United States
Environmental Protection
Agency
National Risk Management
Research Laboratory
Cincinnati, OH 45268
Research and Development
EPA/600/SR-01/106 May 2002
Source Sampling Fine
Participate Matter: Wood-Fired
Industrial BoilerD
Dave-Paul Dayton and Joan T. Bursey
Fine particulate matter of aerody-
namic diameter 2.5 m or less (PM25)
has been implicated in adverse health
effects, and a National Ambient Air
Quality Standard for PM25 has been pro-
mulgated (July 1997) by the U. S. Envi-
ronmental Protection Agency. A national
network of ambient monitoring stations
has been established to assist states
in determining areas which do not meet
the ambient standard for PM25. For such
areas, it is important to determine the
major sources of the PM25 so the states
can devise and institute a control strat-
egy to attain the ambient concentra-
tions set by the standard.
A tool often used by states in ap-
portioning ambient PM25 to the sources
is a source-receptor model. Such a
model requires knowledge of the PM25
chemical composition emitted from
each major source contributing to the
ambient PM25 as well as the chemical
composition of the PM25 collected at
the receptor (ambient monitoring) sites.
The report provides such a profile for a
wood-fired industrial boiler equipped with
a multistage electrostatic precipitator
(ESP) control device. Along with the
PM25 emission profile, data are also
provided for gas-phase emissions of
several organic compounds. Data are
provided in a format suitable for inclu-
sion in the EPA source profile data-
base, SPECIATE.
This Project Summary was developed
by the National Risk Management Re-
search Laboratory's Air Pollution Pre-
vention and Control Division, Research
Triangle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Source profiles for PM emissions to
date have been limited mostly to PM of
aerodynamic diameter of 10 m or less
(PM10) and have consisted of only el-
emental composition data. For certain
sources, where the emitted PM is carbon-
aceous in nature, a knowledge of the
organic species present in the PM can
greatly improve the utility of the profile for
source apportionment purposes. There-
fore, the principal objective of this study
was to acquire such organic speciation
data for an industrial wood-fired boiler in
a form and of a quality suitable for use in
source-receptor modeling.
The industrial boiler selected for test-
ing was a relatively modern, watertube,
pneumatic, vibrating stoker-type unit fu-
eled with wood/bark chips via a continu-
ous screw-feed conveyor belt. When
operated at the design heat input rate,
the boiler generated 74,843 kg/hr
(165,000 Ib/hr) of superheated steam, 65.3
atm/404 °C (960 psig/760 °F). Emissions
were controlled by a multicyclone type
dust collector followed by a multistage
ESP.
Fuel, acquired from the local munici-
pality, was derived principally from tree
trimming waste that had passed through
a chipper/shredder. This material was
stored outdoors in a large pile until used
so that the type, age, and moisture con-
tent of the wood fuel fed to the boiler
varied from day to day. Sampling was
conducted over a 2-day period in August
2000.
-------
Experimental Methods
To relate source emissions measure-
ments to ambient pollutant concentrations,
it is important that the measurement tech-
niques used to characterize and quantify
source emissions are consistent with the
techniques used for ambient pollutant
measurements. A source sampling strat-
egy to eliminate many potential biases
between source and ambient sampling
measurements is to use the same sam-
pling and analytical techniques for both
source and ambient monitoring. There-
fore, for this study, a source sampler was
used which diluted the hot exhaust emis-
sions (ca. 50:1 dilution) with cleaned am-
bient air to near ambient temperature and
pressure prior to collection of the source
samples. Additionally, emissions were
provided a sufficient residence time at
near-ambient conditions prior to sample
collection such that semivolatile com-
pounds in the diluted exhaust could dis-
tribute between the gas and particle
phases as they would in the ambient en-
vironment.
The custom-built dilution sampler fol-
lowed the basic design originated by L.M.
Hildemann et al., but was modified to
allow for continuous automatic adjustment
of exhaust and dilution air flow rates and
for ease in assembly and transportation.
Dilution air was ambient air which had
passed through a high-efficiency particu-
late air (HEPA) filter, an activated carbon
bed, and a final TeflonŽ membrane filter
before mixing with the exhaust gas. Inte-
grated samples of the cleaned dilution air
were also collected during the course of
a test run, and exhaust samples were
corrected for any contaminants which may
have remained in the dilution air after
cleaning.
Particulate samples were collected for
organic analyses using two different types
of sampling arrays. One array consisted
of a cyclone with a d50 cut point of 2.5 m,
followed by a series of two 200 mm long
annular denuders coated with XAD-4
resin, followed by a quartz fiber filter, fol-
lowed by two polyurethane foam (PUF)
plugs. The other array was the same but
without the annular denuders. The pur-
pose of the annular denuders was to re-
move gas-phase semi-volatile organic
compounds (SVOCs) prior to the quartz
filter so as to avoid a positive adsorption
artifact on the quartz filter sample. Quartz
filters were prebaked at 550 °C for 12 hr,
and annular denuders and PUF plugs
were solvent-cleaned prior to use to re-
move organic contaminants prior to sam-
pling. Teflon membrane filters were used
to collect PM25 samples for mass, inor-
ganic ion, and elemental measurements.
Two dinitrophenylhydrazine (DNPH)-
coated silica gel cartridges in series were
used to collect volatile carbonyl com-
pounds. An electropolished stainless steel
canister was used to collect gas-phase
samples for analysis of volatile organic
compounds (VOCs) and selected air toxic
compounds.
The organic and elemental carbon con-
tent of the quartz filter PM25 samples was
determined by the National Institute of
Safety and Health (NIOSH) Method 5040
using a Sunset Laboratory thermal opti-
cal transmittance instrument. Individual
organic compounds present in the PM25
samples collected on the quartz filters
were determined by extracting the filters
with hexane and a 2:1 mixture by volume
of benzene and isopropanol. Prior to ex-
traction, the filters were composited as
necessary to achieve a total of approxi-
mately 0.5 mg of organic carbon and were
spiked with a mixture of deuterated inter-
nal recovery standards.
An aliquot of the combined organic ex-
tract was derivatized with diazomethane
to yield methyl esters of any fatty acids.
An aliquot of the methylated extract was
further derivatized with bis(trimethylsilyl)
trifluoroacetamide-N,O-bis(trimethylsilyl)
acetamide to convert compounds such
as levoglucosan and cholesterol to their
trimethylsilyl derivatives. Both derivatiza-
tions were performed in order to allow
these compounds to be separated and
eluted from a gas chromatography col-
umn. Gas chromatography/mass spec-
trometry was used to identify and quantify
the individual organic compounds present
in the extracts. Target compounds were
positively identified by comparing mass
spectra of the analytes with those ob-
tained for over 100 authentic compound
standards.
The organic compounds were quanti-
tated by referencing each compound to
one or more deuterated internal standards
spiked into the sample to correct for losses
of the analytes which may have occurred
in the compositing, extracting, concen-
trating, and derivatizing steps. An exten-
sive set of standards of target compounds
at known concentrations, which also in-
cluded the deuterated internal standard
compounds, was used to establish 3- or
5-point calibration curves from which the
concentrations of the analytes were de-
termined.
PM mass was determined gravimetri-
cally by weighing the Teflon filters before
and after sample collection. These filters
were maintained at 20-23 °C and a rela-
tive humidity of 30-40% for a minimum of
24 hr prior to weighing on a microbal-
ance. Elemental analysis was carried out
by x-ray fluorescence spectrometry. Wa-
ter soluble ions were analyzed by ion
chromatography.
Carbonyl compounds collected on the
DNPH-coated cartridges were analyzed
using EPA Compendium Method TO-11a.
Air toxics and speciated nonmethane or-
ganic compounds were determined ac-
cording to the procedures of EPA
Compendium Method TO-15 and an EPA
technical assistance method, respectively.
Results
Total speciated nonmethane organic
compound mass emission rates were
13.19 and 3.45 mg/kg fuel for sampling
days 1 and 2, respectively. Total VOC
(speciated plus non-speciated) mass emis-
sion rates were 22.64 and 6.49 mg/kg
fuel on sampling days 1 and 2, respec-
tively. Emission rates of total gas-phase
carbonyl compounds ranged from 8.27
(day 1) to 2.90 mg/kg fuel (day 2). Aver-
age PM25 mass emission rates were 3.54
and 1.23 mg/kg fuel for sampling days 1
and 2, respectively.
An explanation for the observed differ-
ence in mass emission rates of both gas-
eous and PM25 emissions between the 2
test days could not be deduced with con-
fidence. Both the boiler and sampling sys-
tem operating parameters were
essentially identical for both days. The
only apparent variable which may have
contributed to the difference in emission
rates was the nature of the chipped wood
fuel itself since the fuel was selected from
different locations in the large on-site stor-
age pile during the 2 days of testing.
However, only one composited wood fuel
sample from the storage pile was ana-
lyzed, so any significant difference in
boiler feed between the two tests could
not be determined.
Concentrations of propylene, methyl-
ene chloride, and benzene in the diluted
exhaust gas were higher than typically
found in ambient air. However, the con-
centrations of propylene and methylene
chloride in the diluted exhaust gas were
about the same as found in the cleaned
dilution air, indicating that these two com-
pounds were present in both ambient air
and in the source emissions at about the
same level. Benzene concentrations in
the diluted exhaust gas were nearly 3
orders of magnitude higher than concen-
trations observed in the dilution air, indi-
cating that this compound was emitted
from the source at relatively high levels
compared to ambient levels.
Elemental carbon and organic carbon
content of the PM25 collected on quartz
filters was found to be highly dependent
on whether an XAD-coated denuder was
-------
inserted in the sampling line prior to the
filter. Without the denuder, the amount of
organic carbon found on the quartz filters
was 2.6 times the amount found with the
denuder, thus providing confirmatory evi-
dence for a positive adsorption artifact on
the quartz filters when no denuder was
present.
Of the major water soluble ions, only
sulfate and potassium ions were above
quantitation limits. Total potassium as
measured by x-ray fluorescence
spectrometry agreed well with water-
soluble potassium determined by ion chro-
matography. Silicon was the element
found in greatest concentration in the
PM25, perhaps originating from the fire-
brick lining of the boiler.
Individual organic compounds compris-
ing the organic carbon fraction of the
PM25 emissions consisted mostly of poly-
nuclear aromatic hydrocarbons (PAHs),
alkanes (>C15), alkanoic acids (>C8),
and the iso- and anteiso-alkanes.
Levoglucosan, a marker compound for
biomass combustion, was found in the
particulate matter but not in the relatively
large amounts characteristic of open burn-
ing of biomass material or emissions from
wood-fired domestic appliances. Resin
acids (e.g., pimaric, isopimaric, and
sandarapimaric acids) used as markers
for softwood combustion and methox-
yphenols used as markers for hardwood
combustion also were not found above
quantification limits. Therefore, the organic
compound emission profile for the wood-
fired industrial boiler was very unlike pro-
files for residential woodstoves and
fireplaces and for biomass open burning.
This observation is not unexpected since
the combustion regime for the boiler was
substantially different from that of
woodstoves and open burning and since
the boiler emissions were controlled by
an ESP whereas wood burning appli-
ances and open burning are typically un-
controlled.
A complete compilation of the over 300
gas- and particle-phase speciated com-
pounds is available in tabular form in the
full report in a format suitable for inclu-
sion in the EPA SPECIATE emission pro-
file database.
D-P. Dayton and J.T. Bursey are with the Eastern Research Group, 1600
Perimeter Park, Morrisville, NC 27560-2010.
N. Dean Smith is the EPA Project Officer (see below).
The complete report, entitled "Source Sampling Fine Particulate Matter
Wood-Fired Industrial Boiler," will be available at http://www.epa.gov/
ORD/NRMRL/Pubs or as Order No. PB2002-105739 (Cost: $44.00,
subject to change) from:
National Technical Information ServiceO
5285 Port Royal RoadD
Springfield, VA 22161-0001D
Telephone: (703) 605-60000
(800) 553-6847 (U.S. only)
The EPA Project Officer can be contacted at:
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711-0001
-------
United StatesD
Environmental Protection Agency D
CenterforEnvironmental Research InformationD
Cincinnati, OH 45268D
Official Business
Penalty for Private Use
$300
EPA/600/SR-01/106
PRESORTED STANDARDD
POSTAGES FEES PAIDD
EPAD
PERMIT No. G-35D
------- |