600A99056
Fine Particle Emissions from Residual Fuel Oil Combustion:
Characterization and Mechanisms of Formation
William P. Linak*, C. Andrew Miller
Air Pollution Prevention and Control Division, MD-65
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711 USA
Jost O.L. Wendt
Department of Chemical and Environmental Engineering
University of Arizona
Tucson, AZ 85721 USA
Prepared for presentation at
5lh International Conference on
Technologies and Combustion for a Clean Environment
July 12-15,1999
Lisbon, Portugal
••"corresponding author
tel: (919) 541-5792
fax:(919)541-0554
e-mail: linak.bill@epa.gov
-------
ABSTRACT
The characteristics of particulate matter (PM) emitted from residual fuel oil
combustion in two different types of combustion equipment were compared. A small
commercial 732 kW,rated fire-tube boiler yielded a weakly bimodal particulate size
distribution (PSD) with over 99% of the mass contained in a broad coarse mode, and only
a small fraction of the mass in an accumulation mode consistent with ash vaporization.
Bulk samples cotlected and classified by a cyclone indicate that 30 to 40% of the total
particulate emissions were less than approximately ~.5 pm aerodynamic diameter (PM2,s).
The coarse mode PM was rich in char, indicating relatively poor carbon burnout,
although calculated combustion efficiencies exceeded 99%. This characteristic behavior
is typical of this type of small boiler. Larger scale utility units firing residual oil were
simulated using an 82 kW laboratory refractory-lined furnace. PM emissions from this
unit were in good agreement with published data including published emission factors.
These data indicated that the refractory-lined combustor produced lower total but greater
fine particulate emissions, as evident from a single unimodal PSD centered
approximately around 0.1 pm diameter. Bulk cyclone segregated samples indicated that
all the PM were smaller than PM2,s, and loss on ignition (LOI) measurements suggested
almost complete char burnout. These results have particular significance in considering
the effects of fuel oil combustion equipment type on the characteristic attributes of the
fine PM emitted into the atmosphere, and their ensuing health effects.
In both experiments, PSDs were determined by electrical mobility for the
submicron range and by laser light scattering for the larger particles, with a small region
of overlap. The PSDs, size segregated particle analyses, and observed particle
morphology of both sarr:ples provided insight into mechanisms governing the formation
of fine particles during residual oil combustion. Metals do not vaporize until the char
burnout regime of residual oil combustion. Once liberated, these trace elements form
new particles via homogeneous nucleation followed by particle growth by way of
coagulation and heterogeneous condensation processes. Differences in the PSDs are
explained by differences in the heat extraction rates and resulting flame quenching typical
of the two combustion systems. The large coarse mode and high LOI values determined
for the small fire-tube boiler suggest incomplete char oxidation due to flame quenching
and only partial vaporization of the inherent trace elements. The large accumulation
mode and nearly zero LOI values for the refractory-lined combustor suggest almost
complete char oxidation and vaporization of the inherent trace element species. These
conclusions are supported by thermodynamic and aerosol growth models.
2
-------
INTRODUCTION
. "
This paper comprises the first of a two-part~ommunication describing results o.
collaborative research between combustion engineers and inhalation toxicologists. The
objective is to try to establish links between the characteristics of size segregated
speciation of fine particulate maller (PM) in combustion exhausts and mechanisms
through which fine PM causes respiratory distress. The work reported here is concerned
with the engineering portion of the work, and focuses on both characterizatiun and
mechanisms of fonnation of ash aerosol emissions from the combustion of residual fuel
oil.
In a previous study, Miller et al. (1998) explored the relationship between residual
fuel type and composition and the physical and chemical characteristics of PM produced.
Four different residual fucl oils, including two grades and three sulfur contents, were
burned in the same in-house 732 kW (2.5x 106 Btulhr) rated fire-tube boiler. In addition
to tests examining four different residual fuel oils, one oil was burned under conditions
designed to simulate off-optimum boiler operation. In contrast to this previous work, the
current paper explores, for a single residual fuel oil, the mechanistic relationships
, .
between combustion configuration and the physical and chemical characteristics of PM
produced. Whereas the previously reported work examined the effects of fuel oil
compositions, this paper describes the effects of different combustion equipment types.
The results of both studies are used as the basis for ultimately linking measures of acute
pulmonary damage to engineering variables.
Fine particles in the air have been of considerable environmental interest in recent
years because of a number of research studies correlating short term exposure of ambient
levels of fine PM with acute adverse health effects (Wilson and Spengler, 1996). These
studies were summarized by the EPA (U.S. EPA, 1996a; Bachmann et aI., 1996) and
reviewed by EPA's Clean Air Scientific Advisory Committee (CASAC) which concluded
there was evidence linking ambient fine PM concentrations and adverse health effects
(Wolff, 1996). These studies were the basis for a revision of the National Ambient Air
3
-------
Quality Standards (NAAQS) for PM that included a standard for PM less than 2.5 J.lffi in
diameter (PM2.s) (Federal Register, 1997).
In the ambient atmosphere, ultrafine PM, which includes a portion of PM2.5, is
composed primarily of sulfates, nitrates, condensed organics, carbonaceous soot, and
inorganic aerosols formed during high temperature processes such as the ~ombu.stion of
fuels containing trace quantities of metals and other impurities (U.S EP A, 1996a; Linak
and Wendt, 1993, 1994). Formation of these small particles is heavily influenced by
condensation and other gas-to-particle conversion processes. In contrast, the coarse
fraction of PM tends' to be composed of particles formed by mechanical (e.g.,
fragmentation, grinding, crushing, and entrainment) processes. Because they are formed
by different mechanisms, the fine and coarse fractions of PM are likely to have
significantly different compositions. Particle composition has been identified as one of
the possible factors driving the adverse health effects associated with exposure to amb:ent
PM (National Research Council, 1998).
Health effect researchers have identified at least two aspects of particle
composition that appear to exacerbate health damage from particles. The first is related
to water-soluble transition metals such as copper (Cu), iron (Fe), vanadium (V), nickel
(Ni), and zinc (Zn) present in the particles (Dreher et al., 1996a, 1996b, 1997). The
second aspect of particle composition that might be important as far as health damage is
concerned is aerosol acidity in general. In addition to these composition related facets,
ultrafine particles (those particles less than 0.1 11m in diameter), regardless of
composition, have been identified as potential factors influencing mechanisms for these
health impacts (U.S. EPA, 1996a). Particles with all of these characteristics (transition
metals, acidity, and ultrafine size) are contained in the PM generated from the
combustion of residual fuel oils (e.g., No.5 and 6 fuel oils). Hence, one might
hypothesize residual fuel oil combustion to be suspect, as far as emission of toxic fine
particles is concerned.
4
-------
Unfortunately, the hypothesis that residual oil combustion is the prime .~ource of
fine particles causing respiratory distress is not consistent with epidemiological data.
Residual fuel oils are used in significant quantities in only selected regions of the
country. Discounting sales of Bunker C oil, the majority of which is likely to be burned
by ships well away from continental coastlines, Figure 1 shows significant residual oil
usage in the Northeast and Southeast regions. There is little residual oil usage in the
Midwest or along the Pacific coast. Additionally, the 13 other Great Plains and Rocky
Mountain states combined (not included in Figure 1) account for less than 1 % of the
residual fuel oil sale!; in the U.S. However, adverse health effects associated with
exposure to fine PM are not limited to the Northeast and Southeast regions of the U.S.
(Dockery et at., 1993). This fact suggests that sources of fine PM other than (or in
addition to) those related to residual fuel oil combustion must also be important.
EXPERIMENTAL
Particle characteristics and emissions are compared from two types of combustion
systems. These can be considered to represent extremes of a range of the practical
conditions under which fuel oil is burned. Although they may not represent specific
boiters in all respects, they are investigated here with a view to determining how this
range of combustion condit:':ms influence the cha,racteristics of fine particles, and the
mechanisms which form them. The first system is a sman fire-tube boiler, in which
combustion occurs in tubes surrounded by water or steam. These types of small boilers
have large heat transfer surfaces, small combustor volumes, relatively small gas residence
times, cold walls, and high gas quenching rates, and often produce emissions with
relatively high carbon contents due to unburned carbonaceous oil residues or char. The
second system is a refractory-lined research combustor designed to simulate the
time/temperature environments of larger. utility boilers and incinerators. In large utility
boilers the water to be boiled, rather than the combustion gases, is contained in tubes.
These systems, including the refractory-lined combustor, operate at higher temperatures
with lower gas quenching rates. As will be discussed later, particle emissions from this
5
-------
system contain very liule unburned carbon and more closely approximate emissions from
large oil-fired utility boilers reported in the literature.
Fire-tuhe Boiler
Experiments were performed using EPA's North American, three-p~s, ~re-tube
package boiler (NAPB), which is a practical, commercially available residual fuel oil
combustion unit (Figure 2). The burner was a North American model6121-2.5H6-A65,
rated at 732 kW (2.5x106 Btulhr) and has both a natural gas ring and an air-atomizing
center nozzle capable of firing No.2 through No.6 fuel oils. The main fire-tube has a
diameter of 46 cm (18 in), and each of the 24 second-pass tubes and 20 third-pass tubes
has an inside diameter of 6.4 cm (2.5 in). The boiler has 27.9 m2 (300 ft2) of heat transfer
surface, and can generate up to 1090 kglhr (2400 Ihlhr) of saturated steam at gauge'
pressures up to 103.4 kPa (15 psig). Heat is extracted from the steam through a heat
exchanger and rejected to an industrial cooling water system that provides a simulated
boiler load. Oil temperature can be controlled using an in-line electric heater to maintain
consistent fuel viscosity. Oil and atomizing air pressures are independently controlled to
ensure proper oil atomization. The NAPB is equipped with several sampling ports located
at the exit of the boiler (see Figure 2). Temperatures at these sampling ports ranged from
450 to 550 K (350 to 530 OF).
Refractory-lined Comhustor
Experi~ents were also performed using the laboratory horizontal tunnel
combustor also presented in Figure 2. Although this system has a maximum rating of 82
kW (280,000 Btufhr), these experiments were conducted at 59 kW (200,000 Btufhr).
This refractory-lined research combustor was designed to simulate the time/temperature
and mixing characteristics of practical gas- and oil-fired combustion systems. Fuel oil
and combustion air were introduced into the burner section through an International
Flame Research Foundation (IFRF) moveable-block variable-air swirl burner. This
burner incorporates a pressure atomizing oil nozzle positioned along its center axis.
6
-------
Swirling air pass~s through the annulus around the fuel injector, promoting flame
stability and attachment to the refractory quarl. A high{i.\Virl (IFRF Type 2) flame with
internal recirculation (swirl No.=1.48) was used. Gas and aerosol samples were taken
from a stack location 5.9 m from the burner quarl. The temperature at this loca~on was
approximately 670 K (745 oF).
Measurements
Both the fire-tube boiler and refractory-lined combustor are fuUy instrumented,
with continuous emission monitors (CEMs) for carbon dioxide (C02), carbon monoxide
(CO), nitrogen oxides (NOx), oxygen (02)' and sulfur dioxide (S02)' A data acquisition
system was used to continuously record CEM measurements as well as boiler steam and
flue gas temperatures. The flue gases from both units pass through a manifold to a
facility air poIlution control system (APCS). Induced drafts (necessary for the APCS
operation) were imposed on both systems. All experiments were performed at a
stoichiometric ratio (SR) of 1.2 (20% excess air). No air preheat was employed.
Particulate Sampling and Analysis
PM emissions were sampled using a variety o~ methods. Standard EP A Method 5
and Method 60 sampling and analytical procedures were used to determine total
particulate and metal concentrations (Garg, 1990; EPA, 1994a, 1994b). Metal analyse~
were performed using acid digestion and inductively coupled plasma (ICP) mass
spectrometry. Other metal analyses were determined by x-ray florescence spectroscopy
(XRF). Additional samples were analyzed by x-ray adsorption fine structure (XAFS)
spectroscopy. As described by Galbreath et al. (1998) and Huggins and Huffman (1999),
XAFS spectroscopy is a non-destructive element-specific structural probe that is useful
for determining trace element speciation and forms of occurrence in chemically and
structurally complex materials like combustion ash. XAFS spectroscopy is Intended to
compliment other techniques (such as rcp and XRF spectroscopy) and was used here to
help determine sulfur speciation within the collected PM.
7
-------
PSDs were detcnTIined by three techniques: electrical mobility and inertial
impaction for sampled aerosols and light scattering for in-situ in-stack analyses.
Extractive samples were taken for collection by inertial impaction and electrical mobility
analyses using an isokineticaerosol sampling system based on the modified designs of.
Scotto et at. (1992) and Linak et a1. (1994). In order to minimize in-probe gas and
aerosol kinetics, the sampling system dilutes and cools the aerosol sample using filtered
nitrogen immediately after sampling. Dilution ratios, here approximately 5: I, are
determined for each experiment and can be verified independently by the measurement of
a nitric oxide tracer gas. Diluted samples were directed to a Thermo-Systems Inc.
scanning mobility particle sizer (SMPS). The SMPS.classifies and counts particles
within a working range of 0.01 to 1.0 ~m diameter using principles of charged particle
mobility through an electric fie1d. The SMPS was configured to yield 54 channels evenly
spaced (logarithmically) over a 0.01 to 1.0 ~m diameter range.
Extracted samples during the tire-tube boiler experiments were also directed to an
Andersen Inc. eight-stage, 25 Umin. (0.9 felmin), atmospheric pressure in-stack cascade
impactor. This instrument is designed to collect samples on nine stages (including the
afterfilter) less than approximately 10 ~m in diameter for subsequent gravimetric and
chemical analysis. The in-stack cascade impactor was operated using a modified
California Air Resources Board (CARB) Method 501 procedure (CARB Method SOl,
1990). Procedure modifications are discussed in detail elsewhere (~i11er et at., 1998).
During the refractory-lined combustor experiments, a MSP Inc. 10-stage, 30 Umin (1.06
felmin) Micro-Orifice UnifonTI Deposit Impactor (MOUDI) was used. This atmospheric
pressure impactor is designed for enhanced small particle resolution.
In addition to the clectrical mobility and inertial impaction devices, which require
an extracted sample, in-situ light scattering PSDs were taken usin~ an Insitec Inc. particle
counter sizer velocimeter. Although this instrument is designed to measure both particle
size and velocity, the velocity feature was not used here. This instrument determines
particle size by measuring the light scattering intensity of particles which pass through a
sampling volume established within the combustor stack by a laser focused through a set
8
-------
of quartz optical access portS. The working range of this device was approximately 0.3 to
100 flm diameter which slightly overlapped and extended' the PSD data collected by the
SMPS.
SampleS were also collected on silver membrane filters and analyzed using a SEM
equipped wtth nn energy dispersive x-ray (EDX) spectrometer. This provided
morphologica:t information of individual particles. Particles were extracted.from the
stack location using the same sampling system and dilution as used by the SMPS
described above. However, these particles were directed through a stainless steel filler
holder containing a 47 mm silver membrane filler. Sampling times of approximately 30-
60 s provided a sufficient quantity of particles for analysis. Silver filters were used to
improve conductivity and minimize particle charging caused by the electron beam.
It was necessary to collect larger quantities of size segregated PM for the
toxicological studies (described in a companion paper) and XAFS analyses. These were
collected using the large dilution sampling system (see Figure 3), which is capable of
sampling 0.28 m3/min (10 fe/min) of flue gas. Isokinetic sampling conditions were not
possible for these large dilution samples. The sample passes through a modified Source
Assessment Sampling System (SASS) cyclone and is then diluted with clean ambient air
[2.8 m3/min (100 ft3/min)] by means of a perforated con~ assembly. Rapid unifonn
dilution within an approximate 3 s residence time mixing chamber cools the sampled
gases and PM to approximately ambient temperature. The resulting PM is collected on
large (64.8 cm, 25.5 in) Teflon coated glass fiber filters, transferred to sampling jars, and
made available for subsequent chemical, physical, or biological analysis. This sampling
system was originally designed and constructed to sample particulate and condensable
emissions from municipal and hazardous waste incineration facilities. Further details
regarding the dilution sampler's construction and operation are presented by Steele et al.
(1988). The SASS cyclone preseparator produces 50 and 95% particle collection
efficiencies at approximately 1.8 and 2.5 flm aerodynamic diameter, respectively, at
standard conditions.
9
-------
Experimental Approach and Conditions
The current study investigated the effects of combustion configuration by
comparing the characteristics of the fire-tube boiler PM reported previously (Miller et at.,
1998), with the PM characteristics of the same fuel burned in the refractory-lined
combustor. The high sulfur No.6 oil.used in both cases contained 2.33% sulfur.
Operational characteristics for.both systems included similar oil temperatures (380 to 400
K, 230 to 250 oF) and stoichiomctries (SR=1.2), corresponding to stack O2 concentrations
of approximately 3.5% (dry).
Figure 4 presents an oil droplet PSD for the high sulfur No.6 fuel oil using the
Delavan Airo Combustion air atomizing type oil nozzle (model 30615-84) from the fire-
tube boiler. Fuel oil and atomizing air were maintained at approximately equal pressures
between 200 and 240 kPa (29 and 35 psi g) during boiler operation. Figure 4 indicates
that this atomizer produces oil droplets with a relatively narrow PSD and a mean
diameter of between 30 and 40 ~m.
The refractory-lined combustor experiments used a similar air atomizing oil
nozzle manufactured by Spraying Systems Co. (model Air Atom 1/4-JS5). However, in
order to be incorporated within the smallcr IFRF burner, a stock atomizer required
extensive ext~rnal modifications. Critical internal dimensions were unaffected.
Although no oil droplet PSD is available for the refractory-lined combustor experiments,
the PSDs are believed to be similar to those for the boiler studies, and differences in
carbon burnout may be attributed to differences in temperature history rather than in
droplet size.
RESULTS AND DISCUSSION
EQuilihrium Predictions
10
-------
p
Multicomponent equilibrium calculations can provide insight into which species
are thermodynamical1y stable at flame and flue gas temper.£Hures. Previous literature on
equilibrium predictions exist in the work of-Wu and Biswas (1993), Frandsen et n1.
(1994), and Owens et a!. (1995), although there are relatively few results wiLh the
inorganic ash concentration levels pertinent to residual oil combustion. As always, the
accuracy of ~qui1ibrium results depends on the accuracy of the thermodynamic data
available, and on the avai1ahility and inc1usion of thermodynamic data for al1 important
species containing the elements in question. In addition, concentration predictions based
on equilibrium calculations may not be realized in practical systems, because they do not
take into account kinetic or mixing limitations. Thermochemical predictions were
determined using the CEA computer code for calculating comp1ex chemical equilibrium
compositions (McBride et a!., 1993). Tahle 1 lists the metal species considered in these
calculations, together with the appropriate references. Of interest is the thermodynamic
partitioning between vapor and condensed phases, as wen as the partitioning between
various species. Input conditions, summarized in Table 2, represent combustion of the
high sulfur No.6 oil at a stoichiometric ratio of 1.2. The fuel ultimate analysis and trace
element concentrations were taken from previously published results (MilIer et a1., 1998).
However, only trace elements with concentrations greater lhan 1 ~g/g of oil were
considered in the calculations.
Figure 5 presents the results of these equilihrium calculations for seven trace
elements [V, Ni, Zn, Fe, CU,lead (Pb), and chromium (Cr)]. Mass fractions of the
e1ements (as indicated species) are plotted against temperature. Based on the species and
thermodynamic data listed in Tahle I, the elemental dew points are calculated to be
approximately 1800 K for V, 1700 K for Ni, 1200 K for Zn, 1700 K for Fe, 1400 K for
Cu, 1100 K for Pb, and 900 K for Cr. These predictions indicate that, at the
concentrations present in fuel oil, all seven elements might vap0rize within the
combustion environment, depending on the local temperatures to which the ash
constituents are exposed. The dewpoints are lower than those calculated for pulverized
coal and incineration applicatior.s, because of the low ash and metal concentrations
(Linak and Wendt, 1994), and appear to indicate possible vaporization of Cr, for
11
-------
example, which is usually considered to be nonvolatile. In fact, Figure 5 indicates that
Cr~03(s) is not a predicted condensed species 1I.t intermediate temperarures. This is in
'contrast to similar calculations performed using higher Cr concentrations simulating an
incineration environment (Linak et at., 1996). These previous calculations examined a
system containing 100 ppm. Cr compared to the current system which contains
approximately 0.03 ppm,.. It should be emphasized that only the species in Table 1 are
considered in these calculations, and the presence of even ~ma11 quantities of other
elements may affect the predicted behavior. For example, even small amounts of
chlorine can affect metal speciation and volatility (Linak and Wendt; 1993, 1994). Also
evident from Figure 5 is that all seven elements are predicted to form sulfates at lower
temperatures (approximately 700 to 1100 K, 800 to 1520 OF). This preferred speciation
is likely to affect the solubility and perhaps the bio-avaiJabiJity and toxicity associated
with the trace elements. However, it is possible that kinetic limitations at these low
temperatures could limit the formalion of thermOtjynamically stable metal-sulfate species.
Particle Size Distributions
The upper panel of Figure 6 presents representative particle vc.,Ilume distributions
for the fire-tube boiler (open symbols) and refractory-lined combustor (shaded symbols)
experiments. The inset in the upper panel shows more delail in the ultrafine particle size
range. Together, these electrical mobility and light scattering measurements span 4
decades of particle diameters (0.01-100 Jlm). The fire-tube boiler PSD is the same data
as plotted in Figure 3 of Miller et a1. (1998). The boiler PSD,indicates that most of the
particle volume is associated with large (coarse mode) particles greater than 10 Jlm in
diameter. The open symbols in the inset show that even the fire-tube boiler produces a
small accumulation mode with a mean diameter of between 0.07 and 0.08 Jlm, but that
this accumulation mc/de is much smaller than that for the refractory-lined furnace (shaded
symbols). Thus, both configurations produced an uItrafine mode, but only the fire-tube
boiler produced a bimodal PSD with a very large and dominant coarse mode.
12
-------
As discussed by Miller et nt. (1998), this type of bimodal PSD is consistent with a
mechanism of metal vaporization/nucleation/coagulaLion/condensation and incomplete
~..
burnout of residual fuel cenospheres. In order to understand how these two modes arise,
it is necessary to review the various important controlling mechanisms. Williams (1976)
describes the sequence of events associated with residual.oil combustion. Initially, fuel
oil is atomized and sprayed though the burner with combustion air to produce an initial
droplet size distribution ranging from less than 10 to approximately 100 pm in diameter
(see Figure 4). As these fuel droplets are heated, the lighter organic components
vaporize, diffuse into the gas phase, and react (self-ignite) with the oxygen forming
diffusion flames around each droplet. As the process proceeds, fractional distillation,
liquid-phase cracking, and thermal decomposition add vapor-phase fuel species to the
flame front, causing considerable swelling of the remaining heavy tar droplet. As the
volatile components are depleted, the remaining heavy tars collapse to form a
carbonaceous residue (or char) with an open cell structure called a cenosphere. The
flame front recedes into the cenosphcre resulting in heterogeneous combustion of the
carbonaceous residue at a rate of about 10% that of the initial droplet These porous
particles, composed primarily of carbon and ash, may be quite large (comparab1e to or
even larger than the original fuel droplet) and have relatively low density. As oxygen
diffuses into the cenospheres, heterogeneous combustion results in high internal
temperatures promoting vaporization of the inherent miner~l matter, essentially all of
which is bound directly to the organic matrix; As char combustion conc1udes, these
cenospheres produce the large supermicron (coarse mode) PSD.
Figure 7 presents a scanning electron microscope (SEM) image of oil char
collected from the fire-tube boiler, and the sponge-like morphology that is apparent
clearly suggests swelling and pore formation. The size of this coarse mode will be
strongly dependent on the combustion environment and particularly dependent on
processes that influence flame quenching. Likewise, the extent of ash (meta1)
vaporization is also strongly dependent on the extent of carbon burnout. When
combustion is incomplete, a large portion of the metals remain trapped in the unburned
char particles, and never escape into the vapor phase. However, as the combustion gases
13
-------
cool, those metals which have vaporized will condense on existing surfaces or, if
supersatur, 1 In partial pressures are large enough, wilt nucleate to form new particles.
Condensation on existing submicron partic1es is preferred because these particles
typically dominate the available surface area. The distinctive submicron peak (between
0.07 and 0.08 ~m diameter) is indicative of particles formed by nucleation, coagulation,
and condensation of materials that have vaporized in the high temperature region of the
combustor. However, when large portions of the metal constituents fail to vaporize, as in
the fire-tube boiler, the accumulation mode will be very much smaller than when they do
vaporize, as in the refractory-lined combustor (see Figure 6).
In stark contrast to the fire-tube boiler volumePSD, Figure 6 indicates that the
refractory-lined combustor volume PSD ~e;haded symbols) is composed exclusively of a
narrow s'.:bmicron accumu1ati(m modc~ with a mean diameter of approximately 0.1 ~m.
The light scattering measurements indkated an almost complete absence of particles
greater than 1 ~m diameter. This is consistent with the lack of any dilution sampler
cyclone catch, the negligible LOI values, and the greenish-yellow color of the filter PM.
The volume PSD for the refractory-lined combustor is the result of nearly complete char
burnout and nearly complete vaporization of the trace elemente;. These species have
undergone nucleation, condensation, coagulation, and agglomeration processes to
produce the distinctive accumulation mode shown in Figure 6. Comparison between the
areas under the submicron volume PSDs for the two types of equipment suggests that
only a very small fraction « 1 %) of the ash is vaporized in the fire-tube boiler. In
contrast, well over 99% of the ash vaporizes in the refractory lined furnace.
Comparison of Calculated and Measured Mass Concentrations
Table 3 presents a comparison of calculated and measured mass concentrations
for the two experimental systems. It should be noted that these masses include both ash
and unburned carbon, when present. The first two columns presen.t typical total volume
concentrations for the boiler and combustor experiments as determined by the electrical
mobility and light scattering measurements, and are obtained by integration of the PSD
14
-------
data presented on the upper panel of Figure 6. Tobl volumes are detennined by the sum
of accumulation and coarse mode values. Mass concentrations, calculated from these
""-,
measured volume concentrations. and those measured by EPA Method 5, are presented in
the next two sets of columns. A density of 3.5 glcm3 was assumed for the accumulation
mode PM characterized by the electrical mobility measurements, and a density of 1.85
glcm3 was assumed for the coarse mode PM characterized by the light scattering
measurements. Note that the material densities used here are not the same as ash bulk
densities. These assumed densities, yielded 180 mg/ m3 for calculated mass
concentrations for the boiler and 93 mg/mJ for the combustor. Mass concentrations of
184 and 93 mg/mJ were independently measured for the boiler and combustor
experiments. respectively. The assumed density of 3.5 g/cmJ for the accumulation mode
PM was chosen as a reasonable value based on the hypothesis that these particles are
composed of condensed trace element vapor (sulfates and oxides). If one estimates that
the light scattering measurements did not include approximately 25% of the coarse mode
volume corresponding to PM larger than I 00 ~m diameter (see Figure 6), then the
assumed density would need be revised to approximately 1.4 g/cmJ (instead of 1.85
g/cmJ) to reconcile the calculated and measured mass concentrations. Both values are
reasonable based on densities reported for activated carbons and coal chars and based on
the extensive porous particle morphology evident from Figure 7.
Emission Factors
Note that the measured mass concentration of 93 mg/mJ determined from the
refractory-lined combustor experiments ean be converted into an emissions factor of
approximately 10.5 lb/lOJ gal (1.26 gIL). This value is comparable to the emission factor
of9.2lb/lOJ gal (1.10 g/L) for No. 6 residual oil-fired boilers larger than lOOx106 Btulhr
(29.3 MW) published in AP-42 (U.S. EPA. 1996b). This comparison lends Jurther
support to the supposition that the refractory-lined combustor adequately simulates the
combustion environment of larger industrial and utility boilers. Note that Table 3, Figure
6, and the lack of any dilution sampler cyclone catch indicate that the PM emissions from
this unit are essential1y all less than 2.5 j.Jm aerodynamic diameter. As reported by Miller
15
-------
et al. (1998), tho rnnge of emission factors determined for the fire-tube boiler was
approximately twice thnt of oil fired utility boilers. However, dilution samples for these
experiments indicate that only 30 to 50% of the PM mass emissions had an aerodynamic
diameter less than 2.5 J.lm. Hence, the fine PM emission factor for utility boilers may
well be greater than that of fire-tube boilers.
Emissions results from this study cnn be compared to values from the literature.
Goldstein and Siegmund (1976, 1977) examined the effect of fuel type and combustion
modifications on PM emissions from a small 37 kW (50 hp) fire-tube boiler. They report
similar PM emissions of approximately 180 mg/m3 with carbon contents of up to 80%
while burning a similar 2.2% sulfur No.6 fuel oil. They nlso noted that efforts to
increase PM burnout shift the PSD toward the submicron range. Conversely, Cheng et aI.
(1976) and Bacci et al. (1983) examined PM emissions from 30 MW (lxlOl Btulhr) and
320 MW (lxlO9 Btulhr) fuc1 oil fired power plants, respectively. PM emissions from
these units were reported to be 87 mg/m3 and 40 to 50 mg/m3, respectively, even though
the 30 MW unit was equipped with a multicyclone PM control system. Bacci et al. also
recount particle densities r.,nging from 1.4 to 1.9 glcm3 for particles between 3 and 5 J.lm
diameter, considerable enrichment of sulfur, V, and Ni on particles smaller than 0.5 J.lm
diameter, and. morphological and mineralogical data indicating the presence of
cenospheres and micron scale crystals composed of sodium vanadium oxide hydrate.
They go on to sta'e that, even at this large scale, increased combustion chamber
temperatures and improved oil atomization improve particle burnout and overall
reduction of the carbonaceous PM.
Elemental Composition and Particle Size
Table 4 presents a comparison of the trace element emissions for the fire-tube
boiler and refractory-lined combustor experiments. As expected, these concentrations are
similar due to the fact that both systems fired the same high sulfur No.6 fuel oil.
However, in contrast to the PM from the boiler which exhibited high values for loss on
ignition (LOI) ranging from 60 to 85%, blank-corrected results of filter samples from th~
16
-------
combustor tests indicate no mass lost on -ignition. The sum of the concentrations of the
seven analyzed elements listed in Table 4 for the combustor eXp'eriments account for 21.6
'- ,
mglm3 or approximately 23% of the total mass emissions. However, if these clements are
assumed to exist as sulfates, they then account for 67.1 mg/m3 or approximately 72% of
the total mass emissions. In fact, XAFS spectroscopy indicated that, while a large
portion (40-60%) of the sulfur measured in the fire-tube boiler PM existed as unoxidized
organic sulfur (predominantly thiophenic sulfur), essentially all (99%) of the particulate
bound sulfur in the refractory-lined combustor samples was in the form of sulfates.
Figure 8 prescnts elemental mass fractions vs. particle diameter for six metals and
sulfur from the refractory-lined combustor tests. Impactor substrates from replicate
experiments were analyzed by XRF and ICP spectroscopy for the six metals. Sulfur and
sulfate loadings were determined by XRF and ion chromatography (IC) analysis,
respectively. Figure 8 plots elemental mass fractions normalized by dlogDp vs. the
characteristic aerodynamic diameter for each impactor stage. The data indicate that, in
general, XRF and ICP analyses produce similar results, and that the emissions of these
elements are concentrated in particles approximately between 0.08 and 0.2 !lm diameter.
This COITcsponds to the accumulation mode determined by the electrical mobility
analyzer (see Figure 6). Figure 8 also indicates that sulfur and sulfates are present within
particles of the same size. It is interesting to note that Figur~ 8 also indicates the
presence of a small but distinct mode at approximately 1 Ilm diameter. This small mode
is noted in the replicate V, Ni, and Zn measurements determined by XRF and ICP
analysis, and possibly in the Cu XRF analysis. This mode is not evident for the more
volatile metal, Pb. These data are consistent with the conjecture that less volatile metals
are released very late during the heterogeneous combustion of the cenospheric carbon and
only as the neighboring carbon atoms are oxidized, while more volatile metals can more
easily vaporize from interior surfaces and diffuse to the gas-phase environment exterior
to the char particle. The 1 Ilm diameter particles noted in Figure 8 may be the result of
nonvolatile metal coalescence as the last vestiges of carbon are consumed. Note that the
cOITesponding volume PSD in Figure 6 does not indicate the presence of this small mode,
17
-------
although this may b~ th~ result of relatively poor sensitivity of the light scattering
measurements at the lower limit of its operational range.
Model Predictions
Experimental results were compared to model predictions using a multi-
component aerosol simulation code (MAEROS). The purpose of these calculations was
to illustrate that the resulting PSDs for both systems could be predicted using the same
modeling approach involving char particle burnout (or premature quenching) followed by
humogeneous nucleation and coagulation of a portion of th~ PM mass. A secondary
objective was to determine the extent to which differences in coagulation rates caused by
large differences in vaporized mass and resulting number densities might account for the
differences in the measured PSDs. The MAEROS code was developed by Gelbard and
Seinfe1d (1980) to simulate the dynamics of a spatially homogeneous aerosol. The
algorithms solve a discrete form of the general dynamic equation (ODE) which describes
the spatial and temporal evolution of a particle size distribution including convective and
diffusive terms. The ODE can include terms to describe: coagulation due to Brownian
motion, gravity, and turbulence; particle deposition due to gravitational settling,
diffusion, and thermophoresis; particle growth due to condensation of a gas; and time
varying sources of particles of different sizes and chemical compositions. MAEROS is
intended as a 'gel1P,ral tool to supply necessary algorithms to solve the ODE for aerosol
and other particulate systems. Figure 9 iIIustrates the predicted evolution of an aerosol
due to coagulation. The particle size domain (0.001 to 20.0 Ilm diameter) was divided
into 13 geometrically equal sections or bins. Coagulation was the only mechanism
considered. All other mechanisms (condensation, nucleation, deposition, etc.) were
disabled. Two cases con-esponding to the fire-tube boiler and refractory-lined combustor
conditions were examined.
To examine particle evolution within the fire-tube boiler, 99.8% of the 180 mg/m3
PM mass was assigned to the various sections so as to simulate the coarse mode mass
PSD illustrated in Figure 6. The remaining 0.2% of this mass was assigned to section 2
18
-------
(0.0021-0.0046 J.1m diameter) .to simulate the nucleation of n vapor fume. This
<.Ii~trib~tion between fine and coarse mode PM corresponds tQ tlJ.eratios detennined by
\. ,
the electrical mobility and tight scattering instruments (see Table 3). The initial number
concentration was 1.5xl016/m3, and it is important to note that, while the nucleating
vapor comprised only 0.2% of the PM mass, it comprised well over 99.99% of the PM
number. System pressure and temperature were maintained at 1.01x105 Pa (1 atm) and
810 K (1000 OF) to simulate post-flame conditions. Following from this initial
distribution, Figure 9 presents three mass distributions which follow the evolving aerosol
through three orders in time (t=O.1, 1.0, 10 s). Note that coagulation does not change the
total aerosol mass and that the areas under all three curves represent 180 mg/m3. Number
concentrations, however, are affected, At 0.1 s, the number concentration has fallen to
1.1xlO'6/m3, and the avemge nuclei size has grown slightly. At 1.0 and 10.0 s, the
distributions have grown somewhat further into the 0.01 to 0.1 J.1m diameter range. These
times represent a range of typical residence times within combustion systems. At 10 s,
number concentrations have fallen to 1.2x IOJ4/m3 and the nuclei are accumulating in a
mode with a mean diameter of approximately 0.02 pm (see Figure 9 insert). The
distribution of the coarse mode PM is relatively unaffected by these processes.
In contrast to the boiler simulation, 100% of the 90 mglm3 PM mass measured in
the refractory-lined furnace was assigned to section 2. Again, this initial distribution
corresponds to the partitioning seen in Table 3, and indicates complete oxidation of the
oil char and vaporization of the inorganic elements followed by homogeneous nucleation
of these elements to fonn 20 to 40 A particles. The initial number concentration is
2.7xlOlI/m3. In comparison to the boiler simulation, Figure 9 indicates that these aerosol
nuclei tend to coagulate much more quickly, due to the much larger number
concentration of nuclei particles. At 1 and 10 s, as number concentrations fall to
8.2xlOt4 and 7.6x101J, respectively, coagulation slows considerably causing the aerosol to
accumulate into a mode approximately between 0.07 and 0.2 pm diameter. Coagulation
processes predict that the accumulation mode produced from the refractory-lined
combustor would be larger in diameter compared to the accumulation mode produced
from the fire-tube boiler. This prediction is reflected in the measurements. The
19
-------
accumulation mode measured from the refractory-lined combustor matches the predicted
PSD (see Figure 6), while the predicted accumulation mode for the fire-tube boiler (0.02
Jlm diameter) slightly underpredicts the measured value (0.07 Jlm diameter).
Heterogeneous condensation of a portion of the vapor on existing nuclei is a competing
process which has been shqwn to increase particle growth rates (Davis et aI., 1998). Note
that the coagulation mechanism described in Figure 9 also does r.ot include the effect o~
differing fractal properties of any agglomerates that may be formed (Matsoukas and
Friedlander, 1991). It has been assumed here that only spheres result from the
coagulation process.
CONCLUSIONS
The effects of combustion configuration on fine particle emissions from residual
fuel oil combustion were examined. A laboratory scale refractory-lined combustor,
which was shown to simulate combustion conditions of a large utility residual oil fired
boiler (as far as particulate emission factors were concerned) produced fly ash particles
with an essentially unimodal PSD with a mean diameter of approximately 0.1 J.1m.
Conversely, a pilot scale lire-tube package boiler produced particles with a weak bimodal
size distribution, which included a small fraction (-0.2%) of the mass with particle
diameters less thail 0.1 Jlm and a large fraction (-99.8%) of the mass with particle
diameters between 0.5 and 100 J.1m. Here the large particles were shown to consist of
large porous carbonaceous cenospheres resulting from poor carbon burnout, and this
characteristic is not uncommon for that class of equipment. Although the total particulate
mass concentrations in the flue gas of the refractory-lined combustor were less than half
of those of the fire-tube combustor, ultrafine particle concentrations of the refractory-
lined combustor were notahly larger than those of the fire-tube boiler. Volume PSDs
obtained from two independent particle sizing instruments were, with only a few very
reasonable assumptions, consistent with independently measured total mass emission
rates for both equipment types.
20
-------
The results presented here provide insight into mechanisms of fine particle
forma.tion from residual oil combustion. For the refractory-lined combustor, where very
few large particles were formed, the PSD was ne:.lrly unimodal with a mean diameter of
approxi!nately 0.1 J..Im. These particles were composed primarily of trace species
containing V, Ni, Zn, Fe, Cu, and sulfur. Additionally, thes~ particles contained very
little carbon (based on LOI) and the particulate bound sulfur was speciated almost
exclusively as sulfates. The fire-tube boiler produced PM with weak bimodal behavior.
A portion (predominantly metals and sulfur) reporting to a small mode with a mean
diameter of approximately 0.7 to 0.8 J..Im. and, a large portion (comprised primarily of
char) producing a broad coarse mode with a mean diameter of approximately 40 to 50
J..Im. Both of these types of behavior provide circumstantial evidence for a mechanism of
fine particle formation from residual oil combustion as follows. Commonly considered
nonvolatile metals are likely released into the gas phase during the last stages of carbon
burnout. and this is why the accumulation mode for particles formed from vapor
nucleation was so small for the fire-tube boiler. For the refractory-lined combustor,
where char burnout was nearly complete, most of the nonvolatile metals were released
. ~\,.'...
into the gas phase. except for a small residual amount (not detected in the volume PSDs -
Figure 6) that may have coalesced with other metal species at the very end of char
burnout. Volatile metals, such as Ph. may vaporize earlier in the process and diffuse
through the char to the gas phase. Hence, the Pb-containing particulate PSD has a strictly
unimodal behavior.
Further insight was gained through mathematical modeling of the aerosol
dynamics, the predictions from which were compared to experimental measurements. In
each case, it was assumed that 2 to 4 nm nuclei formed from a vapor-phase metal fume,
and then allowed to grow by coagulation. The fire-tube package boiler measured PSDs
were consistent with a model in which 0.2% of the final particulate mass (approximately
0.4% of the fuel oil ash) was at one time nuclei (formed from vaporized trace elements),
and evolved via coagulation. These nuclei do not readily interact with the coarse mode of
unburned carbonaceous cenospheres due to the large differences in particle number
concentrations. The refractory-lined combustor PSDs were predicted using the same
21
-------
model, but with 100% of the final particulate loading being assigned to nuclei in the 2 to
4 nm range. This arrangement yielded a PSD which compared weB with the measured
PSD. Hence, these theoretical calculations strongly supported the mechanisms of fine
particle formation described above.
ACKNOWLEDGMENTSIDISCLAIMER
Portions of this work were conducted under EP A Purchase Order 8CR244NASX
with J.O.L. Wendt and EPA Contract 68-C-99-201 "ith ARCADIS Geraghty & Miller.
The authors gratefully acknowledge the contribution$ of C. Elmore, C. King, E. Squier,
and D. Janek of ARCADIS Geraghty & Milkr (0 th~ ~xperimental efforts, and to S.
Wasson, P. Groff, and M. Calvi of EPA-APPCD and G. Huffman and F. Huggins of the
University of Kentucky for analytical support. The research described in this article has
been reviewed by the Air Pollution Prevention and (.,ntrol Division, U.S. Environmental
Protection Agency, and approved for publication. Tr..: contents of this article should not
be construed to represent Agency policy nor does m~:1tion of trade names or commercial
products constitute endorsement or recommendation for use.
22
-------
p
REFERENCES
Bacci, P., Del Monte, M., Longhello, A., Piano, A., Prodi, F., Redaelli, P., Sabbioni, C.,
and Ventura, A., Characterization of the particulate emission by a large oil fuel fired
power plant, J. Aerosol Sci., 14(4),557-572 (1983).
Bachmann, J.D., Damberg, RJ., Caldwell, J.C., Edwards, C., and Koman, P.O., Review
of the National Ambient Air Quality Standards for Particulate' Matter: Policy Assessment
of Scientific and Technical Information, EPA-452/R-96-013 (NTIS PB97-115406),
Office of Air Quality Planning and Standards, Research Triangle Park, NC (1996).
Barin, I., Knacke, 0., and Kubaschcwski, 0., Thermochemical properties of inorganic
substances, Springer-Verlag, New York, NY (1973).
Bann, I., Knacke, 0., :md Kubaschewski, 0., Thermochemical properties of inorganic
substances, supplement, Springer-Verlag, New York, NY (1977).
Barin, I., Thermochemical data of pure suhstances, VCH Verlagsgesellschafl, New York,
NY (1989).
Bann, I., Thermochemical data of pure suhstances, VCH Verlagsgesellschaft, New York,
NY (1993).
CARB Method 501- Detcrmination of Sizc Distribution of Particulate Matter Emissions
from Stationary Sources. Slate of California Air Resources Board Stationary Source Test
Methods: Volume 1 - Mcthods for Determining Compliance with District Nonvehicular
(Stationary Source) Emission Standards. Adopted March 23, 1988; amended September
12, 1990.
Chase, M.W. Jr., JANAF thermochemical tables, 3rd ed., parts 1&2, American Institute
of Physics, New York, NY (1986).
Cheng, R.J., Mohnen, V.A., Shen, T.T., Current, M., and Hudson, J.B., Characterization
of particulates from power plants, J. Air Poll Control Assoc., 26(8), 787- 790 (1976).
Cox, J.D., Wagman, D.O., and Medvedev, V.A., CODATA Key Values for
Thermodynamics, Hemisphere Publishing Corp., New York, NY (1989).
Davis, S.B, Gale, T.K., Wendt, J.O.L., and Linak, W.P., Multicomponent coagulation and
condensation of toxic metals in combustors, 27h Comb. (Int.) Symp., 1785-1791, Comb.
Inst., Pittsburgh, PA (1998).
Dockery, D.W., Pope, c.A., III, Xu, X., Spengler, J.D., Ware, J.H., Fay, M.E., Ferris,
B.G., Jr., and Speizer, F.E., An association between air pollution and mortality in six U.S.
cities, New EnglandJ. Med., 329, 1753-1759 (1993).
23
-------
Dreher, K., Costa, D., Horrman, A., Bonner, J., nnd O~('Irnjo- Vargas, A., Pulmonary
toxicity of urban air particulate matter (PM), Air & Waste Management Association
Meeting on Measurement of Toxic and Related Air PollutantS, Research Triangle Park,
NC, May 1996a.
Dreher, K., Jaskot, R., Richards, J.H., Lehmann, J.R., Winsett, D., Hoffman, A., and
Costa, D., Acute pulmonary toxicity of size-fractionated ambient air particulate matter,
Amer. J. Respir. Crit. Care Mell., 153(4:AI5) (1996b).
Dreher, K., Jaskot, R., Lehmann, J.R., Richards, lH., McGee, J.K., Ghio, A.J., and
Costa, D.L., Soluble transition metals mediate residual oil fly ash induced lung injury, J.
Toxieol. Environ. Health, 50, 285-305 (1997).
Ebbinghaus, B.B., Thermodynamics of gas phase chromium species: the chromium
chlorides, oxychlorides, fluorides, oxytluorides, hydroxides, oxyhydroxides, mixed
oxyfluorochlorohydroxidcs, and volatility calculations in waste incineration processes,
Combust. & Flame, 101, 119-137 (1993).
Ebbinghaus, B.B., Thenllodynamics of gas phase chromium species: the chromium
oxides, chromium oxyhydroxidcs, and volatility calculations in waste incineration
processes, Combust. & Flame, 93, 311-338 (1995).
EPA Test Method lA - Sample and Velocity Traverses for Stationary Sources with Small
Stacks or DuClS, in 40 CFR Part 60 Appendix A, Government Institutes Inc, Rockville,
MD, July 1994a.
EP A Test Method 5 - Determination of Particulate Emissions from Stationary Sources, in
40 CFR Part 60 Appendix A, Government Institutes Inc, Rockvi11e, MD, July 1994b.
Federal Register, 62 FR 38652, July 18, 1997.
Frandsen, F., Dam-Johansen, K., and Rasmussen, P., Trace elements from combustion
and gasification of coal - an equilibIium approach, Prog. Energy Combust.'Sci., 20, 115-
138 (1994).
Galbreath, K.C., Zygarlicke, C.J., Huggins, F.E., Huffman, G.P., and Wong, J.L.,
Chemical speciation on nickel in residual oil ash, Energy &Fuel.s, 12,818-822 (1998).
Garg, S., EP A Method 0060 - Methodology for the Determination of Metals Emissions in
Exhaust Gases from Hazardous Waste Incineration and Similar Combustion Processes, in
Methods Manual for Compliance with the BIF Regulations: Burning Hazardous Waste in
Boilers and Industrial Furnaces, EPA/530-SW-91-010 (NTIS PB91-120006), pp. 3-1
through 3-48, Oflice of Solid Waste, Washington, DC, December 1990.
24
-------
p
Gclbard, F., nnd Seinfc1d, J.H., Simulation of mullicomponent aerosol dynamics. J.
Colloid Interface Sci., 78(2), 485-501, (1980).
Goldstein H.L., and Siegmund, C.W., Innuence of heavy fuel oil composition and boiler
combustion conditions on particulate emissions, Environ. Sci. & Tee/mol., 10(12), 1109-
1114 (1976).
Goldstein H.L., and Siegmund, C.W., Particle emissions from residual fuel fired boilers:
influence of combustion modifications, J. of Engineering Power, 371-377 (1977).
Gurvich, L.V., Veyts, LV., and A1cock, C.B., ThellTIodynamic Properties of Individual
Substances, 4th ed. Vol. 2, Pts. 1&2, Hemisphere Publishing Corp., Washington, DC
(1991).
Huggins, F.E., and Huffman, G.P., Speciation of clements in NIST particulate matter
SRMs 1648 and 1650. J. Hazardous Materia!s, in press (1999).
Linak. W.P., and Wendt, J.a.L., "Toxic metal emissions from incineration: mechanisms
and control," Prog. Energy C0111bu.rr. Sci., 19, 145-185 (1993).
Linak, W.P., and Wendt. J.a.L.. "Trace metal transformation mechanisms during coal
combusrion," Fuel Proces.'i. Technol., 39,173-198 (1994).
Linak, W.P., Srivastava, R.K., and Wendt, J.a.L., Metal aerosol formation in a laboratory
swirl flame incinerator," Combust. Sci. Techno!., 10 1 (1-6), 7-27 (1994).
Linak. W.P., Ryan, J. V.. and Wendt. J.a.L.. FOllTIation and destruction of hexavalent
chromium in a lahoratory swirl flame incinerator, Combust. Sci. Techno!., 116-117, 479-
498 (1996).
Matsoukas, T., and Friedlander, S.K., Dynamics of aerosol agglomerate fOllTIation. J.
Colloid Infe1face. Sci., 146(2).495-506, (1991).
McBride, B.J., Gordon, S., and Reno, M.A., Coefticients for ca1culating thermodynamic
and transport prope11ies of individual species, NASA Technical Memorandum 4513
(1993).
Miller, C.A., Linak, W.P., King, C., and Wendt, J.a.L., Fine particle emissions from
heavy fuel oil combustion in a firetubc package boiler. Combust. Sci. and Technol., 134,
477-502, (1998).
National Research Council, "Review of EP A' s Particulate Matter Research Program,"
National Academy Press, Washington, DC, 1998.
25
-------
Owens, T., Wu, C.Y., and Biswas. P., An equilibrium analysis for reaction of metal
compounds with sorbents in high temperature systems. Chem. Eng. Comm., 133. 31-52
(1995).
Scotto. M.A. Peterson. T.W., and Wendt, J.O.L., Hazardous waste incineration: the in-
. . ~
Sltu capture of Icad by sorbentsin a laboratory down-now combustor, 24' Comb. (Int.)
Symp., 1109-1118, Comb.lnst., Pittsburgh. PA (1992).
Steele, W.J., Williamson, AD., and McCain. J.D., Construction and operation of a 10
cfm sampling system with a 10: 1 dilution ratio for measuring condensable emissions.
EP N600-8-88-069 (NTIS PB 88-198551), Air and Energy Engineering Research
Laboratory. Research Triangle Park, NC, April 1988.
TAPP - Thelmodynamic Propel1ies Softwan::, V2.2, ES Microware Inc., Hamilton, OH
(1995).
U.S. Department of Energy. "Fuel Oil and Kerosene Sales 1997," DOE/EIA-0535(97),
Energy Information Administration, Washington, DC, August 1998.
U.S. Environmcntal Protection Agency, National Ambient Air Quality Standards for
Particulate Matter: proposed decision, 40 CFR, Part 50 (1996a).
U.S. Environmcntal Protection Agency, Compilation of Air Pol1utant Emission Factors
Volume 2: Stationary Point and Area Sourccs. AP-42. 5th Ed., U.S. EPA, Office of Air
Quality Planning and Standards, Rescarch Triangle Park, NC, November 1996b.
Williams, A. Fundamentals of oil combustion, Prog. Energy Combust. Sci., 2. 167-179
(1976).
Wilson, R., and Spengler. J.D., cds., Pal1ic1es in Our Air: Concentrations and Health
Effects, Harvard Univ. Press, Cambridge, MA (1996).
Wolff, G.T., Closure by the Clean Air Scientific Advisory Commiuee (CASAC) on the
staff paper for particulate matter, EP A-SAB-CASAC-L TR-96-008, U.S. Environmental
Protection Agency, Washington, DC, June 13, 1996.
Wu. C.Y., and Biswas, P., An equilibrium analysis to determine the speciation of metals
in an incinerator, Combust. & Flame, 93, 31-40 (1993).
26
-------
"D
Table 1. Vanadium, niclcd, zinc,lron, copper,lead, and chromium species considered (or equilibrium calculations.
peC1C:S
I e / I
VN 112/73 NiO L2I84 ZrI) LI/67 F~~, J3I78 ClIO J 12177 PbO JI2/71 CrN JI2/73
m 112/73 Ni(CO), B5I89 Zn(s) CODA89 J9/66 Cu, J9I66 PbS J6m CrO(OII), BE9S
VO 112/73 NiO,H, U/84 Zn(1) CODA 89 Fe(OH), 112166 Cu(s) CODA89 Pb J9/63 Ct(OH), BE9S
V(sJ J6173 Ni(s) 112/7 6 ZnSO,(a) Ui79 Fe~, 112166 Cu(1) CODA89 PbA BS/89 Cr(OH), BE9S
V(I) J6I73 Ni(1) 112/76 ZnSo,(b) Ui79 Fe, (s) BAR73 CuO(s) JI2/77 Pb(s) TPIS91 Ct(OH), BE9S
VN(s) 112/73 NiS(.) 112/76 ZnO(s) BMS/84 Fe,N(a) BAR77 CuO,H,(s) J6/66 Pb(1) TPIS91 Ct(OH), BE9S
VO(s) JI2/73 NiS(b) JI2/76 ZnO~H,(s) L7n6 Fe,N(b) BAR77 CuSO,(s) J6I66 PbO(rd) 11 2/71 Cr(OH), BE9S
VO(l) JI2/73 NiS(1) 11 2/76 Zn (s) BMS/84 Fe(a) 13m Cu,0(s) J 12/77 PbO(yw) JI2171 CrO(OH), BE9S
~O,(s) 112173 NiS,(s) 13m ZnS(1) BMS/84 Fe(b) 13m ~O(I) 112/77 PbO(1) JI2171 cragH BE9S
,0,(1) 112/73 NiS,(1) 13m Fe(c) 13m Cu, 6:(S) J6/66 PbO,(s) JI2171 CrO( H), BE9S
V,O,(I) J6I73 Ni,S, (I) JI2/76 Fe(d) 13n8 CUC ,(s) BAR77 PbS(s) J6m era BE9S
~O,(ll) J6I73 Ni}S ,(11) 11 2/76 Fe(1) 13m CuFc,o,(I) BAR77 PhS(/) J6m cra BE9S
,0,(1) J6173 N.,s,(1) 11 2176 FeC,O,(1) 13178 CuFc,O,(2) BAR77 ~(S) J I 2171 ~2\n. BE9S
~O,(S) J6I73 ~S,(S) 13m FeO(s) J6/65 CU~O,(3) BAR77 ,(s) B5/89 BE95
° (I) J6173 N" O,(s) IB 1 993 FeO(l) J6I65 C (s) BAR73 PbSO,(I) BAR 73 crail BE95
VO~,(s) USBM87 C~iO,(s) IBI993 FeS(a) J9m Cu}'c,O,(I) BAR77 PbSO,(2) BAR13 CrOOH BE95
i C(s) BS/89 FeS(b) J9m Clife,O,(2) BAR77 PbSO,(1) BAR73 Cr(s) J6I73
NiCO,(S) BS/89 FeS(c) J9m Cucfut,(1) BAR77 Cr(1) J6I73
NiFe,o,(I) BAR73 FcS(1) J9m u (I) BAR73 CrN(s) JI2m
NiFe&,(2) BAR73 FeSO,(s) J6/66 Cu.5(2) BAR73 Cr.N(s) J 12173
Ni I) BAR73 FeS.(s) J9m Cu;S(3) BAR73 Cr;O,(s) JI2m
Ni0(2) BAR73 Fe,O,(s) J6I6S Cu,S(/) BAR73 - Cr,O,(I) J 12173
NiO(3) BAR73 Fe$,O,,(s) J6/66 Cr55'(S) BE93
fe ° (s) J6I6S Cr( ),(s) BE93
~ CdeO,(s) IBI993 Cr.(SO.>,(s) BE93
-J 1','(OII).(s) 1B199.' CrJ"O,(s) BE9.'
1'«OII);(s) ID 199) Cr:NiO,(s) BE93
CuFe.O,( I) BAR77 C'r,C:(s) BE9J
r"F..;O.C2J nAln7 ('r,C,Cs, DE9.'
('IIF~,o,C.'1 IIAR77 (",O.(s, IIE')'\
Cu,Fe,o ,(I) BAR 77 Cra;(S) BE93
Cu,fe,o,(2) BAR77 CrO,(I) BE93
CU,Fe,o ,(I) BAR 77 crS(1) BAR77
CrS(2) BAR77
J9/63: JANAP 9/63 - C!use (1986)
J6I6S: JANAP 9/65 - Chase (1986)
112165: JANAP 12165 - C!use (1986)
J6I66: JANAP 6/66 - Chase (1986)
J9/66: JANAP 9/66 - Chase (1986)
J 12166: J ANAP 12166 - Chase (1986)
112/71: JANAP 12/71- Chase (1986)
J6I73: JANAP 6173 - Chase (1986)
112173: JANAP 12173. Chase (1986)
112/76: JANAP 12176 - Chase (1986)
J3m:JANAP3n7 -Chase (1986)
J9m: JANAF 9m. C!use (1986)
J12I77: JANAP 12177. Chase (1986)
13m: JANAP 3178 - Chase (1986)
J6I79: JANAP 6179 - Chase (1986)
J9/84: JANAP 9/84 - Chase (1986)
......
w
L 1/67: Lewis 1/67 - McBride el 01. (1993)
L7n6: Lewis 7n6 - McBride cl aI. (1993)
Un9: Lewis 3n9 - McBride el 01. (1993)
L2I84: Lewis 2184 - McBride CI aI. (1993)
1.3/84: Le\
-------
p
,
I
i
Table 2. Elcment11 concentrations used for equilibrium calculations:
g
g-mo cs
H
N
S
o
H2
Ash
FUEL TRACE g g-moles
ELEMENTS)
Cr l.05c-4 2.02e-6
Cu 3.50c-4 5.51e-6
Fe 2.lOc-3 3.76e-5
Pb 4.50e-4 2.17e-6
Ni 3.00e-3 5.11e-5
V 2.20c-2 4.32e-4
Zn 7 AOc- 3 1.13e-4
AIR L g-molcs
O2 260.29 11.62
N2 97Y.55 43.73
I Basis: 100 g of oil.
2Not applicable.
)Trace elements> 1 ~g!g of oil.
420% excess air.
28
p
-------
t)
Table 3. Comparison of calculated mass concentrations detennined from measured volume PSDs with measured mass concentrations determined by EPA M-S.
VOLUME MASS MASS
CONCENTRA TIONS CONCENTRATIONS CONCENTRATIONS
(MEASURED) (CALCULATED) (MEASURED)
Clan' Icm') (mg/m') I (mg/m')l
BOILER COMBUSTOR :' '"IlLER COMB USTOR BOILER COMBUSTOR
SMPS 8.8e+I 2. 7e+4 3.Ie-l 9 .3e+ 1 . .
INSITEC 9.8e+4 1.9e+2 1.8e+2 3.6e-l . .
TOTAL 9.8e+4 2. 7e+4 1.8e+ 2 9.3e+ I 184 (6.2») 93
IAssumed fine mode panicle density" 3..5 gJcm' based on typical densities for vanadiur.l species. Assumed coarse mode particle density.. I.85 gJcm).
~easured mass concentrations determined using EPA Method 5.
'Number in parentheses indicated standard deviation based on multiple measurements.
~
(0
-------
p
-- "
"",
,
i
I
Table 4. Comparison of trace element emissions for high sulfur No.6 oil from two
combustion systems (J.lg/m ').
ELEMENT BOILER COMBUSTOR
Sb 7.7 -
As 6.3 -
Be 0.09 -
Cd 3.5 -
Cr 11. -
Cu 170. 200.
Fe 74U. fWD:
Pb 89. -
Mg 1200. 1700.
Mn 16. -
.
Hg <2.2 -
Ni 1200. 1400.
Na - 2100.
V 9800. 12000.
Zn 3300. 3000.
30
-------
p
LIST OF FIGURES
Fig. 1
fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
1997 U.S. residu:l1 fuel oil s:lles by location and end use. Total sales equal
47.3x 109 L (12.5xl09 gal). The 13 other Great Plains and Rocky
Mountain St:ltes combined :lccount for less thnn 1 % of total sales (adapted
from U.S. DOE. 1998).
Two EPA combustion systems.
... .
EPA dilution sampling system.
Fuel oil droplet PSDs from the fire-tube boiler atomizer.
Residu:lI fuel oil equilibrium predictions.
Me:l5urcd volume and ca1cu\:1tcd mass PSDs.
SEM image of fuel oil char from the firc-tube boiler.
Elemental PSDs detcrmined from impactor samples from the refractory-
lined combustor.
Predictcd evolution or PSDs via coagulation.
31
-------
p
. "\.<
Figure 1
CD
(/)
:)
"C AL
c::
W AR
S FL
<5 GA :::::
Q3 KY
:J LA
U.
-' MS
co
:J NC .:.:.:.:.:
"C .. ..
"(i) SC::
~ TN
(5 1)( ::.
(/) VA
~ 'IN
Cf)
f'-.
0>
0>
,-
CT
DC
DE
MA
MD
ME
NH
NJ
NY
PA
AI
VT
IL
N
1A
M
MN
MO
OH .
Vv1
AK
CA
HI
OR
WA
Northeast
38%
[ill Vessel Bunkering - 40%
[ZI Electric Utility - 37%
D Industrial - 15%
EJ Commercial - 6%
I;.'SI Oil companJ
i52] Military - 2%
D Other
Midwest
3%
Pacific
18%
o
500 1000 1500 2000 2500
Million Gallons
32
-------
p
Fire-tube Boiler
Refractory-lined
Combustor
Figure 2
To aIr pollutton control system ..
ElectrIcal mobility, Inerttallmpactors
Dllutton sampler, In-situ light scatterIng
I. 1 m -I
Gas sample port
Fuel 011
'.
Steam out
Natural gas
Water In
- To air pollution control system
In-sItu light scattering
Dilution sampler, electrical mobIlity, Inertlallmpactlors
I" 1m ~I
33
Combustion air
-------
. -..:
....
Dilution cone
p Orifice
meter
Dilution air blower
Dilution air
90 cfm
Sample blower
Exhaust
Figure 3
34
-------
15 No.6 fuel oil droplet PSD
L.
Q) Fire-tube boiler
.......
Q) 2.54 em from atomizer tip
E
.~ 400 K, 200 kPa cP
-0 \
o ' ,
c: 0
Q) 10
.~ 0
0>
- 0
0 0
CIJ
.......
Q) 0
a.
0 5 0
L.
"C
-
0
....... 0
c:
Q)
u 0
'-
Q)
0...
1 10 100
Droplet diameter (!-.1m)
Figure 4
35
-------
1.
O.
'. . O.
\,...,
1.
O.
(/)
O.
'0
1.
a.
(/)
"0
O.
.....
.~
"0 O.
.S
(/) 1.
ro
.....
c
O.
E
-------
1e+3
1e+S
5e+2
.
o
o
o
o
..........
('I')
E
~
('I')
E
2; 5e+4
c.
o
Q)
o
=0
3;
-0
Oe+O
~ 0.01 '
.
.
.
electrical mobility
0.1
o
o
10,:
q:.'.
'0
o
o
o
light scattering
~
.1
.
.
.
Oe+O
30
.
20
,
..
.
.
o Fire-tube boiler
. Refractory-lined combustor
..........
('I')
E
-
Q)
E
-
10
. .
.
c.
o
Q)
.Q
~
-0
.
.
.
.
0.01
0.1
1
Diameter, Dp (IJm)
10
100
Figure 6
37
-------
p
,...
Figure 7
38
-------
1.5
1.0 Vanadium
OXRF
0.5 . ICP
0.0
1.5
D.. 1.0 Nickel
0 OXRF
0> . ICP
0 0.5
~ 0.0
"C 1.5
-d' Zinc
~ 1.0 OXRF
::J . ICP
(/) 0.5
CCS
CD
E 0.0
....... 1.5
c:
(]) Iron
E 1.0 OXRF
(]) . ICP
Q) 0.5
"@
.......
0 0.0
....... 1.5
-
0 Copper
c: 1.0
0 OXRF
~
U 0.5
CCS
~
-
"C 0.0
(]) 1.5
.~
Cd 1.0 Lead
E OXRF
~
0 0.5'
Z
0.0
1.5
Sulfur
1.0 0 XRF (8)
. IC (804.2)
0.5
0.0
0.01 0.1 1 10 100
Aerodynamic diameter, Dp (Ilm)
Figure 8 39
-------
-
C")
E
-
0>
.s 10
c
0
~
~ 5
+-'
C
Q)
()
c
~ 0
0 () 10
CI)
CI)
Ct1
:2: 5
Fire-tube boiler
180 mg/m3
10
5
0.001
0.01
100 0.001
0.1 1
Diameter, Dp (IJm)
10
Figure 9
"
Refractory-lined combustor:~
90 mg/m3
t = 0.1 5
t = 1.05
'"
t = 10 s
100
0.01
10
0.1 1
Diameter, Dp (IJm)
------- |