable for the formation of prompt NO can readily arise at the interface between fuel rich and fuel lean eddies and the surrounding medium. Prompt NO is considered to be formed via CN-group containing intermediate. Several experiments have shown relatively large concentrations of HCN near the reaction zone and the rapid decay of HCN corresponds with a rapid formation of NO.

An additional cause for augmentation of NO formation is due to temperature fluctuations. In turbulent flames instantaneous temperature levels will exceed the time average temperature level and, during these periods, NO formation will be increased above the level predicted on the basis of the time average temperature. The temperature fluctuations occur due to both variations in the mixture ratio and bulk movement of the flame. These fluctuations affect the time-mean NO-formation rate in particular regions, since the Zeldovich NO formation is strongly and non-linearly temperature dependent.

Nitrogen dioxide concentrations are generally negligibly small compared to NO concentrations. Relatively large concentrations of NO2 can be formed in the combustion zone, followed by subsequent conversion of the NO2 back to NO in the post-flame region. NO2 is thus generally considered to be a transient intermediate species, which only exists at flame conditions. Rapid mixing of hot and cold regions of the flow in turbulent flames can result in a rapid quenching of the NO2, resulting in relatively large NO2 concentrations in the cooler regions of the flow, followed by subsequent emission in the exhaust gases.

It is concluded that the principal mechanism for NO formation is thermal NO via the Zeldovich mechanisms. When the fuel contains nitrogen compounds, additional NO can be formed, due to oxidation of the nitrogen containing compounds. Further quantities of prompt NO can be formed, due to radical overshoot, CN reactions or temperature fluctuations. In most practical combustion devices the temperature is the most important factor governing the formation of NO and significant reductions in NO emissions are achieved by reducing both local and overall temperature levels.

Organic Pollutants

Organic pollutants, often referred to as unburned hydrocarbons, are the consequence of incomplete combustion of hydrocarbons in the fuel. The inability to complete the combustion of the fuel affects the combustion efficiency and the primary aim of combustion engineering is to maximize the combustion efficiency and, therefore, minimize the emission of hydrocarbons. The most important reason for incomplete burning of fuel is insufficient mixing between the fuel, air and combustion products. If the air and fuel are completely mixed on the macro-scale and sufficient time is given for reaction to take place, after mixing on the micro-scale, burning could take place at stoichiometric fuel–air mixture ratios. Temperatures require to be sufficiently high for reaction to be completed during the period of contact between fuel and air and combustion products provide the heat source for raising the air and fuel mixtures above the ignition temperature. When insufficient mixing takes place, excess air—above the stoichiometric ratio—is required and the poorer the mixing the more excess air is required in order to complete combustion. Excess air reduces the combustion efficiency due to the lowering of temperatures and the increase in heat losses arising from heat being convected out of the combustion system via the increased mass flow rate of exhaust gases. The general aim of minimizing pollutants and maximizing combustion efficiency is achieved in practice by increasing the efficiency of mixing between fuel, air and combustion products, increasing the residence time, and minimizing the amount of excess air used.

Some hydrocarbon species are not considered to provide a serious health hazard, as these organic pollutants are almost entirely inert from a physiological point of view. Other hydrocarbons, such as the polynuclear organic compounds, have been shown to cause cancer when deposition is above certain threshold concentrations. The more highly reactive hydrocarbons are participants in the production of smog, whereas other organic compounds may be virtually unreactive in this respect. The relative concentrations of hydrocarbon emissions are greatly influenced by the composition of the fuel. For fuels containing large concentrations of olefins and aromatics, exhaust gases contain relatively high concentrations of reactive hydrocarbons and polynuclear organic compounds (POM). Correlations have been found between the concentration of high molecular weight hydrocarbons in the fuel and high levels of POM emission. Pyrolysis synthesis and hydrocarbon reactions in the flame result in the formation and emission of hydrocarbon compounds which are products of combustion. The complete understanding of the history of organic compounds in the combustion process requires detailed knowledge of individual pyrolysis and oxidation reactions of individual hydrocarbon species. Hydrocarbons decompose thermally via chain reactions, which are initiated by uni-molecular decomposition of the parent hydrocarbons. Synthesis reactions occur during combustion, leading to the formation of complex hydrocarbon molecules in the post-combustion region of flames with simple hydrocarbon fuels.

Oxidation mechanisms of hydrocarbon species are divided into two principal regions: low temperature (T 〈 1000 K) and high temperature regions (T 〉 1000 K). Most of the hydrocarbon species found in exhaust gases are formed in the low temperature regions. The high temperature oxidation mechanisms differ from those at low temperature because of the decreased importance of several intermediate species, such as peroxides, hydroperoxides, aldehydes and peracids.

Before oxides of nitrogen were considered to be a major pollutant, the emission of hydrocarbons from many combustion systems was reduced to very low levels by the use of efficient mixing and high combustion efficiency was achieved with low excess air. In these systems, temperature levels were high—favouring the formation of oxides of nitrogen. The focus of attention on oxides of nitrogen as a major pollutant led, initially, to the use of off-stoichiometric mixture ratios, which resulted in the lowering of temperatures and, consequently, decrease in emission of NOx, but with reductions in combustion efficiency. Satisfying the legislative requirement of maintaining both emissions of hydrocarbons and NOx below statutory limits has only been achieved by careful control of mixing patterns, temperature levels and residence time distributions throughout the system.

Carbon Monoxide

Carbon monoxide was one of the first combustion products to be recognized as a pollutant. Because of its hazard to health and its fatal effect under uncontrolled conditions, careful monitoring and control of CO emissions have been introduced in most industrial plants, but the main source contributing to ground-level concentrations of CO in cities is the petrol engine, burning a rich mixture. Dangerous levels of CO are also found in buildings where combustion chambers and exhaust ducts are not securely sealed. Carbon monoxide is formed as an intermediate species in the oxidation of carbon-containing fuels. The reaction of CO to CO2 is almost exclusively due to the elementary reaction

(4)

Since this is the sole major mechanism for the conversion of CO to CO2, it has been concluded that all carbon initially present in the fuel will form CO. All efforts in controlling emissions of CO are, therefore, concentrated on the completion of oxidation of CO rather than attempting to inhibit its formation. If sufficient oxygen and residence time is available at flame temperatures CO concentrations fall to very low levels after reaction, to form CO2. Maximum CO concentrations in flames are generally larger than the equilibrium values for adiabatic combustion of the reactant mixture. The levels of CO detected in exhaust gases are lower than the maximum values found within the flame but are significantly larger than equilibrium values for the exhaust conditions. Both the formation and destruction of CO in combustors are kinetically controlled. CO formation is one of the principal reaction paths in the hydrocarbon combustion mechanism. The principal CO formation reaction is due to thermal decomposition of the RCO radical. Bowman in his examination of CO kinetics, shows that it is possible to use a quasi-global model for CO formation by a one-step reaction in which the hydrocarbon fuel reacts with molecular oxygen to form CO and H2. The rate of oxidation of CO to CO2 is relatively slow compared to the CO formation rate. In hydrocarbon flames, which generally have relatively large H concentrations, the oxidation of CO is very slow and may, in many cases, be neglected. Whilst the various elementary steps for the oxidation of CO are known, the steps by which CO is formed are not known, except for simple fuels such as methane.

Sulphur Oxides

Sulphur dioxide, SO2, and sulphur trioxide, SO3, are formed from the sulphur-containing compounds in the fuel after reaction with oxygen. The deposition of sulphur oxides sulphates and sulphuric acid on metallic surfaces causes severe corrosion problems, both inside and outside combustion equipment. Corrosion is usually associated with conversion of SO2 to SO3, which then hydrates and condenses as sulphuric acid on cooled surfaces. SO3 concentration in flames can greatly exceed thermodynamic equilibrium concentrations, based on the molecular reaction between SO2 and O2. Rapid cooling of combustion gases can result in the freezing of SO3. When residence times are short, as occurs in some boilers, only 5% of SO2 is oxidized to SO3. In automobile engines, over 40% conversion to SO3 is possible. It was thought at one time that the oxides of sulphur problems could be reduced by not burning fuels with high sulphur contents. At times of fuel shortage it becomes necessary to burn whatever fuel is available and, under these conditions, the fuel needs to be desulphurized prior to introduction into the combustion chamber or special provision has to be made for sulphur removal prior to emission from the combustion chamber.

ANALYTICAL AND MODELLING TECHNIQUES IN TURBULENT COMBUSTION

Boundary Layer Equation Methods

Very significant advances have been made in recent years in the field of mathematical modelling as used particularly in the fields of fluid mechanics, heat transfer and chemical kinetics. The principal equations for conservation of mass, momentum, energy and individual species have been formulated in terms of the governing differential equations. Because of the three-dimensional and turbulent nature of most combustion systems, it is not possible to solve the fundamental equations even with the use of the largest available computers. Assumptions must be introduced to allow the simplification of the equations. Formulation of the chemical kinetic models involves the identification of the critical rate-limiting steps. Depending upon the particular situation, many processes can be ignored on the basis that they are too slow compared to the faster chemical reactions.

The fundamental equations including those of continuity, Navier–Stokes and the heat and mass transport equations are analysed numerically using finite-difference techniques and computers with large memories. Appropriate forms of the finite-difference equations are selected and the choice of grid size for the finite-difference computations is limited by the memory size and execution time of the computer. For examination of some combustion and pollution problems, where formation and destruction processes occur on the microscale, it becomes necessary to introduce a very fine mesh size.

Turbulence plays an important role in the fluid flow in almost all combustion chambers. Since the time dependent Reynolds’ equations are too complex to solve directly, the approximations of the correlation terms relating fluctuating components of velocity, temperature and concentration are introduced. Caretto examines the range of models which have been introduced for solution of turbulent fluid flow problems. These include models based on the kinetic energy of turbulence, turbulent viscosity and length scale parameters. The special problems associated with viscous sub-layers in wall regions are studied by using various techniques for determining wall fluxes. Despite the complexity found when dealing with non-reacting turbulent flows, major developments are taking place in the formulation of models for turbulent flows including chemical reaction. Account is taken of the effects of turbulence on combustion and the effects of chemical heat release on fluid flow. Heat transfer by radiation plays an important role in many high temperature combustion systems and flux methods have been developed for modelling the radiation flux.

Caretto derives the finite-difference forms of the basic equations and he discusses the problems associated with non-linearity of equations, influence of pressure fields and three-dimensional recirculating flows. He concludes that the potential for applying finite-difference equations to pollutant modelling is great, but that at present no good comparison is available between theory and experiment. Further, efforts should be concentrated on improving the quantitative knowledge of the mechanics of turbulence and its interaction with chemical kinetics since these processes are more important than making further advances in computing techniques and numerical simulation.

Bilger in his review of turbulent jet diffusion flames develops a basic theory for reaction rates and diagnoses the role of kinetics in the detailed flame structure. He concentrates on an examination of the burning of gaseous fuels as turbulent jet diffusion flames and examines the various models that have been put forward attempting to deal with the problems of unmixedness and the effects of combustion on the fluid mechanics of the system. Particular emphasis is given to the effects of fluctuating and spatially varying density on the computation of velocity and scalar fields. On the basis of recent measurements of concentration fluctuations in non-reacting flows, it is shown that the probability density function is not Gaussian in form and that this therefore imposes theoretical constraints on many previously accepted models. Theories of the nitric oxide formation in turbulent diffusion flames require examination in the light of mounting experimental evidence that O-atoms are being found in super-equilibrium concentrations and thus account needs to be taken of O-atom overshoot on nitric oxide formation in flames. In Bilger’s development of the basic theory of flame structure he shows the relationships between the conservation equations and the Flame Sheet Model, relations between conserved scalars, and the shifting equilibrium reaction model. On the basis of mixing experiments he develops a mixing analogy and shows the effect of the reaction rates on reaction zone structure.

In the theoretical analysis of turbulent combustion systems it has been the practice to eliminate terms in the equations on the assumption that their order of magnitude was negligible compared with other terms in the equations. More recent analyses are showing that some of these assumptions are not valid and particular care needs to be taken in the examination of the influence of terms involving correlation between fluctuating components of temperature, concentration, density and velocity. Bilger places particular stress on the use of Favre averaging in which all quantities are weighted by the instantaneous density before averaging. In the analysis of the boundary layer equations, Favre averages, fluctuations, cross-correlations and probability density functions are used. The density fluctuations are shown to have a direct effect on turbulent fluxes and it is shown that density fluctuations can generate or suppress turbulence and also alter the structure of the turbulent flow field.

The possibility of the existence of flame generated turbulence has been recognized for many years. For flames in tubes the evidence and explanation of flame generated turbulence is relatively well established. In unconfined or partially confined turbulent diffusion flames, the evidence is less clear and direct evidence of the generation of turbulence as a consequence of chemical reaction in flames is not yet available. It can be concluded that in confined flows where strong pressure gradients rise, flames can have large effects on turbulence. In free turbulent flows there is evidence of increase in magnitudes of fluctuating components of velocity, as measured by laser anemometers and that intermittency is different from that in uniform density flows.

The significance of probability density functions in combustion flows is becoming recognized, but there is as yet no evidence in the literature of measurements under combustion flow conditions. There is, however, some information obtained in non-reacting flows and these are being examined as to their possible effects on analysis and prediction in turbulent flames. The basic properties of probability density functions, their modelling and the use of approximate methods are discussed in detail by Bilger. Recommendations are made as to the type of measurements which are required in order to provide further insight into the structure of diffusion flames and the role that non-equilibrium kinetic effects play in these flames.

Population Balance Approaches and Mixing Concepts

In the field of chemical engineering and chemical reactors a number of concepts have been formulated relating the mixing of reactants with chemical reaction. It is only relatively recently that these concepts have been introduced into the field of combustion and Pratt summarizes the present state-of-the-art by dealing with the effects of spatial fluctuations and intermittency of temperature and specie concentration as they effect the chemical-kinetic coupling on energetic and pollutant formation processes. In chemical reactor theory, population balance concepts and stochastic mixing models are used for describing the interaction of chemical reaction and turbulent fluctuations. During the process of mixing fuel and air, Pratt distinguishes between stream mixing when the fuel and air are introduced in separate streams and mixing takes place across the stream boundaries, and age mixing or back mixing during which partially or wholly burned gases are mixed with the unburned fuel-air mixture in order to produce flame stabilization without the aid of an external energy source. The degree of mixing between fuel and air streams determines the local fuel-air ratio and this in turn governs the local rate of chemical reaction and heat release. Mixing between streams of partially mixed and streams of burnt gases affect the temperature distributions as well as local fuel-air ratios. In order to obtain flame stabilization and high volumetric heat release rates, many combustion chambers have recirculating flows which provide the required degree of age mixing. Pratt describes how the effectiveness of stream mixing is studied by injecting inert tracers and measuring tracer concentrations as a function of time at various points within the system and at the exit from the system. In terms of statistical quantities based on concentration measurements, the degree of stream segregation and unmixedness is defined. The probability density functions of inert tracer concentrations for a number of systems is illustrated by Pratt. When a pulse of inert tracer is injected, the age mixing effectiveness can be determined by measuring the time-varying tracer concentration at a point within the system. By this means distribution functions for molecular ages at a given point and the degree of age segregation are defined. At the exit plane from a combustor the residence time distribution function is shown to be related to the age distribution function.

In chemical reactor theory, the probability balance is primarily concerned with the interaction of turbulence mixing and chemical reaction. Macro mixing effects are treated solely through consideration of the residence time distribution, and frequency function. Distinction is made between the a priori probability density with which a particle that enters the reactor at time t = 0 will leave the reactor at time t and the a posteriori probability density of the fraction of fluid particles which entered at time t = 0 and which exist during the time interval (t, t + dt). Reactors are separated into two principal types—plug flow reactors in which all particles entering the reactor together, leave together, and in which no age mixing or axial dispersion is permitted, and perfectly-stirred reactors for which the exit stream has the same concentration as that of the homogeneous reactor contents.

In combustion systems it is necessary to initially achieve macro mixing, but combustion can only be completed when mixing takes place at the molecular level. Distinction is thus made between macro mixing referring to mixing between streams, and back mixing by convective recirculation, and micro mixing which occurs at the molecular level and is dependent upon molecular diffusion rates. Pratt gives various examples demonstrating the importance of the various types of mixing as they occur in typical combustors. In order to apply the mixing concepts to practical combustion systems, a stochastic model is introduced for combustion with stream mixing. Turbules of fluid are allowed to interact with turbules of any age or residual lifetime and turbules of any age are allowed to exit from the reactor. This model corresponds to a perfectly stirred reactor with global age segregation varying between zero (micro mixed) and unity (macro mixed) as the coalescence/dispersion frequency parameter varies between infinity and zero. Pratt argues that alternatives need to be found to the traditional, fluid mechanical approach based on computational models for higher-order correlation terms which appear upon Reynolds’–decomposing of the governing differential equations. He shows that the experiments obtained to date with stochastic Monte Carlo, coalescence/dispersion models for simple combustion flows are sufficiently encouraging to expect considerable progress to be made by using chemical reactor concepts.

THE INTRINSIC TURBULENT STRUCTURE

Turbulence plays an important role in mixing processes in almost all practical combustors. In the past the statistical approach has been adopted for analysis of turbulent flows, based upon the assumption that turbulence is a stochastic or random process. Almost all of the statistical methods involve averaging, either with respect to time or space. It is common to invoke the Reynolds assumption by describing the velocity, temperature, pressure and concentration at any one instant of time, as the sum of a time-average and a fluctuating component. Turbulent flows are then treated, both analytically and experimentally, by introducing an effective or turbulent eddy viscosity and treating the flow as being similar to that of a laminar flow. In recent years the basic concepts of turbulent flow have been re-examined and new ideas and approaches are in the process of development which are of particular significance to the problem of the influence of turbulence on combustion as well as the influence of combustion on turbulence.

There is now sufficient experimental evidence that has demonstrated the existence of structures in turbulent shear flows and a detailed examination of this physical structure leads to the conclusion that coherent structures are present in all turbulent flows. These coherent structures are large in scale, can be clearly identified as they form at wall or fluid interfaces and retain some element of coherence for long periods of time so that they can be recognized far downstream from their point of formation. In jet, wake and boundary layer flows, repetitive and orderly eddy structures have been clearly identified and their growth and interaction with the surrounding fluid is being studied by the use of sophisticated experimental and data processing techniques. By coupling flow visualization experiments using high-speed photography, with trigger and multiple high frequency response probes, coherent recurring structures have been shown to occur in mixing layers as they occur in wakes, boundary layers and jets. Further, it has been shown that the classical statistical measuring techniques only provide quantitative information of time-average properties and do not give information on the detailed structure. By the use of conditional data sampling techniques which separate the more significant components from the background noise, clearly defined patterns have been shown to emerge and these are related to clearly identifiable representative physical characteristics in the flow pattern.

From the above discussion it is apparent that effort needs to be concentrated on obtaining more information of the instantaneous conditions of combustion flows rather than relying on the more classical time-mean and exchange coefficient analytical approaches, if progress is to be made in predicting pollutant emissions. The use of coalescence and dispersion models, intermittency techniques and higher order, Favre averaged cross-correlations are steps forward in this direction. Recent experiments in cold jets, together with the intermittency and unmixedness data for heated heterogeneous and burning flows suggest that turbulent diffusion flames are dominated by large coherent eddies mixing with the flow. These eddies entrain both fuel and air and their motions can produce regions of unburned fuel bounded by air; fuel bounded by air mixed with hot products and also air/product mixtures either within the eddies or at their boundaries. These eddies have been referred to as Coherent Structures, as they persist for relatively long periods before breaking up. The microscale mixing required for combustion is associated with the small scales of motion within and at the boundaries of these eddies. As the velocity fields of the large eddies have been seen to be fairly repetitive, it is reasonable to propose that the micro-scale mixing and hence flame structures associated with the eddies are basically repetitive from one eddy to the next. There thus seems some hope of developing better physical models of details of the flow in terms of these eddies and their interactions as they travel through the flow. Furthermore, the existence of such eddies suggest that the coalesence and dispersion modelling described by Pratt may in fact have a physical basis in terms of real eddies in the flow. It should be possible to modify Pratt’s approach to model the structure and behaviour of real large eddies more closely. For example, in simple jets, the large eddies are found to grow with increasing residence time by the coalescing of adjacent eddies, without subsequent break-up or dispersion. These large eddies produce the intermittent forms for time histories of concentration, velocity and temperature, as described by Bilger, and a more detailed knowledge of these eddies will allow more accurate analysis of intermittency from equations for combustion conditions, and in particular of the cross-correlation terms. Knowledge of the structure and motion of the larger eddies could lead eventually to a better prediction of temperature and concentration fluctuations, and correlations within the flow as well as the overall flow patterns. This in turn would lead to a better prediction of pollutant emissions.

PRACTICAL COMBUSTION SYSTEMS

The application of fundamental principles to the practical problems of minimizing pollutant emission is examined by Mellor for gas turbine engines, Heywood for spark ignition engines and Henein for diesel engines. Each author provides details of the combustion chambers in each type of engine, and discusses the particular design and combustion problems that arise in practice. Design modifications are suggested in order to meet the legislation introduced in the United States by the Environmental Protection Agency.

Gas Turbine Engines

Emission of pollutants from aircraft is mainly of concern in urban areas with large airports and predictions of the contribution of aircraft to pollution in the metropolitan Los Angeles area for 1980 are carbon monoxide, 13.6%, unburned hydrocarbons, 2.5% and oxides of nitrogen, 5.7%. Regulations controlling the emission of pollutants from aircraft are being mainly confined to the conditions of taxi/idle, take-off, climb-out and approach during which the aircraft is below 0.9 km altitude. Examination is being made of the influence of emissions of oxides of nitrogen on ozone concentrations in the stratosphere for supersonic transport, but long term effects of such pollution have not been conclusively proved. The principal pollutants from aircraft gas turbine engines are unburned fuel and other hydrocarbons (HC), CO, NOx and smoke. Sulphur oxides are not usually taken into account since aviation fuels contain very small concentrations of sulphur.

Combustion efficiencies of gas turbines are very high and in almost all cases they are above 98%. Also the differences between aircraft combustor designs of the various manufacturers are not very great, though designs have been proposed involving radical changes to the presently accepted form in which the chamber is divided up into primary and secondary combustion zones followed by a dilution zone. Most of the emitted pollutants are formed in the primary zone and are generally ascribed to poor mixing. The destruction of pollutants is incomplete due to insufficient residence time. Since liquid fuel is introduced in the form of a spray, vaporization and burning of liquid droplets needs to be taken into account. Mellor shows the relative importance of the time scales affecting droplet evaporation, injected liquid fuel, eddy dissipation in shear layers, fuel ignition delay and NO formation. He shows how these affect the special problem of achieving stable combustion when the overall fuel-air equivalence ratios are of the order of 0.2 as well as the problem of blow-off–relight hysteresis that is found in gas turbine combustors. Mixing between hot recirculating burnt gases and air in the recirculation zone as well as mixing between air and fuel play a dominant role in combustion performance. It is shown how pollution can be reduced by variation in the characteristic times. These are achieved practically by improving fuel atomization, increasing air velocities injected with the fuel, increasing air flow rates through the primary zone and changing local temperature distributions so as to increase chemical reaction rates for HC and CO as well as decreasing chemical reaction rates for production of