|Table of Contents|
Table of Contents
Chapter 1. Introduction
Chapter 2. Soot formation
Chapter 3. Experimental methods
Chapter 4. Evaluation of the optical system
Chapter 5. Soot morphology and chemical analysis of PAHs adsorbed on the soot particle surface
Chapter 6. Experimental results and discussions
Chapter 7. Conclusions and recommendations
CHARACTERISTICS OF SOOT FORMATION AND BURNOUT IN TURBULENT RECIRCULATING FLAMES
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1987
DEDICATED TO MY PARENTS
The author would like to thank Dr. Charles L. Proctor II, chairman of the supervisory committee, for his valuable assistance in conducting this project. Special thanks are extended to Dr. Alex Green, Dr. Roger Gater, Dr. John Eyler, and Dr William Lear for serving as committee
This research effort was conducted under a contract from the US Air Force, Tyndall Air Force Base, Environics Division, Panama City, Florida. The author would like to
thank the personnel of the Environics Division for their support and friendship throughout the 2-1/2 years that he spent working with them. Special thanks are extended to Major Paul Kerch, Dr. Joe Wander, and MSgt. Dan Stork.
The author would like to acknowledge Dr. William Shultz and Ms. Mary McGill for chemical analyses of the soot samples, and Dr. Richard Irey and Dr. Art Sterling for valuable technical discussions.
Special thanks are extended to Mrs. Becky Hoover and her daughters Pam and Kim for their friendship and help in preparing this manuscript.
TABLE OF CONTENTS
ACKNOWLEDGMENT .................................... iii
ABSTRACT .......................................... vi
I INTRODUCTION ............................... 1
II SOOT FORMATION ............................. 7
Introduction .......................... 7
Soot Precursors ....................... 8
Neutral Species Mechanisms ............ 8
Ionic Mechanisms ...................... 14
Soot Emission Models .................. 23
Experimental observations ............. 30
III EXPERIMENTAL METHODS ....................... 39
Modular Burner ........................ 40
Physical Sampling ..................... 43
Temperature Probe ... 47
Integrated Particle sizing System 47
IV EVALUATION OF THE OPTICAL SYSTEM ........... 60
Cold Flow ............................. 62
Hot Flow .............................. 68
Reproducibility of Soot Data .......... 75
V SOOT MORPHOLOGY AND CHEMICAL ANALYSIS OF PAHs
ADSORBED ON THE SOOT PARTICLE SURFACE ....... 77
Morphology ............................ 77
Chemical Analysis of PAHs Collected on
the Soot Particle Surface ............. 79
VI EXPERIMENTAL RESULTS AND DISCUSSIONS ....... 88
Experimental Conditions ............... 89
Velocity Measurements ................. 90
Temperature Measurements .............. 97
Soot Measurements ..................... 112
VII CONCLUSIONS AND RECOMMENDATIONS ............ 152
Conclusions ............................ 152
Recommendations ......................... 154
REFERENCES ................................... o ... 155
BIOGRAPHICAL SKETCH ............................... 164
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of philosophy
CHARACTERISTICS OF SOOT FORMATION AND BURNOUT IN TURBULENT RECIRCULATING FLAMES
Chairman: Charles L. Proctor, II Major Department: Mechanical Engineering
The present study represents an investigation of the effect of fuel type, fuel stream heat content, nitrogen dilution, and air jet velocity on soot formation rates and
particle burnout in a highly recirculating, turbulent-type flame. Soot particle size and flux measurements have been made using an optical probe based on a large angle ratioing
technique to measure the intensity of forward scattered light from individual particles at two off-axis angles.
Chemical analyses of soot samples have been made using a gas chromatograph with a flame ionization detector (FID), and a morphological analysis of soot samples has been made using a scanning electron microscope (SEM). Physical
probes have been used for temperature measurements and extraction of soot particles.
Chemical analysis of the composition of the polycyclic
aromatic hydrocarbons (PAHs) extracted from soot samples collected at the face of the burner and on a filter located
downstream in the exhaust system suggests that multiple, convergent pathways, rather than one chemical mechanism, lead to the formation of high molecular weight PAHs and soot.
Net soot production was found to be the result of the competition of soot particle formation and burnout. The fuel type and the fuel stream heat content appear the main
parameters that determine the flame's propensity to soot. The addition of nitrogen to a fuel stream increases the difference in the net soot production among the fuels investigated. Dilution by nitrogen decreases more effectively the oxidation rate of soot particles in flames that use fuels of lower heat content.
Unburned hydrocarbon particles are detrimental both to human health and to the efficiency of fuel-powered engines. Both of these problems must be acknowledged by any program to develop technologies that would use fuels having a broader range of properties, which eventually would lead to
increased soot emissions. The goal of modern engine designers is to devise combustors giving optimum performance while satisfying environmental quality requirements. Therefore, an understanding of soot
formation and destruction in combustion chambers is a prerequisite for achievement of this goal.
Soot emissions involve several interactive processes that occur simultaneously in a combustion environment. Such processes include chemical and physical aspects of soot formation and growth, fluid mechanics involved in the
transport of mass, momentum, and energy between different regions of the flames, and radiative heat transfer from the flame to the surroundings.
Researchers in this field continue to attempt to devise laboratory scale experiments that can decouple the different processes occurring in soot formation. For
example, shock tubes and simple flames have been used to
examine the chemistry of formation of soot particles and several theories based on neutral or/and ionic mechanisms have been proposed to describe the chemical processes associated with soot formation. These theories were based on experimental observations on the fate of gaseous products and particles in flames. To date, no theory has been universally accepted because of the complexity of the subject. These complexities result from the very fast
rates of chemical reactions leading to soot formation and from the dependence of the relative rates of competing pathways leading to the formation of soot under varying experimental conditions.
Notwithstanding the importance of chemistry in
understanding the incipience of soot particles, chemistry, alone, does not exclusively explain the formation of soot in flames. Such physical processes as particle growth, which accounts for most of the soot loading (Prado and
Lahaye, 1981), and fluid mechanics are no less important than the chemical processes in understanding soot formation. Modeling all the processes that are involved in the formation and evolution of soot particles in flames
requires an extensive database, which is not available at this time. Current limitations in state-of-the-art analytical techniques and in computer storage preclude detailed modeling of all the steps of soot formation.
Because of the complexities involved in developing a detailed phenomenological model of soot formation, only few
predictive models for the rate of soot formation and consumption have been proposed. Most of the proposed models were based either on correlative equations
dependent on experimental conditions but not on chemical kinetics of the system or on global or quasi-global kinetic concepts that use unrealistically simplified
chemical kinetic equations. At the present time, the simplistic approach of these models generally affords moresatisfactory descriptions than do detailed approaches to the prediction of soot formation in a well defined combustor configuration; however, it must be recognized that development of models has been constrained by the paucity of detailed and accurate measurements of the
fundamental chemical and physical parameters that control soot formation. Determining and elucidating the effects of these parameters on soot formation will contribute to a database for developing more efficient predictive models.
Experimental diagnostics on gas turbine engines can be very difficult and costly; therefore, gas turbine engines were not an attractive alternative to the simpler
fundamental laboratory burners. A bluff body burner was used in this study to provide both reasonable simulation of the combustion process in gas turbine engines and simplicity of operation. Burners of this type, which were designed to provide an intermediate combustor for practical
studies (Roquemore et al., 1983), produce axisymmetrical turbulent flames that simulate a wide range of practical combustors.
Evaluation and development of models of soot formation and destruction are possible only after a) accurate
characterization of species and particles as they exist in a realistic flame and b) determination of the fundamental parameters that control the different processes involved in
the formation and destruction of soot. Accordingly, this investigation was designed to meet two main objectives, which can be described as follows: first, to evaluate the
performance of an optical diagnostic system for both particle size and count rate measurements, with emphasis on
characterization of soot particulates in combustion and to perform the necessary adjustments for optimization of data
acquisition; second, to identify and assess the main parameters that control soot formation in well defined experiments.
The optical diagnostic systems used in this study
consist of an Integrated Particle Sizing System (IPSS) for particle size and population density measurements. A
thorough analysis of the IPSS was undertaken to evaluate its performance as a particle sizer and particle counting instrument in both reacting and non-reacting flows.
Polystyrene latex spheres of known size were used to calibrate and verify the performance of the instrument in a
non-reacting medium. The system-s response for size and particle count rate measurements in non-reacting and reacting flows assessed the viability of the instrument.
The aim of this work was to improve understanding of processes of soot formation and destruction by using stateof-the-art techniques to measure soot particle diameter and number density in conjunction with temperature and velocity profiles in a highly turbulent recirculating flame. Any change in the operating conditions of the combustors would strongly affect the formation and destruction rates of soot
particles as well as the temperature and velocity profiles in the flames. The thrust of this work was to devise well
conceived experiments to decouple the different processes that occur in combustion flames and determine the main parameters that control the soot formation and subsequently the oxidation rates of soot particles.
The formation and evolution of the soot particles in flames were argued to be strongly affected by the temperature history, residence time, and fuel structure (Santoro and Semeriian, 1985). Accordingly, detailed
measurements of soot particle size and number density, temperature, and velocity were used to investigate the
different processes involved in soot formation and burnout in a turbulent recirculating flame. Given the importance of mixing in turbulent recirculating flames, flowfield mapping was performed to localize the regions of greatest
soot particle loading and to trace the mean path of soot particles through the flame.
Temperature, velocity, and species concentrations are interdependent state parameters that vary with the operating conditions. Their separate effects on soot formation can be examined by using a selective set of independent controlling parameters. The controlling parameters that were investigated in this study are fuel type, fuel flow rate, fuel heat content, and fuel and air jet velocities. Nitrogen dilution was used to control the fuel jet velocity and the fuel jet heat content.
A general overview on the work that has been
accomplished in the area of soot formation will be given in chapter II. The experimental equipments used in this work will be described in chapter III, followed in chapter IV by a thorough analysis of the Integrated Particle Sizing System to evaluate its overall performance as a particle sizer as well as a particle counter. Soot morphology and chemical analysis of soot samples obtained from different flames will be covered in chapter V. The experimental
results and discussion of this research effort will be presented in chapter VI. Finally, in chapter VII the main conclusions will be drawn and recommendations for future areas of research will be offered.
A GENERAL REVIEW OF SOOT FORMATION Introduction
The soot formation process which occurs during
combustion of hydrocarbon is yet to be understood by the community working in this field of research. Although
several theories of soot formation have evolved over many years of research, none of them have been universally accepted because of the complexity of the subject. These theories historically were derived from experimental observations, and advances in determining their validity have only recently become possible through the development of new experimental techniques. These new techniques,
which permit the study of molecular species and particles as they exist in flames, include mass spectrometry (Calcote, 1963; Michaud et al., 1981), laser light scattering (Kent et al., 1981; D'Alessio et al., 1977), laser induced fluorescence (DiLorenzo et al., 1981), laserinduced ionization (Smyth and Mallard, 1981) and other intrusive or non-intrusive techniques.
This chapter summarizes the work that has been
accomplished in this field and -justifies the experimental approach of this research program; it includes the proposed mechanisms leading to the formation of the first solid soot particles, recent experimental developments in the study of
soot formation in shock tubes and flames, and a review of some analytical/correlative models for soot emissions.
The primary soot particles formed in flames consist of spherical or near spherical elementary particles which have mean diameters in the range of 10-50 nanometers, and a carbon-hydrogen atom ratio in the range 8:1-12:1 (Calcote, 1981).
Soot forms in hydrocarbon flames by a succession of fast processes that occur within few milliseconds. In the first process, nucleation embryonic gaseous species (soot precursors) undergo a series of gas phase reactions to form
the first solid particles. In the second, surface growth, small solid particles coalesce to form large spherules. In the final process, the spherules aggregate to form chains.
Neutral Species Mechanisms
As attempts were made to develop comprehensive
phenomenological models for soot formation, the combustion chemistry of neutral species in flames has given birth to new theories. The development of new experimental techniques for sampling species detected in flames has been very helpful in understanding soot phenomena. The new findings, however, have discredited many theories that had
attempted to explain the principal routes leading to soot formation. Workers in the field of carbon formation in flames have failed to agree on a neutral species mechanism
that describes the phenomenological process of soot formation. No attempt will be made to discuss all the various theories because they have been reviewed by several authors such as Palmer and Cullis (1965), Gaydon and Wolfhard (1970) and others; however, the most recent viable neutral-species mechanisms, which have received support, will be discussed.
Homann and Wagner (1967, 1968) suggested that intermediate hydrocarbons have an acetylenic character that enables them to be added to other hydrocarbons and still retain their radical character. They argued that the rate of polyacetylene formation can be explained only by radical reactions and that the polyacetylenes are subsequently rearranged to an aromatic graphite-like structure and soot. This scheme was proposed based on their observations of the fate of the polyacetylenes and polycyclic aromatics in the oxidation region of the flames they studied. The concentration of polyacetylenes was observed to decrease while the concentration of the polycyclic aromatics was increasing in the oxidation region. From considerations of the results of other workers (Cullis et al., 1967), Bonne et al., 1965; Thomas, 1962), Calcote (1981) concluded that the cyclization process in the polyacetylene scheme would be too slow to account for the rapid formation of soot particles in the post oxidation zone. Further, the unfavorable thermochemistry of large CnH and CnH2 species might explain the rapid decrease in their concentrations.
Homann and Wagner suggest the possibility of C2H or other radicals attacking a polyacetylene molecule to give a branching radical with ring closure. The latter approach was been reinforced by the work of Crittenden and Long (1973) who concluded that an aromatic ring with a twocarbon side chain (i.e., C6-C2) can be formed from polyacetylene branch radicals by ring-closure or by cyclization. The latter species, still having radical character, will grow to small carbon particles as further polyacetylenes are added.
Glassman (1979) considered a model where both free radicals and ions have a role in the formation of incipient soot particles. He emphasized the importance of strongly conjugated molecules having resonance structures on the stability and reactivity requirements of soot formation. As pointed out by the author, such strongly conjugated molecules as butadiene and benzene undergo rapid reactions
with each other, with ions, and with other neutral molecules and radicals. They are also stabilized by their polar resonance structure. Although butadiene has less resonance energy than benzene, its ability to accept a positive ion and form an even more reactive carbonium ion can make it the most likely precursor to nucleation. This can be seen in the following example:
H2C=CH-CH=CH2 + H+ CH3-C + H-CH=CH2 H3-CH=C + H=CH2
According to Glassman's model butadiene forms from
acetylene in premixed flames, and probably reacts further to form larger precursor radicals, which undergo further addition and dehydration via a combination of reactions involving radical and ionic compounds. The products of these reactions, under the conceptual scheme developed by Glassman, undergo Diels--Alder cyclization leading to polynuclear aromatics. Another feature of butadiene, as noted by Glassman, is that of all the fuels tested, it has the greatest tendency to soot in a diffusion flame. The principal objection to the model is that the concentration of butadiene falls very rapidly even before soot particles are observed (Calcote, 1981).
Bittner and Howard (1981 a&b) reported mole-fraction profiles of stable and free radical species in a benzeneoxygen-argon flame near the sooting limit (equivalence ratio = 1.8). They concluded that PAH observed in sooting flames cannot be formed by condensation reactions of intact aromatic rings, and that much of the carbon that ends up as
soot may come from non-aromatic hydrocarbon intermediates. Their flux profiles for C6H6 and C2H2 indicate that the reaction of OH, rather than 0, with benzene may be the
primary source of C6H60, followed by production of CO and C 5H 6; however, acetylene is not an early product in the ring fragmentation.
A sequential mechanism is outlined by these workers for the production of C1, C2, C3, C4 ,and C5 from benzene that is consistent with the flux profiles of intermediate species. They observed that as the fuel equivalence ratio is increased past the sooting limit, a sequential growth from polycyclic aromatic hydrocarbons of 120-210 daltons to heavier species occurs, and then soot is formed. They point out that rapid growth of aromatic structures by free-radical addition may result in a stabilization of the intermediate without formation of six-membered rings.
Among the possible free-radical addition schemes proposed for PAH and soot formation, reaction of the phenyl radical
(C6H5) with vinylacetylene has been found the most favorable for stabilizing the intermediate to form a six membered ring.
Stein (1978) suggested that concentration profiles of
the most reactive species cannot, in themselves, explain the chemistry of polycyclic aromatic hydrocarbon (PAH) and
soot formation. He conducted studies on the equilibrium behavior of PAHs in hydrocarbon systems at high
temperatures to elucidate the chemical steps leading to their growth and finally to formation of soot particles. one important finding of his studies is that above 1700K, the concentration of highly condensed PAHs passes through a minimum. He concluded that since only highly condensed aromatic hydrocarbons (i.e., benzene, coronene,) are capable of extensive growth, this minimum may constitute an
effective thermodynamic barrier to formation of soot
particles at high temperatures. A polymerization scheme is given below by the author for the first six members of the most thermodynamically favorable high-temperature polymerization route. In this scheme C2 and C4H2 refer only to the number and type of atoms added in each step and do not refer to specific chemical species or mechanisms.
C 4H 2C 4H 2C2
cihk + c-4 9 *
C 4H 2C 2C2
Another pathway for soot formation is given below for soot and PAH formation through direct aromatic condensation
in which aromatic units remain intact. The constituent oTerphenyl is at minimum equilibration concentration relative to both larger and smaller molecular
intermediates in this polymerization pathway. Stein
concluded that soot formation must occur through PAHs that represent the most favored thermodynamic path.
As a summary of the above discussion concerning
neutral mechanisms leading to soot formation, one can conclude that each of these hypothetical mechanisms has
been criticized in one or more of its intermediate steps. This demonstrates the lack of rigor in their concepts,
primarily owing to the complexity of the subject.
It has long been recognized that flames conduct electricity and therefore contain ions. The lack of a
unifying mechanism involving neutral and free-radical nucleation processes and the weaknesses encountered in the mechanisms proposed above have encouraged the consideration of an ionic mechanism as a possible process of soot nucleation in flames.
There are now improved mass spectrometric techniques that permit measurements of detailed ionic profiles in both fuel-lean and fuel-rich flames. Such techniques are
allowing researchers to examine the ion chemistry involved in the formation and loss of ions in flames (Goodings et al., 1979a&b).
Calcote (1981) reviewed many of the proposed
mechanisms of soot nucleation, both neutral and ionic, and presented a case favoring an ion--molecule scheme in which the ion C3H3 + plays a leading role. This scheme is based on the C3H3 + molecule's abundance and its characteristic
profile in a flame front. Before examining such an ion-molecule scheme and other possible routes leading to soot formation, we will present some cited experimental observations (Olson and Calcote, 1981a) that'had provided a basis for a possible ionic mechanism. They are 1) a positive correlation is observed between the growth of large ions and the appearance of soot; 2) the concentration of large ions in diffusion flames and the onset of soot formation are both increased by electrophilic molecules; 3) the propensity of a fuel to soot is directly related to its
ability to produce ions; 4) large concentrations of ions appear in pyrolysis experiments; 5) reactions of oxygen atoms with acetylene yield very rapid ionic polymerization;
6) concentration of the molecular ions peaks at the same location as the soot precursors; 7) the concentration of chemi-ions is estimated to be large enough to account for the concentration and rate of particles formed.
Wittig and Lester (1978) concluded from their
experimental observations on the effects of electric field that positive ions appear to play a role in the nucleation of carbon particles; however, it is not clear whether it is the charge on the ion or its chemical nature that is important in initiating the formation of carbon particles. Ionic Mechanisms
The overall process of soot formation from primary
molecular species to soot aggregate is similar for mechanisms involving both neutral and ionic species. The
difference is that for ionic mechanisms, the nucleation sites, along with PAHs and virgin particles, are ionic in character.
The primary objection to an ionic mechanism is the
small relative concentration of ions in flames as compared to hydrocarbon radicals (Michaud et al., 1981, Homann and Wagner, 1968). Figure 2-1 shows the concentration profiles
of key neutral and ionic species as functions of distance above a burner (Calcote, 1981). Examination of the data shows effectively that concentrations of ions are orders of magnitude less than of polyacetylene or free radicals;
however, the concentration of large positive ions exceeds that of soot particles, so it is clear that sufficient ions exist to serve as nuclei for particle growth (Wersborg et al., 1975). Also, it can be seen from this figure that the
build-up and decay of relevant species seem to be consistent with a sequential ion-molecule reaction leading to a rapid build up of large ions (300-1000 amu range) and then to formation of soot.
Although many ionic mechanisms leading to soot formation have been postulated, there is still no complete
agreement among workers for an ionic scheme describing the nucleation process. The controversy regarding any proposed
ionic route can be attributed to the sparse experimental information available on how positive ions initiate the growth of soot nuclei. It has been generally accepted that
1017 I I I I I
a 1012 PRECURSORS C14H
1011 LARGE POSITIVE IONS
too I I f I I I I
0 1 2 3 4 5 6 7 DISTANCE ABOVE TURNER. cm
Figure 2-1. Number Densities of Various Species Found
within Flames. Reprinted from the Article
by H. Calcote (1981).
formation of primary ions occurs in a very thin, reactive zone (Bittner and Howard, 1981), via the chemi-ionization reaction first proposed by Calcote (1962):
CH +0 -d CHO+ + e- (1)
or CH + C2H2 C3H3 + e (2)
Reaction (1) has been postulated to account for the large concentration of C3H3+ observed in fuel-rich flames (Knewstubb and Sugden, 1959). However, this reaction has been discounted as a major contributor to the total ion density (Kinbara and Noda, 1971 & 1973, Miller, 1973; Peeters and Van Tiggelin, 1969). Extensive proton transfer reactions that neutralize CHO+ follow rapidly so that this ion is never observed in large concentrations. Numerous flame ions obtained from secondary reactions have led investigators in flame chemistry to suggest various hypothetical reaction schemes in an attempt to explain the evolution of ion populations from the upstream region, through the reaction zone to the burnt gas region of a flame. Such a scheme includes
CHO+ + H20 H30+ + CO (3)
CHO+ + C2H20 C2H3O+ + CO (4)
C2H3O+ + C2H2 C3H3+ + CH20 (5)
CHO+ + C2H2 CO + C2H3+ (6)
CH3+ + C2H2 C3H3+ + H2 (7)
Reaction (3) leads to the formation of the dominant ion H30+ observed to persist downstream of the reaction zone.
Reactions (4) and (5) were suggested by Olson and Calcote (1981a&b) to account for the large concentration of C3H3 + in near stochiometric and lean flames. Reactions
(6) and (7) have also been suggested by the same workers as a possible reaction route that does not require oxygenated species to account for the large concentration of C3H3+ in rich flames (Olson and Calcote, 1981a&b).
Calcote (1981) has presented a case for an ion-molecule scheme in which C3H3 +plays a leading role in the
subseqent ion--molecule reactions in flames. This has been further elucidated by experiments of Olson and Calcote (l98lb), who have employed a mass spectrometer to sample positive ions at different equivalent ratios in premixed flames. Their results show that C3H3+ is the most abundant ion in fuel-rich acetylene and benzene flames. Further, as the flames are made richer, a drastic increase in large ion concentrations occurs at the expense of small ions (Figure 2-2).
Similar observations concerning the abundance and characteristic profile of C3H3 + throughout the flame front have been reported by Michaud et al. (1981). Although they have recognized the importance of C3H3+ for the subsequent
ion chemistry in the burnt gas region, they have been reluctant to accept an ionic mechanism leading to soot formation.
Experiments have been conducted by Smith et al. (1982), and Eyler (1984) to show the effect of hydrocarbon ionic structure on reactivity. In order to compensate for
Large Ions C3H3
... _7 9.v .
10-11 .C19H11 /
1.6 2.0 2.4 2.8 Equivalence Ratio
Figure 2-2. Effect of Equivalence Ratio on Peak Ion Currents
for 2.0 KPa Acetylene-Oxygene Flames. The Shaded
Area Indicates the Minimum Equivalence Ratio for
Soot Formation: Reprinted from the article by
the lack of thermochemical data for isomeric hydrocarbon ions, two C3H3+ structures (cyclic and linear isomer) have been investigated by Eyler (1983). Both have been made to undergo rapid reactions with a variety of unsaturated molecules, thus leading to higher molecular weight ions through condensation reactions.
C3H3 + +C2H2 C5H5+
|IC4H2 C7H5 +
Vinckier et al. (1977), in their O-atom reaction studies in flow tubes, observed that CHO+ is the dominant primary chemi-ion observed for real hydrocarbon flames, but other chemi-ions can be expected (C2H30+ has been observed in several flames). It has also been suggested by these authors that the secondary chemi-ion is obtained through the following proton transfer reaction:
CHO+ + C2H2 C2H3+ + CO
C2H3 then polymerizes rapidly with acetylene to form C2nH2n+l and C2nH2n-l.
+ C2H2 + C2H2 + C2H2
C2H3 C4H5 C6H7 8H3 etc
+- H 2 H -2H2
4H3 + C6H5 -8:H7 etc
These early ionic polymerization are interpreted as the nucleation steps in soot formation in rich flames.
Brill and Eyler (1981) proposed a possible sequential ion--molecule polymerization if C2H2+ ions were present in sooting flames:
+ + C2H2 + + + C2H2 + +
C2H2+ 0 C4H2+, C4H3+ C6H4 C6H5
The rate constants for the condensation reactions above are smaller for each successive step. The next overall step in the nucleation process is the growth of larger hydrocarbons by further addition of the building block CxHy, to the primary intermediate ion produced above. The growth process occurs by rapid condensation. Condensation--elimination reactions are followed by a rapid rearrangement from linear to a more stable polynuclear structures (Stein, 1983). This is due to the
characteristic of gaseous ions to rearrange to their most stable structure (Calcote, 1981)
Like the proposed neutral species mechanisms, the ionic mechanism also encounters some weaknessess in its formulation. The sequential ion--molecule mechanism of growth proposed by several of the workers above has been debated by Michaud et al. (1981). Because of the small concentration of ions in flames, they suggest that the source of large hydrocarbon ions can be explained as an equilibrium ion-molecule reaction between C3H3+ and larger neutral species. Goodings et al. (1979b), in their interpretation of the ion chemistry in rich CHy-C2H2-O flames, showed the relatively small importance of
condensation reactions when compared to proton transfer and, to a lesser extent, charge transfer throughout a flame in the chemical ionization of neutral species. Further, they have concluded, from their observations of hydrocarbon
anions downstream of the reaction zone, that these anions could be formed by proton transfer from neutral hydrocarbon
species which are resistant to destruction, by associative detachment with the H atom, i.e.,
CnHx C nHx + e
Finally, the latter authors have suggested that the negative ions might explain the neutral chemistry in the early chemical stages leading to soot formation.
Soot Emission Models
Numerous attempts have been made to develop a
comprehensive kinetic model for soot formation; however, models that estimate the rate of soot production in flames are very few. Although the hypotheses put forward by
several workers (above) were sought to be interpretive of the chemical aspect of soot formation, the number of elementary chemical reactions associated with the process of formation of soot particles are excessively large for inclusion in a predictive model. Furthermore, in real combustion flames, emission of soot particles is controlled not only by chemical kinetics but also by heat transfer and fluid mechanics. The interaction between these processes makes modeling of soot emissions even more difficult.
Few cited models on soot formation and oxidation will be discussed in this section. The purpose of this discussion is not to validate or refute any of the proposed models, but rather to give some insight to the complex problem of modeling soot emissions.
The most comprehensive model for predicting the rate of formation of soot from gaseous hydrocarbons is the one
proposed by Jensen (1974), who attempted to use detailed kinetic mechanisms to deal with all stages of soot formation. He suggested a model to describe the production of soot from a gaseous hydrocarbons in five steps: gas reactions producing radical fragments (fuel pyrolysis), nucleation, coagulation, growth, and oxidation. He considered fragments such as C21 C3, and C2H as initial nuclei. Although the predictions made for a methane flame
agreed with experiments, the utility of this model is limited because little information is available on the
intermediate elementary reactions in the pre-particle chemistry.
Tesner et al. (1971) developed a simpler model which uses a global rather than a detailed approach to characterize soot formation. It is formulated in terms of
equations that relate the rate of formation of active nuclei to the experimentally determined rate of formation of soot particles. In contrast with Jensen's model, Tesner's model does not specify the chemical nature of the
radical nuclei. The rate equations describing the formation of the radical nuclei and the soot particles are expressed as
= no + (f-g) g0Nn
n =concentration of active nuclei
no= spontaneous origination rate of the active nuclei f-g = linear chain-branching coefficient
go= linear chain-breaking coefficient N = concentration of soot particles a and b are empirical kinetic coefficients
The value of a was determined analytically to be equal to the product of b and the maximum measured concentration
of soot particles Nm at a corresponding time tin. Subsequently, the values of the other coefficients (N, f-g, and b) were found using an iterative technique.
The rate of spontaneous formation of radical nuclei is evaluated by means of a rate equation:
n= AC exp (-E/RT)
in which A is the frequency of oscillation of atoms along the bond being broken, E is the activation energy, and C is the concentration of fuel molecules.
Magnussen (1981) expanded such a two-step approach to the study of soot formation in turbulent flames to include the interaction between chemistry and turbulence. He
assumes that chemical reactions are confined to regions
consisting of fine structures, described as laminar sheets surrounding big eddies, generated by intense turbulent molecular mixing. The reaction volume and the mass transfer rate between these fine structures and the
surrounding fluid have to be known in order to treat the chemical reactions within this space.
A modified version of this model was devised by Galant et al. (1984). The calculations of soot concentration
calculations reported by Magnussen and Galant using both laboratory and large scale diffusion flame show satisfactory agreement with their corresponding experimental results.
A model based on a so-called "quasi-global" scheme was introduced by Edelman and Fortune (1969). This model is a compromise between the complex detailed mechanism and the simplistic but less informative global reaction scheme.
The quasi-global kinetic mechanism is an extension of the notion of a single-step overall reaction to a set of important reactions. This scheme characterizes the partial oxidation of large hydrocarbon molecules by sub-global
steps coupled with a set of finite, reversible reactions for which sufficient information exists, and treats soot formation similarly to the formation of other species.
Edelman et al. (1979) developed an expression for the rate of soot formation in a jet-stirred reactor.
R= A T a [HC~b [021c exp (-E/RT)
where a, b, and c are dimensionless constants, E is the activation energy, and [HCJ is the concentration of unburned hydrocarbon.
The rate of soot formation and other global rates are evaluated by direct comparison of predictions with experimental data. The quasi-global mechanism predicts the unburned hydrocarbon and oxygen concentrations and
temperature required in the global reaction for soot formation. Net soot production can be estimated using the following conservation equation (Farmer et al. 1981):
Cn C c- =A R -R
S sQ t ox f
where in, p, V, c s, C S''and A't are total mass flow rate, exhaust gas density, reactor volume, outlet and inlet
soot concentrations, and the total surface area available for oxidation, respectively. The reaction rates for the soot formation and oxidation steps are Rf and Rox, respectively.
A quasi-global kinetic model for toluene and isooctane developed by Edelman et al. (1983) showed an
acceptable fit with experimental data. A set of Exxon jetstirred combustor experiments was the primary source of data to establish the model.
Khan and Greeves (1974) presented a semi-empirical Arrhenius type equation for the local rate of soot
formation in diesel engines. The phenomenological aspect of soot formation was not suggested in this model.
Furthermore this model is based on experimental values of local temperature, unburned hydrocarbon concentration, and the local equivalence ratio which can change from one experiment to another.
The oxidation step is very important in
understanding soot particle emissions. The multitude of potential oxidants (0, 02, OH, CO) that can participate in the oxidation step, depending upon conditions, prevents
complete agreement on a universal formula that will best predict the oxidation rate of soot particles in flames.
Radcliffe and Appleton (1971) suggested that the surface mechanisms for oxidation of pyrolytic graphite and of soot are similar. A semi-empirical expression,
originally proposed by Nagle and Strickland-Constable (1962) for oxidation of graphite, was used by different workers (Chul and Appleton, 1973; Edelman et al., 1979; Appleton, 1973) to predict the rates of oxidation of soot. The rate of oxidation of soot per unit area of soot surface is formulated as follows:
R ox = 12 k A P o2 x k B P o2 (1-X) 1+k z P o2
where k t_
This rate equation is based on the theoretical model of the surface oxidation rate mechanism originally proposed by Blyholder et al., (1958). The values of ka, kb, kt, and kz were obtained by Nagle and Strickland-Constable (1962) to fit their measurements.
ka = 20 exp (-30/T) g cm1 sec-1 atm-1
kb = 4.46 exp (-7,460/T) g cm2 sec-1 atm-1
kt = 1.51 105 exp (-48,800/T) g cm-2 sec-1 atm-1
kz = 21.3 exp (2,060/T) atm-1
Lee et al. (1962) measured rates of oxidation of soot in a laminar hydrocarbon flame. Soot particles and samples of combustion gases as well as temperature were measured at different cross-sections of a hollow soot column. A semiempirical rate equation that is dependent on the temperature and partial pressure of oxygen was suggested:
Rox = 1.085 104 P02 T-1/2 exp (-E/RT)
The first-order dependence of rate upon the partial pressure of oxygen was criticized by Appleton (1973). He concluded from his experimental results in a shock tube that at fixed temperature and low oxygen partial pressure, the surface oxidation rate is first order in partial pressure of oxygene and approaches zero order at high pressures.
Neoh et al. (1981) investigated the potential oxidants of soot in both fuel-rich and fuel-lean flames. Their experiments showed OH contribution to be more important than 02 in the oxidation of soot in atmospheric flames
between 1575 and 1865K. Comparison of their experimental rate to the rate predicted from the Nagle--StricklandConstable formula agreed for lean flames; however, discrepancies between the two rates became larger as the flames were made richer in fuel. According to Neoh and coworkers, the neglect of the contribution of OH in the
predictive formula of Nagle and Strickland-Constable was the cause of the observed discrepancies.
To interpret soot-related measurements of chemical
species and concentrations of particles in flames, it is necessary to determine the parameters that affect the processes governing soot formation and oxidation and to attempt to explain the drastic changes in the evolution of the soot field that occur between fuel-lean and sooting flames.
It has been recognized that the equivalence ratio correlates the tendency of fuel to soot (Street and Thomas,
1955) ; however, this parameter is based on a reaction scheme where only C02 and H20 are produced in oxygen richflames. This is not the case at the sooting limit where carbon monoxide is in excess of C02' Takahashi and Glassman (1982) proposed a more-general formalization for
expressing the tendency of fuel to soot. It is based on a stoichiometric reaction in which the carbon is converted to carbon monoxide and the hydrogen is converted to water. The proposed variable for indicating the tendency of hydrocarbons to soot in premixed flames is the ratio (C +
H/2)/O, where the numerator expresses the stoichiometric oxygen requirement on this new basis and the denominator is the actual experimental value of oxygen at the onset of sooting; however, in diffusion flames soot, particles are f ormed in the f uel side of the f lames (Kent et al 198 1) where the concentration of oxygen is insignificant.
Hence, the equivalence ratio is not an appropriate parameter that can correlate the sooting tendency of a fuel
in a diffusion flames. Next, the recent developments in the study of soot formation in shock tubes and flames will be reviewed.
Pyrolysis of Fuel
Shock tube techniques have been very useful in the
study of soot formation. They allow one to produce time invariant, controlled conditions of temperature and pressure similar to those in real flames. This permits examination of the temporal behavior of reactant and
product concentration at constant temperature and pressure.
Selected fuels have been investigated by pyrolysis behind incident or reflected shock waves (Graham, 1977; Frenklach et al., 1982;) and are reviewed briefly here.
Influence of the induction time
Initial formation of carbon particles occurs after an induction period that is characterized as a function of temperature, pressure, and molar concentration of hydrocarbons This parameter, although subject to
criticism related to the credibility of its measurement, can be useful when relating soot formation to residence times in particular regions of combustion chambers. Influence of pressure
Frenklach et al. (1983 a&b) observed that for homogeneous hydrocarbon/air mixtures at low pressure (P < .1 bar), the maximum measured soot yield is shifted to
higher temperatures; however, at higher pressures ( 1 to 10 bars) an increase in pressure increases soot yield slightly. This dependence of soot yield on pressure during
soot formation in pyrolysis of aromatic hydrocarbons was interpreted by Frenk lach et al ( 198 3a ) as f ol lows: A decrease in pressure slows down decomposition reactions and at constant soot yields production, an increase in
temperature is needed to compensate for the decrease in total pressure. This explains the shift of a soot-bell maximum to higher temperature. Influence of temperature
Several experimental observations (Frenklash et al., 1982, 1983b; Wang et al. 1981) have shown a bell shaped dependence of soot yield on temperature in the pyrolysis of
a variety of hydrocarbons. Frenklach et al. (1983a) concluded from an extended analysis of xylene pyrolysis that the position of the maximum in soot yield is not universal but rather dependent on observation time,
pressure and extinction mode. Rawlins et al. (1982) observed no soot in toluene pyrolysis below 1500K or above 2500K, and a maximum soot yield of 30-50% of the initial fuel carbon near 2000K.
Mass spectrometric and optical techniques have been used extensively in premixed flames to identify hydrocarbon species (stable products, radicals, and ions) which can serve as detailed probes of combustion chemistry. The
soot formation chemistry in premixed flames is not complicated by the diffusion processes that are characteristic to diffusion flames.
Prado et al. (1981) identified two regimes for soot particles growth in premixed flames. For soot volume fractions less than 10-7, particle growth occurs by coalescent coagulation. At larger soot volume fractions, chain-forming collisions occur simultaneously with surface growth. Laser light scattering and absorption
techniques have been used to characterize particle size and
number density profiles in premixed flames (D'Alessio et al., 1977; Haynes et al., 1981). It has been observed that the number of particles near the reaction zone is large (1011-1012 cm-3) and the corresponding "soot" particles are
small (the smallest particle diameter detectable by laser light scattering methods is about 2.0 nm). Smyth and
Mallard (1981), using laser induced ionization in premixed flames at the sooting limit, measured particle masses in the range 2300-6100 daltons, having diameters of 1.62.2 nm. This is in the range of the lower limit of particle diameters (1.5 nm) that can be detected by electron-microscopic techniques.
Beyond the soot limit, the yield of soot in premixed systems increases very rapidly with C/o ratio. This is the
result of increased nucleation sites for elementary soot particles to form as the carbon--oxygen ratio increases (Wagner, 1981).
The effect of metal additives on soot formation in premixed flames was recently investigated by Haynes et al. (1981). This work investigated the impact of alkalimetals and alkalineearths injected into the post reaction zone of premixed flames. They concluded that these additives do not reduce soot as expected by many workers, but rather, they increase the number of smaller, charged soot particles that subsequently resist coagulation by Coulombic
Glassman (1979) argued that temperature is a very significant parameter associated with processes of soot
formation and oxidation. He supported Millikan's concept (1962) that under premixed conditions the rate of soot
formation is controlled by the competitive process between
the rates of pyrolysis of f uel and oxidation of soot precursors. Because the rate of fuel pyrolysis is less sensitive to an increase in flame temperature than the rate of soot oxidation, raising the temperatures lowers the tendency to soot.
Harris (1982) examined the processes leading to surface growth in premixed C2H4/air and C2H4/Ar/02 flames.
He concluded that acetylene in combustion products is the source of mass for most mature soot particles Furthermore, this author determined that C2H2 occurrence is a first-order with respect to soot growth.
Calcote and Manos (1983) reviewed the literature and investigated the results of soot measurements in both premixed and diffusion flames. They defined a numerical index called the "Threshold Soot Index" (TSI), which can be
used quantitatively to compare and analyze sooting flame data. This soot index ranks the fuels from 0 to 100 (0 least sooty).
Diffusion flames are very important in simulating phenomena occurring in most combustion devices. Unlike
premixed flames or the homogeneous mixture used in shock tube research, the diffusion flame is mixing controlled.
The height at which a diffusion flame starts to smoke and the corresponding luminosity of the flame are considered a measure of the tendency of a fuel to form carbon. Velocity has a direct effect on the height and the
brightness of the flame. As the velocity of the fuel jet increases, both the flame height and the luminosity increase. Finally, a limit is reached where the flame begins to emit soot. Unfortunately, spatially resolved mapping of soot distribution in diffusion flames is difficult owing to steep concentration and temperature gradients, and to the inhomogeneity of the mixture in such flames (Kent et al., 1981) As particles travel through the flame field, they pass through regions where conditions
vary widely. Concentration, Size, temperature, and flow velocity of soot particles are not only dependent on the height in the flame, but also on radial position owing to non-axial flow velocity and particle diffusion.
The development of laser light-scattering measurements and other non-intrusive optical techniques has enhanced the spatial resolution of particle mapping throughout the flame. In their study of soot particle formation and growth in a co-annular diffusion flame, Santoro et al. (1983) observed that the flame can be divided into two regions. The first is characterized as a region of growth in which soot formation dominates. The second is characterized as a region dominated by oxidation processes. Rates of soot production have been found to be higher near the base of the flame where the temperature range is 12001400 OC; The production rate decreases with the distance from the flame (Kent et al., 1981). The soot formed near the base of the flame is transported upward by convection
and radially by both diffusion and convection (Santoro et al., 1983; Kent et al., 1981). Kent et al. (1981) observed that small flames have higher initial soot volume fractions and particle growth, but larger flames have a greater soot volume fraction higher up in the flame because of their longer relative residence times.
Glassman (1979) determined that in diffusion flames, oxidative rates do not increase as fast as pyrolysis rates with respect to temperature. Thus, the higher the temperature, the greater tendency to soot. Wey et al. (1982) confirmed the latter result in their studies. These researchers observed that higher temperatures also lead to large soot particles and agglomerates that are difficult to burnout downstream of the flame. Turbulent Diffusion Flames
The unsteady nature of turbulent diffusion flames results in a nonhomogeneous distribution of fuel in air. Conceptually, this can be represented as pockets or eddies of gas with different fuel-air ratios (Prado et al., 1977) distributed throughout the flame region. Soot is assumed to be formed and contained in these eddies, which are separated by regions containing little or no soot (Magnussen et al., 1979). These inhomogeneities are
attributed to the interaction between turbulence and chemical reactions.
It has been suggested by Becker and Yamazaki (1977) that soot formation and burnout in diffusion flames is essentially mixing controlled. The principal aerodynamic
parameter that controls soot formation is the Richardson ratio (Ri), which relates the effect of buoyancy to momentum forces. Becker and Yamazaki discounted the Reynold's number (Re), as an important parameter, since the action of buoyancy, even at low Reynolds number, increases the downstream momentum flux such that virtually all flames are fully turbulent; however, Magnussen (1981), in his
eddy-like interinittency model, observed that for a given nozzle diameter, variations of the Reynolds number significantly affect the amount of soot formed.
Nishida and Mukohara (1982) demonstrated that when the
inlet air temperature is raised from 50 to 5000C, the gas temperature on the flame axis increases by 100-3000C. The concentration of soot and hydrocarbons was shown to be greater and the decay of soot concentration more rapid in the higher temperature flames.
Characterization of molecular species and particulates
as they exist in a combustion environment is necessary to improve the understanding of the processes that lead to their formation and destruction in that environment. Such
characterization requires a diagnostic technique that does not interfere with the measured properties of interest and
that can provide adequate spatial and temporal resolution of the data.
Physical sampling probes have been used to observe changes in particle size and population density in the
combustion environment; however, they are inherently slow and intrusive, and they require tedious sampling procedures and analysis. The emergence of optical techniques has enhanced the study of molecular species and particulates in combustion. These techniques provide rapid
data acquisition and resolution while causing only minimal disturbance to the actual phenomena. This research effort
stressed the use of an optical technique to measure size and population density of soot particles (Integrated
Particle Sizing System (IPSS)). Physical probes provided necessary support for validation of the IPSS, and they were
also used in this work to complement data collection in such areas as temperature measurement and gas sampling, for which optical instruments were not available.
The burner used for the current study is a nonpremixed axisymmetric bluff-body type combustor exhausting into free air. Studies on this type of combustors were
initiated by the Ai r Force Wright Aeronautical Laborator ies/Aeropropuls ion Laboratory (Roguemore et al., 1983). The simple configuration of such combustors
provides easy optical access for measurement purposes and good flame stabilization.
The bluff-body burner (Figure 3-1) consists of a central fuel pipe surrounded by an annular region through which air flows. The fuel passes through the central pipe and exits through a 1.37-mm diameter nozzle located in the center of a 5.72-cm OD aluminum centerbody. Compressed air
(relatively free of contamination by particles, water, or oil droplets) enters the plenum chamber surrounding the
central fuel pipe, passes through flow straighteners, and exhausts at the burner exit plane, where the mixing process occurs. The fuel, nitrogen dilution, and air flow rates through the system are monitored through an easily accessible control board (Figure 3-2).
Figure 3-1. Schematic of the Bluff Body Burner
Figure 3-2. Control Board
Extractive Probe and Sampling System
Sample extractions at a point in a flame or in the exhaust system were performed with an extractive probe
(Figure 3-3) provided by the University of California Irvine Combustion Laboratory. This sample probe was designed to collect soot samples and combustion gases for chemical and physical analysis. The design considerations of the extractive probe and sampling system were carefully
reviewed (Hack et al., 1981) and will not be presented in this report; however, a succinct description of the probe and the sampling system is presented below.
The probe consisted of a sample transport tube surrounded by an annular jacket through which cooling water circulated. Stainless steel (316) was used for all
surfaces exposed to the sample or the combustion environment. The outside diameter of the probe was 0.375 in (0.953 cm). and the sample tube diameter was 0.12 in (0.305 cm). The overall length of the probe was 25 in (63.5 cm) (from the tip to the mounting plate) with a curvature of 20 in (50.8 cm).
Cooling water was carried through seven 0.043-in (0.109 cm) ID tubes inside the annulus to the probe tip and returned through the remaining annular space. The cooling water was circulated to maintain the probe temperature below the design limit of stainless steel.
Figure 3-3. Extractive Probe
Nitrogen was injected through thirty-two 0.031-in
(0.079 cm) diameter holes organized into eight groups of four holes. Dilution with nitrogen was necessary to minimize particulate deposition on the walls of the
sampling line. A separate port in the mounting plate was included for extracting gas samples.
The sampling system consisted of a filtration bank in which Nuclepore filters (0.47 mm OD) and 0.2 micrometer
pore size were used, a dry test meter (DTM) four rotometers for volume determination, and a vacuum pump for isokinetic sampling.
Direct particle collection.
The face of the burner and a filter (Figure 3-4) located about six feet (183 cm) downstream in the exhaust system were chosen as sites to collect soot particles for chemical analysis. The former was chosen because of the relatively low temperature existing at the face of the
burner (this cold surface allowed trapping of stable products that could be related directly to the intermediate
hydrocarbons that form soot particles). The latter site was chosen to determine the chemical composition of
components that survived the oxidation process and to determine their toxicity in relation to the environment.
Temperature measurements in sooting flames are difficult because soot deposits on a thermocouple causes an
Figure 3-4. Soot Particle Filter.
increase in conductive and radiative losses; however, such
measurements do provide valuable information that can be used with soot particle and velocity measurements to complement the study of soot evolution in flames.
Temperature profiles were obtained with Pt/Pt-10% Rh fine wire thermocouples that were made of 0.006-in
(0.015 cm) diameter coated wires (plat inum--rhodium as a sheat material and magnesia for insulation) with a welded junction of nearly the same diameter.
Integrated Particle Sizing System
This study involves the use of an Integrated Particle Sizing system (IPSS) capable of providing nonintrusive, insitu, real-time measurements of particulate size and
population density in flames. The technique is based on the theory for scattering of a plane, linear ly-polar ized light wave by a single, homogeneous, isotropic sphere of arbitrary size and refractive index, first formulated by Mie in 1908 (Kerker, 1969). Since the technique is based on light scattering from spherical particles, Mie theory is discussed first, followed by the system description.
Mie's theory is based on a formal solution to Maxwell's equations, with appropriate boundary conditions applied to the scattering of light by a homogeneous sphere
of arbitrary size and refractive index. The mathematics
and the corresponding assumptions that led to the
formulation of the solution are given elsewhere (Van de Hulst, 1981; Kerker, 1969) and only a few results relating the solution to the problem of interest are repeated here.
The solution takes the form of an infinite series of Bessel and Legendre polynomials. The radiant flux P
scattered by a particle can be expressed as
P~~ =A' X2 2 2 *l (X,mO)sin 2 +i(XmQ)cos2 sinededo
0 2 Ul 11
4 Tr 1 1
where Io is the incident intensity, X is the wavelength of the incident radiation, r is the distance from the
scatterer, m is the complex index of refraction, 8 is the scattering angle, X = is the size parameter, and 4 is the angle between the scattering plane and the plane of polarization of the incident radiation. The Mie scattering functions for light polarized perpendicular and parallel to the scattering planes are iI and i2, respectively. These functions are related to the amplitude functions as
iI (X,m,8) = S1 (X,m,6)
i2 (X,m,6) = S2 (A,m,O)
Also it is customary to define the partial scattering cross-section C as the power of light divided by the the incident intensity (Sterling,1986) as C = P/I0
The Integrated Particle Sizing System uses a ratioing technique for particle size determination. The intensity ratio of the power of the radiation scattered by a particle
and collected in the forward direction by lenses located at two different angles is
P (0e1) C (6 1)
P ( 2) c (62)
where is the large angle and is the small angle formed by the lens axes with the axis of the incident beam. The
incident laser intensity cancels out, eliminating the problems associated with the Gaussian intensity distribution of the laser beam.
The intensity ratioing technique requires the
determination of the Mie intensity functions il and 2
which can be generated by computer codes developed by a number of workers in this field (Dave, 1968). The input parameters are the size parameter ( wrD/X), the wavelength of the incident radiation, the scattering
angle, the refractive index of the surrounding media, and the complex refractive index of the sphere. A plot of the
intensity ratio versus particle size for scattering at 200 and 600 is presented in Figure 3-5. The computation of
L 1 .5 7-O0.56 i
.7I L (0V
Figure 3-5. Intensity Ratio vs. Particle Size
intensity ratio versus particle size was carried out by Sterling (1986). The particle refractive indexes were for latex (m = 1.60) and soot (m = 1.57 0.56i).
The Integrated Particle Sizing System includes 4 techniques for in-situ measurements of particle size, droplet size, and particle velocity. These techniques are
(1) large-angle intensity ratioing, applicable to particles 0.08 to 0.6 pim, in diameter; (2) small angle intensity ratioing, applicable to particles 0.3 to 3.0 pim in diameter; (3) particle sizing interferometry, applicable to particles 2.5 to 45 Vim in diameter; (4) droplet sizing interferometry, which is applicable to droplets 2.5 to 140 pim in diameter. Because the focus of this work is characterization of submicrometer-sized soot particles, the
large angle intensity ratioing technique was adopted for particle size and number density measurements.
The concept of the large-angle intensity ratioing technique is to collect the forward-scattered light from individual particles as each crosses the probe volume at two off-axis angles. This instrument was used to deduce their respective sizes from the ratios of the collected intensities. The relationship between particle size and intensity ratio is obtained from the solution of Maxwell's equations (Kerker, 1969) describing the scattering of electromagnetic waves as derived by Mie for spherical particles of arbitrary size and
refractive index. The input parameters for this technique
are the collection angle pair and index of refraction of the particles. The angle pair determines the particle size range to be measured.
Random stray signals associated with multiple
scattering can overwhelm the scattering signals from the particles to be measured; however, this instrument can also be used as an optical counter if single-particle
scattering is dominant over multiple-particle scattering. Single particle scattering requires that the laser light
scatter from a single particle passing through the probe volume and propagate directly to the detectors without interfering with any signals from other particles that may be present in the particle cloud surrounding the probe volume. Single particle scattering can be promoted by providing a small probe volume at the test section, which, when combined with a high signal-to-noise ratio and fast data resolution, can enhance the performance of the system as a particle counter.
A schematic diagram of the measurement system is
given in Figure 3-6. The incident radiation is supplied by a 2-W Argon Ion Laser operating in the multiline mode. The 488 nm line used for intensity rationing is separated from the other lines by a dispersion prism. The beam is aligned by a series of mirrors, expanded by a variable beam expander, and linearly polarized in the direction vertical to the plane of scatter by a dichroic mirror. The
q o L6
Mirror Beam Expander [ L7
Apple II Microcomputer
Figure 3-6. Schematic of the Integrated Particle Sizina System.
beam is focused by an 300-mm focal length lens to produce
a waist that, in conjunction with the collecting optics, defines the sample probe volume of the instrument.
The light scattered by a particle passing through the probe volume is collected by two pairs of identical 300mm focal length, f/4 lenses in the horizontal plane of the beam. The two pairs of lenses are contained in two cylindrical enclosures located at two angles off-axis of the incident beam. The angle pair (600/200) was used in
this study to yield measurements of particle size in the range 0.08 to 0.39 pin. The focus of the first lens coincides with the sample point of the instrument. The first lens of each receiving system collects the scattered light and collimates it. The second lens focuses the beam upon a pinhole in front of a photomultiplier tube.
The detectors are RCA 8575 photomultiplier tubes (PMTS) powered by a high-voltage power supply (Bertran Associate, model PMT-20A/N (option 3)). These PMTs utilize a photosensitive cathode and a number of dynodes to convert the scattered light into current. The relationship between the collected light impinging upon the PMTs and the corresponding output current can be expressed as
i = c(F)(lOkE)
where F is the radiant flux striking the cathode, E is the voltage supplied to the PMT, and c and k are constants specific to the PMT; F is equal to the power of the
scattered light, P, passing through the pinhole aperture divided by the sensitive area of the cathode.
Each PMT is equipped with a band pass filter with a bandwidth at half maximum of 2 nm and a transmitance of 20% at 488-nm. These filters discriminate against the background signals generated by stray light or luminous light from flames.
A dual logarithmic amplifier from Spectron Development Laboratories (SDL), Model LA-1000, used to convert the current pulses from the PMTs into positive voltages pulses, is scaled to a maximum output of 10 V for an input current of -1 mA. For pulse widths not less than 10 vs, the amplifier is rated for current amplitudes as small as 1 nA; for pulse widths as small as 2 ps, it is rated for amplitudes as small as 1 pA. The output of the log amplifier can be expressed as
V = 2 log(i/io)
The output pulse voltages from the logarithmic amplifier are sent to a ratio processor (SDL, RP-1001). These signals are first buffered and peak detected. The dc level outputs from the peak detectors are fed to an analogue subtractor circuit which mechanically divides the signals at the large angle (A) by the respective signals at the small angle (B). The subtractor output is amplified to 0 to 10 volts and then converted into an 8-bit binary number by an analogue--digital converter. An output of 10 volts corresponds to a ratio (A/B) of 0.1 and 0 volts corresponds to a ratio of 1.0.
The ratio processor uses the seven most significant bits of the data word for data acquisition; the 8th bit in
the data word is used as a status bit to indicate whether the current data word is valid or invalid. A zero level
indicates a valid data word. This is determined by it meeting three conditions:
a. The B input signal level must be less than 10 volts.
b. The A input signal must be less than the input signal
B and greater than a minimum level of 4 volts for high
speed operation or 2 volts for low speed operation.
c. The width of the A input signal pulse must be greater than 2 ps for high speed operation or 10 ps for lowspeed operation.
The digitized signals are then transferred to a
microcomputer for data analysis. A histogram generator software package resolves the data output into 62 bins, of which each corresponds to a discrete size range. Typical output data of this technique are shown in
Spherical latex particles of known size were used to evaluate the performance of the optical system both as a particle sizer and as a particle counter. These particles,
which were supplied in mixtures in water and surfactant, were 0.5% and 10% solid by weight. They have a density of
1.05 g/ml and a refractive index of 1.57 at 488 nm.
Samples were prepared by diluting a highly concentrated particle suspension in de-ionized water or alcohol. The resulting suspension was passed through a
0.33 un LTX Particles
0 .05 .1 .15 .2 .25 .3 .35 .4 .45 .5 Size (r)
Figure 3-7. Typical Optical Data Output.
constant output atomizer to form droplets. A schematic diagram of the aerosol generation system is shown in Figure 3-8. Clean, dry air was supplied to the atomizer through a 0.343-mm diameter orifice to form a high velocity jet. The suspension was drawn by the jet to the atomizer, where large droplets were removed by impaction on the walls of the atomizer assembly. Excess liquid was removed continuously to a waste container.
A syringe pump (model 975, Harvard Apparatus Co.) was used to provide a regulated hydrosol flow rate to the atomizer. The pump has a synchronous motor and is capable of providing stable constant flow rates. The pump includes a 30-position gear box that allows the liquid flow rate to be varied from 0.027 to 46 cm3/min when using a 50 cm3 standard syringe.
The wet aerosol coming out of the atomizer was passed through an evaporation/condensation aerosol conditioner (TSI model 3072) to improve its monodispersity. A
diffusion drier (TSI model 3062) was used to dry the aerosol; the aerosol was then injected through a nozzle centered just below the optical probe volume for measurements.
clean dry air
pol ydisperse aerosol
liquid drAin it.,evaporation section
liquid /condensation section
silica gel instant output atomizer
diffusion drier monodispersed aerosol
Figure 3-8. Aerosol Generation System.
EVALUATION OF THE OPTICAL SYSTEM
A detailed analysis of the Integrated Particle Sizing System (IPSS) to evaluate its overall accuracy in measuring sizes and number densities of particles and its
adaptability to different testing environments (reacting and non-reacting flows) is a necessary prerequisite for interpretation of empirical data for particulate
characterization in combustion systems. The performance characteristic of the optical system is a strong function of the state and nature of the particle delivered to the probe volume as well as of the medium surrounding it.
The IPSS measures the rate at which particles pass
through the sampling space defined by the intersection at the waist of the incident laser beam and the collecting
optics of the system; however, the effective sampling volume can be affected by the intensity variation in the sample space (Figure 4-1) and the scattering characteristics of the particle to be measured. Therefore some refinement of the measured data for the IPSS can be realized, through probe volume corrections that would take into account both of these effects.
Experiments were performed to test and evaluate the optical system performance as a particle sizer as well as a particle counter in both cold (dispersion of latex
100 ........... Theoretical
90 0 .0 ...> Experimental
80 .' 0"
r 70 0 ..
0 0%50 0 "
0 30 0L20 ". 0
-500-400-300-200-100 0 100 200 300 400 500
Contour Radius (Am)
Figure 4-1. Laser Beam Intensity Distribution.
hot (combustion) media. The results of these experiments are presented next.
Latex particles of known size were used to validate the system's performance as a particle sizing-instrument. These particles were diluted in liquid methanol to avoid scattering introduced by water droplets. Figure 4-2 shows the measured count mean diameter versus the nominal diameter of the particles. The measured diameters of the particles agree well with the corresponding sizes specified by the manufacturer.
The system's performance as a particle counter, in the limits of the measured size range, was investigated by
measuring the particle count rates from suspensions with different latex particle concentrations; however, the flux rate at which particles pass through the probe volume was found to be dependent both on the particle concentration of
the suspension and on the flow rate through the atomizer. Therefore, in order to verify the reliability of the system as a particle counter, the operating conditions of the atomizer had to be determined.
The characteristic operating conditions of the
atomizer assembly were investigated using two suspensions with particle concentrations (Cl) and (cl)/2; (Cl) was the reference particle concentration. The flow rate of the particle suspension was maintained at a known level with a
E 5 0 Data
.2 .25 .3 .35 .4 .45 .5 .55
Nominal Particle Size (,a~m)
Figure 4-2. Measured Diameter versus Nominal Diameter (pim)
syringe pump. The particle count rates versus flow rates for the two suspensions tested are presented in Figure 4-3.
The results are consistent with expectations only at the lowest particle suspension flow rates. The effect of
suspension concentration on the atomizer output decreased when the suspension flow rate was increased. The atomizer
output became independent of the concentration of the suspension concentration when the flow rate was set above
one might speculate that a suspension having a high concentration of particles produced large droplets whose greater inertia caused their impaction on the walls of the atomizer resulting in a decrease of its output. In order to
establish the particle concentration as the main variable controlling the atomizer output, the impaction of the
droplets was reduced to a minimum by using low suspension flow rates.
The measured size distribution of the latex particles was found to be independent of the suspension concentration. Figures 4-4a and 4-4b show the normalized size distribution histograms for a suspension of 0.412-pm latex particle at concentration (Cl) and (Cl)/2, respectively. Virtually no shift in size distribution is observed.
The measured particle count rates for four suspensions
with particle concentrations of (Cl), (Cl)/2, (Cl)/4, and
(Cl)/8 are presented in Figure 3-5 for particle solution
1.2 atceSseso ocnrto C /
0...- Particle Suspension Concentration (Cl)/
z-0 Partice_ SusensioConcetratin___z
2 3 4 5 6 7 8 9 10
Particle Suspension Flow Rate (cm 3/M in)
Figure 4-3. Particle Count Rate vs. Particle Suspension
Volumetric Flow Rate.
S.4- .4N L
Z .2 Z .2
o o ---0 .1 .2 .3 .4 .5 .6 0 .1 .2 .3 .4 .5 .6
Size (Um) Size (pm)
Figure 4-4a & b. Normalized Particle Count Rate
vs. Particle Size.
SSuspension Flow Rte 4.4 cm3/min 0-Suspension flow Rote 3.1 cm3/min
-0-- Suspension Flow Rote 2.2 cm3/min
Z .4S .20
0 I I I I
0 .2 .4 .6 .8
Figure 4-5. Particle Count Rate vs. Particle Suspension Flow Rate (Normalized)
flow rates of 2.2, 3.1, and 4.4 cm3/min. For each particle injection flow rate, the measured particle count rates were
normalized to the arbitrary selected count rate Nl, which was measured for the suspension with a particle concentration (Cl) and injected at the same flow rate. The count rates were found to vary linearly with the suspension particle concentration.
This part of the study was undertaken to characterize
the performance of the optical system in real combustion situations and to perform the necessary adjustments to the optical system for optimization of data acquisition. The adjustments to the receiving optics and to the power supplied to the photomultipliers were investigated and the
reproducibility of the data was verified. validation of and adjustments to the system are discussed below.
The combination of high intensity and number density of the signals obtained from sooting flames resulted in saturation of the signals sent to the ratio processor. This saturation may be interpreted as a result of the PMTs output consisting of current pulse values exceeding the
input ratings of the dual-channel logarithmic amplifier. The latter has two ratings for pulse widths no less than 10 sec for current amplitudes as small as 1 nA, and pulse widths no less than 2 sec for amplitude as small as 1 A.
The high output current was directly related to the high radiant flux of scattered light from the soot particles,
the radiant flux from the flame, and the power supplied to the PMTS. The dependence of the output current on the radiant flux and the PMT supply voltage can be expressed as
where E = supply voltage
F = radiant flux striking the cathode
and C and k are constants associated with a particular PMT.
Therefore, in order to reduce the high radiant flux striking the cathode of the photomultiplier tube, a mask having a I-in diameter hole at the center was placed on
each set of the receiving lenses. The amplitude and the frequency of the current pulse from the PMTs were thus
diminished sufficiently to enable individual processing of the scattered signals from the soot particles.
Measurements of size and number density of soot particles were taken consecutively with and without the masks inside a propane flame. These results are presented in Figures 4-6 and 4-7 for size and number density
measurements, respectively. The use of the masks on the receiving lenses had a dramatic effect on the soot particle number density but had only a slight effect on the size distribution.
The measured size distribution confirmed the
computations conducted by Sterling ( 1986) on the independence of size measurement using the intensity rationing technique upon collection solid angle. Sterling's
results for scattering of light by soot particles are
Without Masks With Masks
4- 4E 06
Z .2 7- .2
0N I -0- n .I
.07 .12 .17 .22 .27 .32 .37 .07 .12 .17 .22 .27 .32 .37
size (jim) size (Uim)
Figure 4-6a & b. Normalized Particle Count vs. Particle Site
(a) without Masks, (b) with Masks on the Receiving Lenses.
.... With Masks
--0- Without Masks
V) 0 i I i I I I
0 .5 1 1.5 2 2.5 3 3.5 4 4.5 5
Axial Locations (x/R, r/R=O)
Figure 4-7. Soot Particle Count Rate at the Centerline
With and Without Masks on the Receiving Lenses.
presented in Figure 4-8; they were obtained by direct calculations of the scattered intensities in the direction
of the optical axis and by integrating the scattered intensities over the collecting lens aperture. The direct and integrated results were found to be nearly identical.
The soot particle number density was found to be higher inside the flame when the masks were used (Figure 4-7. The latter result was expected because of the attenuation of the radiant light striking the cathode. The
soot number density outside the flame was found to be higher when the masks were not used. This result was also
expected because the radiation flux from to the flame was almost absent, so only the laser scattered light from the soot particles struck the cathode of the PMTS.
Masking alone was not sufficient enough to reduce the
amplitude and frequency of the signals from high sooting flames to an acceptable level for the ratio processor.
Subsequently lowering of the voltages supplied to the PMTs afforded the necessary decrease in overall intensity and frequency of the current leaving the PMTS.
The voltages applied to the PMTs were decreased such that the size distribution at a point inside the flame stayed the same. Figure 4-9 presents the results of the voltages that had to be applied to the PMTS in order to obtain the same soot particle size distribution at a point far downstream of the flame, where the effect of the
luminosity was negligible. The signal traces from the logamplifier were monitored with a dual-trace oscilloscope.
1 "0. .... Integrated
.9 --0-- Direct
m .6 0o .5
c .4-'-.30 .05 .1 .15 .2 .25 .3 .35 .4 .45 .5 .55 .6
Figure 4-8. Calculated and Integrated Scattered Intensities
vs. Particle SiZe.
--0-- 0.31 pm soot particles
900 I I I I I
900 1000 1100 1200 1300 1400 1500
Figure 4-9. Applied Voltages on the Photomultiplier Tubes.
The system stability was evaluated for a propene flame
operating at an air/fuel mass ratio of 600 and an air velocity of 14 M/s. The data were taken at the same locations inside the flame on two different days. The results, presented in Figure 4-10, show an adequate
reproducibility of the data with a maximum error of about 10%.
1600 1500 % 1400 -.....
9004 S800 CL 700 .... ... Date : 09/03/86
--- Date :09/02/86
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Radial Locations (r/R, x/R = 5)
Figure 4-10. Reproducibility of Data for a Propene Flame.
SOOT MORPHOLOGY AND CHEMICAL ANALYSIS OF PAHs ADSORBED ON THE SOOT PARTICLE SURFACE
A morphological analysis of different soot samples
collected from different fuels was attempted using a Scanning Electron Microscope (SEM). The soot samples were collected on 47-jim Nuclepore membrane filters with 0.2-jim pore size. The soot particles collected consisted of agglomerate whith chain-like structure with diameters up to 1 pim and small individual spherules with diameters less than 0.1 lim (Figure 5-1). The size and the shape of the soot aggregates collected on the filters are consistent
with what it is reported in the litterature; however, the collection time tc on the filter was found to affect the shape of the soot particles collected. A long collection time led to a greater extent of aggregation of the soot particles on the Nuclepore filters, distorting the
information concerning the real size and shape of the soot
particle aggregate inside the flame.
Figure 5-1 shows photographs of two soot samples
collected from the same acetylene flame but at two different sampling times. The degree of soot aggregation was found much greater in the sample which gathered during
the longer collection time. Therefore, smaller collection 77
(a) t = 3s
(b) t = 5s
Figure 5-1. S.E.M. Photographs of Two Soot
Samples Obtained from the Same
Acetylene Flame but at Two Different Collection Times.
collection times (approximatively 3 seconds) were used for soot morphology analysis.
Soot samples from four flames operating with four different fuels (propane, 1-butene, isobutylene, and acetylene) were characterized by electron microscopy with a total magnification of 40,000 and 48,000. Photographs from each sample are presented in Figure 5-2. As expected, the soot aggregate size and number density for the propane flame were observed to be much smaller than those found in other flames. At equivalent collection times, the soot aggregates from the acetylene flame exhibit a more elongated chain-like structure than the others. Morphology
of the soot collected from the different flames suggests that the aggregates were formed from a multitude of spherules.
Chemical Analysis of PAHS Collected on the Soot Surface
Polycyclic aromatic hydrocarbons (PA~s) formed as products of incomplete combustion of organic fuels are a potential health hazard because of their toxicity. An understanding of the chemical pathways leading to formation
of soot, which is composed of these combustion products, would be of considerable value in understanding the
phenomena leading to soot formation. Characterizing the
chemical mechanism of soot formation is an active subject of research in its own right (chapter 2), and this report will present some observations on the chemical composition
Figure 5-2. S.E.M. Photographs of Soot Samples Obtained
from (a) Propane; (b) 1-Butene; (c) Isobutylene; (d) Acetylene.
of PAHs extracted from soot samples, without speculations about the mechanistic processes involved in their formation.
Samples of soot were obtained from two locations: on the face of the burner, and on a filter (Figure 3-4) located in the exhaust system, about six feet downstream of the burner. The filter used is one half of a Schleicher and Schull binder-free quartz micro-fiber 8x10 inch filter. The filters were conditioned in an oven at 400 0C for 4 hours, cooled, weighed and stored in a dessicator prior to use. Three runs were conducted at air/fuel ratio of 800,
in which pure fuels (Propane, propene, and 1-butene, respectively) were burned. At the end of each run, soot samples were collected from the burner face and the filter.
The samples were stored at 00C until they were extracted for 24 hours with methylene chloride in a Soxhlet
apparatus (Whatman 33 x 80 mm single thickness cellulose thimbles with a light glass wool plug at the top). The glass wool and thimbles were periodically pre-extracted whith the methylene chloride, dried at 500C and 3 torr, and tared. The solvant volume for extraction was approximatively 250 ml with a cycle time for extraction of
about 6 minutes. After extraction, the soluble materials were concentrated to 5.0 ml in a Kuderna--Danish apparatus, at 980C in a water bath. Gas chromatographic analyses were carried out using a Perkin--Elmer Sigma 2000 instrument fitted with a J and W Scientific 30 m, 250 m open tubular column with walls of which was bounded a 0.1- m film of
DB-5. Eluting components were "visualized" using a flame ionization detector (FID). The injection mode was 30:1
split ratio and the linear velocity of helium through the column was 35 cm/sec at 1000C. The injection port and the
detector were maintained at 3000C. The oven temperature was initially set at 500C and was gradually increased to 1250C at a rate of 50C/min and then increased to 3000C at 80C/min, with a 10 minutes final temperature hold. The sample volumes injected were 1 or 2 1.
The analyses of the extracts from soot particles amounts to a study of complex mixtures. Identification of some of the components of the mixture were made on the basis of retention time. The retention times were obtained
for standards (pure compounds) which were analyzed using the same capillary column and operating conditions used for exhaust emission analyses.
Some of the PAHS identified by retention times in this experiment are presented in table 5-1 for all the soot samples collected .
Chromatograms of the products extracted from different
soot samples collected on the filter are presented in Figure 5-3. The chromatograms for the extractable
products were qualitatively identical for all three flames studied, suggesting that, under the conditions of this study, the PAHs collected on the filter reached their final state which is independent of the parent fuel burned.
0 6.25 12.5 18.75 25.0 31.25 37.5 43.75
Figure 5-3. Chromatograms of Chemical Products Extracted
from Soot Sample Collected on a Filter:
(a) Propane Flame; (b) Propene Flame;
(c) Isobutene Flame.
Representative chromatograms of the extractable volatile present in soot collected at the face of the burner (Figure 5-4) show significant qualitative
differences in composition. These products were trapped before reaching chemical equilibrium, because the face of the burner provides a mechanism for premature termination of the polymerization reactions of intermediates present during the various steps of the combustion process. The differences in the distribution of intermediate products suggests that the polymerization process shows some
specifity to the parent fuel during the early stages; however, the similarity in the compositions of products
isolated from the exhaust stream suggests that these different chemical pathways ultimately converge to a common distribution of equilibrium products that is independent of the fuel structure.
The above are consistent with the operation of multiple, convergent pathways rather than one chemical mechanism leading to the formation of high molecular weight PAHs and soot and asserts the complexity involved in exact modeling of soot formation processes.
0 6.25 12.5 18.75 25.0 31.25 37.5
Figure 5-4. Chromatograms of Chemical Products Extruded
from Soot Samples Collected on the Face of
the Burner: (a) Propane Flame; (b) Propene
Flame; (c) Isobutene Flame.
Identification by retention times of some
polycyclic aromatic hydrocarbons
in combustor soot samples
Sample Retention Time Area Identification
(Propene) 30.87 1532 Benzo(b)fluorene
Filter 32.87 7504 Triphenylene
34.03 1113 Benzo(a)anthracene
35.98 21414 Benzo(e)pyrene
36.15 3372 Benzo(a)pyrene
38.73 17738 Benzo(ghi)pyrene
42.66 4521 Coronene
(Peaks present: 31; Peaks identified: 7)
propenee) 12.73 6148 Napthalene
Scraping 20.39 39104 1-5 Dimethylnapthalene
30.87 3696 Benzo(b)fluorene
32.81 82636 Triphenylene
34.07 2319 Benzo(a)anthracene
35.41 15870 Benzo(k)fluoranthene
35.92 15742 Benzo(e)pyrene
36.10 2047 Benzo(a)pyrene
42.89 35700 Coronene
(Peaks present: 33; Peaks identified:9)
(Propane) 12.82 393314 Naphtalene
scraping 16.60 27449 Azulene
30.89 5756 Benzo(b)fluorene
32.86 301390 Triphenylene 34.05 9155 Benzo(a)anthracene
35.37 52578 Benzo(b)fluoranthene
36.07 108229 Benzo(a)pyrene 38.34 85756 Indeno(l,2,3,cd)pyrene
(Peaks present: 58; Peaks identified: 8)
TABLE 5 (Continued)
Sample Retention Time Area Identification
(Propane) 20.33 1337 1,5 Dimethylnapthalene
Filter 29.92 1079 Pyrene
30.89 15140 Benzo (b) fluorene
32.84 18989 Triphenylene
33.09 3473 Chrysene
34.01 4269 Benzo(a)anthracene
35.94 37799 Benzo(k)fluoranthene
36.08 92952 Benzo(a)pyrene
38.32 45467 Indeno(1,2,3,cd)pyrene
42.78 23262 Coronene
(Peaks present: 40; Peaks identified: 10)
(1-Butene) 12.67 78230 Napthalene
scraping 16.07 10661 Azulene
20.36 194765 1,5 Dimethylnapthalene 29.38 30394 Fluoranthene
30.84 2852 Benzo(b)fluorene
32.84 14168 Triphenylene
34.06 3178 Benzo(a)anthracene
35.92 12416 Benzo(e)pyrene
36.10 3476 Benzo(a)pyrene
38.80 25319 Indeno(1,2,3,cd)pyrene
(Peaks present: 28; peaks identified: 10)
(Butene) 26.30 1730 Anthracene
Filter 29.92 1213 Pyrene
30.86 7616 Benzo(b)fluorene
32.82 13651 Triphenylene
35.34 22683 Benzo(b)fluoranthene
35.99 616516 Benzo(e)pyrene 36.13 10107 Benzo(a)pyrene
38.33 1359 Indeno(1,2,3,cd)pyrene
42.90 59110 Coronene
(Peaks present: 37; Peaks identified: 9)
EXPERIMENTAL RESULTS AND DISCUSSION Introduction
A series of experiments were implemented to study the key physico-chemical parameters that control the formation, growth, and oxidation processes of soot particles in a well-defined combustor configuration. It has been argued that formation and evolution of soot particles in flames is
controlled by competition between the rate of pyrolysis of the parent fuel and the rate of oxidation of the soot precursors and the formed soot particles (Milliken, 1962). The rates of pyrolysis of fuel and the oxidation of the soot precursors and soot particles are temperaturedependent. This requires that they be considered functions of interrelated parameters such as fuel structure, residence time, rate of mixing of the fuel with the
potential oxidants, and fuel heat content, which, in turn are determined by the operating conditions. The separate effects of these parameters on the fate of a soot particle was examined using a selective set of independent parameters. The controlling parameters that were
investigated in this study are: fuel type, fuel flow rate, fuel heat content, air/fuel ratio, and air and fuel jet velocities.
This chapter is divided into four parts. The first part deals with the operating conditions and the reasoning behind their selection. The second part is a succinct summary of a companion project performed by Roe (1987) on flow field mapping of a similar combustor. In the third part, the experimental results for temperature distributions are presented. The final part contains the experimental results and a discussion of soot formation and oxidation processes in the present study.
Soot emissions are intimately related to the heat
transfer, aerodynamics, and chemical processes that occur in flames. Although these processes are interrelated, they were investigated separately because simultaneous
measurements in flames are very difficult to achieve. Different operating conditions were implemented to investigate the main parameters that affect the temperature
distribution, the particle velocity distribution, and the chemical mechanisms that affect soot production in flames.
The Integrated Particle Sizing System (IPSS) and the
Laser Doppler Velocimeter (LDV) (Roe, 1987) were used to characterize the soot particle and velocity fields, respectively. Fine wire thermocouples were used to measure
the temperature distribution in flames. The data inside the flames were taken at different locations; locations are
expressed as the dimensionless ratio of displacement from
centerline to the radius R of the bluff body of the burner. The radial profiles of soot yield were taken for one axial location (x/R = 3) outside the flames.
The experimental work involving measurement of soot particle size and flux rates, temperature, and velocity
distributions was divided into four sets of experiments. First, measurements were performed for a number of flames in which the fuel flow rate and fuel species were varied (Table 6-1). Second, a set of measurements was obtained for flames in which nitrogen was added to the fuel to maintain a constant fuel jet velocity (Table 6-2). The latter was designed to show the effect of the fuel flow rate on soot formation in flames operating with similar flow conditions. Third, a set of experiments (Table 6-3) was designed to demonstrate the temperature effect on soot
formation. Nitrogen was added in graduated proportions to different fuel species to maintain a constant total heat content of the fuel stream while varying the fuel jet velocity. The final set of experiments was designed to
investigate the effect of nitrogen dilution on soot field (Table 6-4). Each set of experiments was investigated for three fuels (propene, 1-butene, and isobutylene).
Flowfield mapping of different flames was undertaken by Roe (1987) using a combustor similar to the one used in this investigation. Velocity mapping was used to characterize flow fields behind a bluff-body burner for
Flow Conditions for Experiment No. I (Effect of Fuel Flow Rate)
Experiments Fuel Flow rate Velocity Air/Fuel
No. (cm3/s) (m/s) Ratio
Ii Propene 20.42 13.9 1200
12 Propene 24.51 16.6 1000
I3 Propene 30.63 20.8 800
14 Propene 40.85 27.7 600
15 Propene 61.27 41.6 400
I6 1-Butene 15.28 10.4 1200
17 1-Butene 18.34 12.4 1000
18 1-Butene 22.92 15.6 800
19 1-Butene 30.66 20.8 600
110 1-Butene 45.84 31.1 400
Ill Isobutylene 15.85 10.8 1200
112 Isobutylene 19.02 12.9 1000
113 Isobutylene 23.77 16.1 800
114 Isobutylene 31.70 21.5 600
115 Isobutylene 47.54 32.3 400
1The air reference velocity for this set of experiments is
maintained constant at a value of 14 m/s.
Flow conditions for Experiment No. II (Constant Fuel Jet Velocity)
Fuel Nitrogen Nitrogen/Fuel
Experiment Fuel Flow Rate Flow Rate Velocity Ratio
No. (cm3/min) (cm3/min) (m/s) %
Ill Propene 55.19 9.20 49.9 10
II2 Propene 49.05 18.40 49.9 20
II3 Propene 42.92 27.59 49.9 30
II4 Propene 36.79 36.79 49.9 40
II5 Propene 30.65 46.01 49.9 50
II6 1-Butene 41.89 9.28 42.1 10
II7 1-Butene 37.23 18.67 42.1 20
II8 1-Butene 32.58 28.01 42.1 30
II9 1-Butene 27.92 37.28 42.1 40
Ill10 1-Butene 23.27 46.62 42.1 50
Il14 Isobutylene 43.47 9.28 43.2 10
Il15 Isobutylene 38.64 18.64 43.2 20
Il16 Isobutylene 33.81 27.97 43.2 30
Il17 Isobutylene 28.98 37.26 43.2 40
Il18 Isobutylene 24.15 46.60 43.2 50
Flow Conditions for Experiment No. III ( Constant Heat Release)
Fuel Nitrogen Nitrogen/Fuel
Experiments Fuel Flow Rate Flow Rate Velocity
No. (cm3/s) (cm3/s) (m/s)
IIll Propene 42.30 0.00 28.7
III2 Propene 42.30 18.40 44.4
III3 Propene 42.30 23.58 47.2
III4 Propene 42.30 28.77 50.0
III5 Propene 42.30 37.73 54.9
III6 Propene 42.30 45.28 59.0
III7 1-Butene 30.66 0.00 20.8
III8 1-Butene 30.66 17.92 35.8
III9 1-Butene 30.66 23.11 38.2
IIIl10 1-Butene 30.66 28.30 40.5
IIIll 1-Butene 30.66 39.15 45.4
IIl12 1-Butene 30.66 46.70 48.8
IIl13 Isobutylene 32.21 0.00 21.9
IIl14 Isobutylene 32.21 19.81 38.0
IIl15 Isobutylene 32.21 25.00 40.4
IIl16 Isobutylene 32.21 30.19 42.8
IIl17 Isobutylene 32.21 39.15 46.9
IIl18 Isobutylene 32.21 46.70 50.4
1 The air velocity and the the heat release rate were
maintained equal to 14 m/s and 196 Btu/min, respectively.