EPA 910/9-82-089a
United States
Environmental Protection
Agency
Region 10
1200 Sixth Avenue
Seattle WA 98101
Air & Waste Management Division February 1984
&EPA Residential Wood
Combustion Study
Task 1
Ambient Air Quality Impact Analysis
V^-i Pi tl H t=1JBU
-------
RESIDENTIAL WOOD COMBUSTION STUDY
TASK 1
AMBIENT AIR QUALITY IMPACT ANALYSIS
-------
RESIDENTIAL WOOD COMBUSTION STUDY
TASK 1
Ambient Air Quality Impact Analysis
FINAL REPORT
Prepared by:
John E. Core
Dr. John A. Cooper
Richard T. DeCesar
Dr. James E. Houck
NEA, INC.
Beaver ton, Oregon
Prepared for:
DEL GREEN ASSOCIATES, INC.
ENVIRONMENTAL TECHNOLOGY DIVISION
Woodburn, Oregon.
Under Contract No. 68-02-3566
U.S. ENVIRONMENTAL PROTECTION AGENCY
Region X
Seattle, Washington 98101
Task Manager
Wayne Grotheer
September, 1982
U.S. Envtronmental PwUCtlon
Region V, Library
230 South Dearttffll St?«H
Chicago, ttUortL 60604
-------
THIS REPORT CONSISTS OF SEVERAL DIFFERENT PARTS.
THEY ARE LISTED BELOW FOR YOUR CONVENIENCE.
EPA 910/9-82-089a Residential Wood Combustion Study
Task 1 - Ambient Air Quality Impact
Analysis
EPA 910/9-82-089b Task 1 - Appendices
EPA 910/9-82-089c Task 2A - Current & Projected Air Quality
Impacts
EPA 910/9-82-089d Task 2B - Household Information Survey
EPA 910/9-82-089e Task 3 - Wood Fuel Use Projection
EPA 910/9-82-089f Task 4 - Technical Analysis of Wood Stoves
EPA 910/9-82-089g Task 5 - Emissions Testing of Wood Stoves
Volumes 1 & 2
EPA 910/9-82-089h Task 5 - Emissions Testing of Wood Stoves
Volumes 3 & 4 (Appendices)
EPA 910/9-82-089i Task 6 - Control Strategy Analysis
EPA 910/9-82-089J Task 7 - Indoor Air Quality
-------
DISCLAIMER
This report has been reviewed by Region 10, U. S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policies of the U. S. Environmental Protection Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
-------
TABLE OF CONTENTS
Page
ABSTRACT ii
EXECUTIVE SUMMARY iii
ACKNOWLEDGEMENTS iv
LIST OF TABLES v
LIST OF FIGURES vi
1.0 INTRODUCTION 1
2.0 PROGRAM METHODOLOGY 2
2.1 Air Sampling Program 2
2.2 Analytical Protocol • 6
2.2.1 Gravimetric Analysis 6
2.2.2 X-ray Fluorescence (XRF) 8
2.2.3 Neutron Activation Analysis (NAA) 8
2.2.4 Carbon Analysis 10
2.2.5 PNA Analysis 10
2.2.6 Carbon-14 Analysis 10
2.3 Quality Assurance
2.4 Sample Selection 11
3.0 SOURCE APPORTIONMENT 12
3.1 Chemical Mass Balance 12
3.1.1 Chemical Mass Balance Receptor Model 12
3.1.2 Source Characterization 15
3.2 Carbon-14 16
3.2.1 Fuel Ratio Measurements 16
3.2.2 Contemporary Carbon Source Activity 18
•
3.3.3 RWC Emission Carbon Content 18
-------
Page
4.0 EXPERIMENTAL PROGRAM RESULTS .18
5.0 DISCUSSION OF RESULTS 21
6.0 CONCLUSIONS - . . . . 25
7.0 REFERENCES 27
8.0 APPENDICES
1. Air Monitoring Site Description
2. Source Emission Composition
3. Field Program Sample Summary
4. Source Impact Summaries by Site
5. CMB Source Apportionment Calculations
6. Quality Assurance Program Elements
-------
ABSTRACT
A Study of Residential Wood Combustion
Task 1: -Ambient Air Quality Impact Analysis
The ambient air quality impact of Residential Wood Combustion (RWC)
emissions in eight Pacific Northwest cities was studied during the 1980-
81 space heating season. Seventy-seven worst case fine particle ambient
samples collected in Seattle, Spokane, Ya'kima, Tacoma and Longview,
Washington; Boise, Idaho; Portland and Medford, Oregon were selected
and analyzed for Polynuclear Aromatic Hydrocarbon compounds, 35 trace
elements and carbon. The impact of RWC and other sources was determined
by Chemical Mass Balance calculation and validated against independent
carbon-14 measurements.
This document describes the project design, methodology and results
of the RWC impact monitoring program. The findings of this task form the
basis for projections of future RWC impacts discussed under task 2A of
this series. They also provide the first comprehensive survey of RWC
impacts within Pacific Northwest communities during a period likely to
be typical oi maximum RWC emissions.
-------
EXECUTIVE SUMMARY
Recent increases in the use of wood as a source of residential
space heating within Pacific Northwest states has led to concerns
focusing on public health risks and degrading air quality within many
communities. A comprehensive survey of eight cities in Oregon, Washington
and Idaho was conducted during the October, 1980 - March, 1981 space
heating seasons to provide a better assessment of current maximum 24 hour
RWC impacts on particulate air quality. Seventy-seven selected fine
particle samples from Seattle, Spokane, Tacoma, Longview and Yakima,
Washington; Boise, Idaho; Portland and Medford, Oregon were analyzed for
35 trace elements, carbon, PNA compounds and carbon-14. RWC impacts were
then quantified by Chemical Mass Balance and validated by independent
carbon-14 measurements.
Source apportionment results typically assign from 66% to 84% of the
fine particle mass (less than 2.5 urn) and 36-60% of the TSP mass measured
in residential land use areas to RWC emissions during selected, worst case
woodsmoke conditions. Twenty-four hour, worst cast impacts exceeding 60
ug/m3 were measured at residential sites in Seattle, Spokane, Portland,
Medford and Boise. A maximum impact of 128 pg/m3 (upper limit estimate)
was recorded at the Boise residential site. Benzo(a)pyrene concentrations
as high as 22 ng/m3—a factor of ten higher than background—were
measured in Spokane.
Results from the ambient monitoring program clearly document the
importance of RWC emissions as Che most important contributor to fine
particle mass concentrations in many Northwest communities during worst
case winter particulate air quality conditions.
iii
-------
LIST OF TABLES
Table Page
1 RWC Ambient Sampling Network ............. 3
2 Program Analytical Design .............. 7
3 XRF Minimum Detectable Concentrations ........ 9
4 Relative Wood Type Use
Data Used to Form 1'*C Fuel Samples .......... 21
5 Carbon-14 Analysis Results for Typical Wood Fuels . . 21
6 CMB Estimates of RWC Impacts (October, 1980-March,
1981)
Fine Particle Fraction ................ 4
7 PNA Analysis Results for Selected RWC Samples .... 26
8 Average PNA Concentration for Residential Sites
(ng/nr) (1980-81 Heating Season) ........... 27
9 Comparison of. RWC Impacts Based on CMB and
Radiocarbon Measurements ............... 28
-------
LIST OF FIGURES
Figure Page
1 Source Contributions, Fine Particle Fraction 25
2 1'*C Validation of CMB Impact Estimates 29
-------
1.0 INTRODUCTION
The increasing use of wood (and coal) as a source of residential heat is a
phenomenon common to many states in the Pacific Northwest. For a variety of
economic and political reasons, the public's interest in the use of wood as an
important heating fuel has led to a resurgence in concerns focusing on wood smoke
visibility reduction, odors, public health risks and the "livability" of many
communities.
The use of wood for residential heating, while aesthetically pleasing,
carries a significant potential for adverse health effects to large segments of
the population. The impact of Residential Wood Combustion (RW.C) emissions is
especially severe because plume impacts typically occur at ground level very near
to the source. In addition, the areas of highest RWC emission density often
coincide with the areas of maximum population density and the majority of the RWC
particulate emissions are within the size range deposited within the lungs. RWC
emissions are relatively rich in-carcinogenic organics, toxic pollutants and
respiratory irritants. For all of these reasons, wood smoke
represents an important problem that is of growing public concern.
The chemical products formed in the combustion of wood have recently been
shown to contain 17 priority pollutants, 14 carcinogenic compounds and 6 toxic or
mucus coagulating agents which, when considered in addition to toxic gaseous
emissions and respiratory irritants, collectively represent an important health
risk. :
RWC emissions are becoming increasingly important as a major contributor to
violations of current particulate air quality standards and are implicated in
issues related to visibility reduction, odors and public health. New attention
being focused on the adoption of an Inhalable Particulate National Ambient Air
Quality Standard has also caused concern about the RWC impact on 24 hour standard
attainment. The continuing economic pressures to expand the use of wood and coal
for residential heating, and the limited regulatory pressures restricting the use
of wood, may cause additional concern about the impact of RWC emissions on public
health, aesthetics and the future "livability" of many communities.
A multi-faceted study of current and projected air quality impacts, fuel use
-------
trends, the current status of wood stove emission testing, RWC control strategy
alternatives and indoor exposure has been conducted in response to the current
concern over the RWC problem. This document reports on Task 1 in this series;
the air quality impact of RWC emissions on ambient air quality. This report
focuses on the impact of particulate emissions, although carbon monoxide,
aldehydes, phenols, volatile organic compounds and oxide of nitrogen emissions
are also of potential concern.
The objective of Task 1 was to estimate the maximum 24 hour impact of RWC
emissions in Portland, Oregon, Seattle and Spokane, Washington during the 1980-81
heating season. Historical samples' collected in Longview and Yakima, Washington,
Boise, Idaho and Medford, Oregon were included to provide a broad-based study
representative of Pacific Northwest communities. Other objectives of the program
were to (a) determine the ambient concentration of polynuclear aromatic
hydrocarbons (?NA) compounds associated with RWC emissions (b) validate RWC
impacts calculated by Chemical Mass Balance methods with estimates obtained by
Carbon-14 measurements and (c) develop a data base upon which future projections
of RWC impacts can be based given fuel use trend information.
2.0 PROGRAM METHODOLOGY
The program design of the RWC Ambient Impact Analysis includes six elements,
each of which are important to the program success. Data analysis methods
focusing on the Chemical Mass Balance (CKB) model and Carbon-14 techniques, the
air sampling program, sample selection, analytical protocol, source
characterization and data interpretation methods. Each -are described in this
section.
2.1 Air Sampling Program
The RWC air sampling program was established and operated to define the
ambient particulate impact of residential wood heating emissions in three
Northwest cities; Portland, Oregon, Spokane and Seattle, Washington. Sampling
was conducted at rural, residential, commercial and industrial land use areas .
listed in Table 1. The sampling network can be subdivided into two groups:
-------
Table 1
RWC Ambient Sampling Network
Site //
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
City Location
•
Spokane Country Homes
Spokane Turubull
Seattle Newport Way
Seattle Lake Samamish
Portland Marcus Whitman School
Portland Cams
Medford County Courthouse*
Seattle South Park (Dallas)*
Tacoma Fire Station 12*
Seattle Georgetown*
Yakima County Courthouse*
Longview City Shops*
Spokane County Health*
Spokane Crown Zellerbach*1
Boise Fairview & Liberty St.*
Notes:
i- Located about 1 mile from an
2- Cyclone Sampler (< 1 urn)
3. Integrating Nephalometer
* Historical Samples
+ Located in near proximity to
Land Use/
Site Type
Residential *
Background
Residential
Background
Residential
Background
Commercial
Residential
Industrial
Industrial
Commercial
Commercial
Commercial
Industrial
Commercial
Aluminum Reduction
Sampling Number of Monitoring Instrument
Period (1981) Samples HV SSI Sierra DiclioL
1/31-3/10
10/30/80-3/1/81
2/6-3/8
3/9/81
1/30-3/5
2/6-3/5
12/17/80- 1/16/81
1/15-1/16
1/4-2/9
10/18/80-2/3/81
11/17/80-2/21/81
12/11/80-1/16/81
10/31/80
1/17-3/11
12/9/80-2/23/81
Plant
15 x
5 x
10 x
1
8 x
6 x
1 x
2 x
4 x
3 x
4 x
4 x
1
4 x
9 x
X X
X X
X X
X X
X X
X X
X X
X
X
X
X
X
X
X
X X
Residential development
-------
Sites 1-6 active during the October, 1980 co March, 1981 period and sites 7-15
from which historical samples were selected.
Sites active during the 1980-81 heating season were operated either on a
daily or every other day schedule on a 24-hour basis. All residential sites were
selected on the basis of the apparent level of wood burning activity as indicated
by visible plumes, wood stove stacks and firewood storage piles.- Final siting
was contingent upon availability of power, security considerations, sampler
exposure and adherence to established siting criteria (40 CRF Part 58) .
Characteristics of specific sites are noted in Appendix 1.
Sampling was planned to concentrate on weekends, when greater use of wood
stoves and fireplaces was assumed likely to occur. Flexibility was, however,
provided to sample continuously during any air stagnation episodes with cold
temperatures. Sampling was planned for the Friday-Monday period, plus a mid-week
day, provided weather conditions were sufficiently cold and stable. Weather
forecasts were monitored and sampling decisions were made on a daily basis. The
following criteria were adopted as guidelines in determining whether or not
sampling would be conducted:
1. Predicted minimum temperature of less than 40°F and preferably less than
35°F
2. Daily average wind speeds less than 5 mph
3. Presence of atmospheric inversions
Background site samples were not planned Co correspond to each sample collected
at each residential site. Instead, 3-6 background samples were selected to
represent a range of background levels. Most background site samples did
correspond to a residential site sample, however, providing a number of days for
each city for which CMS analysis can include comparison of simultaneous in-town
and background samples.
Sampling Protocol
The air sampling protocol was intended to capture particulate samples (TSP,
respirable and inhalable) for subsequent chemical analysis, to estimate
residential wood combustion (RWC) impacts. At each residential and background
site, the following instruments were operated for this study:
-------
1. A standard high volume sampler, to measure total suspended particulate
(TSP), on glass fiber filter media (hi-vol).
2. A standard high volume sampler equipped with a Sierra Model 235 High
Volume Cascade Impactor to collect "respirable" (< 2 microns in
particle diameter) aerosols on glass fiber filters (Sierra).
3. A dicholomous sampler to collect a "respirable" or fine fraction (<2.5 urn)
and a coarse (2.5-15 ym) fraction on teflon filters.
4. A size selective inlet sampler designed to collect inhalable particles
less than 15 urn on glass fiber filters was used at the Boise site (SSI) .
Although samplers from EPA Region X, Che Oregon Department of Environmental
Quality and NEA, INC. were used in the network, most of the sampling
instrumentation required was already in regular use at che background monitoring
sites in the Portland area (at Carus) and Spokane area (at Turnbull) , as part of
the regular particulate monitoring network operated by DEQ and the Spokane County
Air Pollution Control Agency (SCAPCA), respectively. A new background site was
established temporarily in the Seattle area for this study. The dichotomous
sampler deployed there was provided by the State of Washington Department of
Ecology (DOE) Laboratories.
Field checks of instruments for Quality Assurance purposes were performed in
Portland by NEA, in Seattle by DOE staff, and in Spokane by SCAPCA and DOE
personnel. Standard operating procedures were developed and provided to field
technicians. Filters for all instruments were preweighed by NEA, logged and
distributed to field technicians in all three sites. Glass fiber filters were
maintained in cold storage from soon after retrieval until final weighing and
analysis. Following sampling, used filters were logged in and weighed by NEA
staff to determine sample mass, then converted to standard units. Corrections
for fine particle deposition on the coarse mode filter were made to gravimetric
and elemental data.
-------
2.2 Analytical Protocol
The analytical program consisted of gravimetric analysis of aerosol
mass, elemental analysis by x-ray fluorescence, organic/elemental carbon by
thermal combustion, Polynuclear Aromatic Hydrocarbons (PNA) by gas chroma-
tography/mass spectroscopy and Carbon-14 by low level B particle counting.
Each of these methods is described below. The sampling and analytical
techniques used were designed to provide aerosol chemistry information needed
to identify RWC impacts by Chemical Mass Balance and Carbon-14 source
apportionment methods discussed in sections 2.4 and 2.5, respectively. Table 2
summarizes the sampling and analytical design used as well as the intended
purpose of the data collection.
2.2.1 Gravimetric Analysis
Hi-vol Sierra sampler filters used at sites 1-6 were preweighed,
conditioned and final weights determined at NEA, INC. using a Torval balance.
Replicate filter weights were routinely conducted on 12% of the filters to
verify that weights were maintained with ± 10% relative precision. Dichotomous
filters were weighed on a Cann Model 27 electro balance with replicate
weighings of 12% of che filters co also insure that weights were held to
within t 10% relative precision. Filter handling, logging, shipping and
calculations were conducted in accordance with NEA standard operating procedures,
Because of the potential loss of volatile organic hydrocarbons, all hi-vol
and Sierra samples were shipped and stored under refrigeration. .
-------
Table 2
Program Analytical Design
Sampling Equipment
Sample Provided
Analytical
Method*
Provides Data Input to
CMir
Carbon-1A
PNAD
NAAQS
Sierra 235 Impactor
Fine Fraction -
Glass Substrate
Combustion
Dichotomous Impactor
Ili-vol Sampler
Size Selective Inlet
Fine Fraction -
Teflon Substrate
TSP Fraction -
Glass Fiber
IP Fraction -
Glass Fiber
XUF,
NAA6
C-142
GC/MS
GC/MS3
C-14
* Gravimetric analysis completed on all samples
Combustion oxidation for Elementa I/Organic carbon
2 Carbon-1A
J PNA compounds by Gas Chromatography-Mass spectrometry
'' Chemical Mass Balance
5 Polynuclear AromaLic Hydrocarbons
6 X-ray Fluorescence/Neutron Activation Analysis
-------
2.2.2 X-rav Fluorescence (XRF)
Fine particle samples collected on dichotomous filters were analyzed for
trace elements using x-ray fluorescence (XRF). The technique is based on the
fact that atoms within the sample are excited from their ground state to higher
energy levels by x-radiation from an x-ray tube. The excited atoms emit discrete
energy x-rays as they return to their normal energy level. The energy of the
emitted x-rays is characteristic of the emitting element and is used to
qualitatively identify the element while the number of x-rays detected is
proportional to the number of atoms present and provides a quantitative measure
of a specific element's concentration through a direct comparison with standards.
Thirty-three elements are commonly measured by this technique, provided their
minimum detectable concentrations are greater than those noted in Table 3.
Intermethod comparison of XRF and neutron activation analysis for about 7
elements is conducted routinely. A detailed description of potential
interferences, calibration calculations and quality assurance procedures is
documented in NEA standard operating procedures (Appendix 6) .
2.2.3 Neutron Activation Analysis (NAA)
NAA is used (a) to measure the concentration of several elements (Na, Mg)
for which quantitative measurements cannot be obtained by XRF analysis and (b) to
provide intermethod quality assurance comparisons for 7 other elements measured
by both XRF and NAA. The technique requires that the sample be placed in a
neutron flux from a nuclear reactor. Elements within the specimen absorb
neutrons and undergo nuclear transmutations forming radioactive elements which,
in turn, emit radiations characteristic of the specific radionuclides. Since the
number of these radioactive elements formed are proportional to the number of
original stable element atoms -present, measurement of the number of specific
radionuclides radiation is an indirect quantitative measure of the stable element
concentrations. Counting was done with 30% and 17% ORTEC HpGe gamma-ray
spectrometers.
-------
Table 3
X-RAY FLUORESCENCE
MINIMUM DETECTABLE CONCENTRATIONS*
STANDARD ANALYSES-i-
Element
AI
Si
P
s
Cl
s
C*
Ti
V
Cr
«n
Fe
Hi
Cu
Zn
Ga
A3
Se
3r
Rb
Sr
1
Zr
Pd
AS
Cd
In
Sn
Sb
Sa
La.
Kg
?b
I
10
3
10
30
20
6
6
2
1
1
1
2
1
1
1
1
1
1
i_
2
4
4
3
30
3
12
14
13
13
15 ;
100 |
100
4
5
Three sigiaa incerrerence-free MDC for particulaces deuosiced on filters «ich 1 jig/ca-
densities sucn as Nuclepore and some cellulose and :eflon filters XDC '- 'ciae)"*"-
(filter aensicy)**-
Tor scretcned teflon filters
Custom analysis for specific elements at special >35C's available upon request.
-------
2.2.4 Carbon Analysis
Organic and elemental carbon (OC/EC) analysis of a sample aliquot is
conducted by heating samples collected on glass or quartz fiber in an oxygen-
free atmosphere at a temperature of 600CC to oxidize organic carbon to C02
which, in turn, is reduced to CHi, in a Mn02 bed heated to 900°C. The methane
is then measured by a flame ionization detector (FID). Elemental carbon is
then measured by introducing 0% into the sample oven and combusting remaining
carbon to C02 which is also reduced to methane and measured by FID. Artifact
EC formed from pyrolysis of OC is detected (and corrections made) through the
use of an optical filter reflectivity system.
2.2.5 PNA Analysis
Selected Sierra and hi-vol filters were submitted for PNA analysis by
gas chromatography/mass spectroscopy. Each sample was extracted for 8 hours
with dichloromethane followed by removal of polar compounds on an alumina
column and solvent evaporation. Internal standards were added to each sample
prior to GC/MS performed as outlined elsewhere.3
2.2.6 Carbon-14 Analysis
Sample analysis for Carbon-1.4 consists of mounting the particle filter
directly on the inner wall of a combustion tube and converting the sample
carbon to high purity C02 by combustion in an oxygen stream over CuO and
Ag-wool at 800°C and 400°C, respectively. Final purification of the sample
COa is accomplished at low temperature (-78°C) distillation and exposure to
Cu-foil at 800°C. Sample mass is then determined by standard gas volumetric
techniques.
' The L1*C content is determined by directly counting the C02 at 1-4 atm
in a miniature low-level gas proportional counter using massive shielding,
meson cancellation and pure counter construction materials. The 1UC decay
rate is based on 8 particle counting.
10
-------
2.3 Quality Assurance
The project quality assurance program consisted of four elements each of
which are briefly described below:
2.3.1 Ambient Monitor Siting
Each of the ambient monitor sites selected for this study was
reviewed with respect to EPA siting criteria described in 40 CFR 58.
Each of the sites is described in Appendix 1.
2.3.2 Analytical Program
A brief description of quality assurance elements associated
with the analytical program are described in section 2.2 and in greater
detail in Appendix 6. Since elemental analysis is central to the CMB
source apportionment method, the XRF elemental analysis quality assurance
plan is briefly discussed below.
The XRF QA program includes all aspects of the ambient sampling
and filter handling protocols. Special precautions were taken in the
field and in the laboratory to avoid sample contamination, loss of
particles during filter shipment and potential for filter identification
problems. Clean room techniques, for example, were used during loading
(and unloading) filters from their holders and specially designed sample
shipment containers were used to minimize the potential for particle
losses due to sample tipping. During XRF analysis, quality control
standards and blank samples were analyzed for each group of 10 samples
analyzed and results were compared to known standard values to maintain
precision to within ± 2%. Sample groups outside of this range were
reanalyzed. Analytical results for several elements (K, Ca, Fe, As,
Br and Pb) were intercompared between excitation conditions and calibration
standards have been verified by extensive interlaboratory and intermethod
analysis.
Similar procedures have been followed with respect to organic and
elemental carbon analysis results to insure that analytical precision and
accuracy is held to within i 15% relative to the sample concentration.
Both internal quality control (replicate analysis of a "standard" filter)
and external standards (filters containing known amounts of carbon) were
routinely analyzed.
11
-------
Gravimetric analysis of the ambient samples included replicate
analysis (12% audit level) of the filters to maintain weighing pre-
cision and consistency of tare and balance calibration checks during
each weighing set to sithin ± 0.004 mg of true value.
2.3.3 Field Monitoring Program
Standard operating procedures were developed for Hi-vol and dichotomous
sampler calibration, maintenance, filter handling and flow audits to insure
consistency within the data set. Each procedure is described in Appendix
6. In addition, notes were maintained to reflect meteorological
circumstances in the near vicinity of the sampling site. These notes
are included in Appendix 3.
2.3.4 Data Management
Two levels of data validation QA were maintained; (a) verification
of the transcribed data against the source documents and (b) examination
and resolution, where possible, of inconsistencies in the data sets.
For example, bivariate plots of elemental measurements (Pb/Br, Al/Si,
etc) were made prior to CMB analysis of the data to identify and
validate outliers.
12
-------
2.4 Sample Selection
Samples submitted for chemical analysis were selected on the basis of (a)
fine fraction (< 2.5 ym) mass loading, (b) light scattering (Bscat) average
where available, (c) relative meteorological stability (wind speed and air
stagnation advisories) and (d) daily minimum temperatures.
The above criteria were used to estimate the probable relative RWC impacts
represented by sample filters. As noted earlier, filters likely to reflect
maximum impacts were selected. Other criteria which influenced selection and
ranking of filters recommended for analysis included:
1. Completeness of data for best CMS analysis
Ideally, respirable particulate samples for both inorganic and carbon
analysis are desirable at both the urban monitoring site, and a suit-
able background site during the same 24 hour sampling period. However,
this ideal situation devotes considerable resources to analysis of
background samples, and was often not possible, due to lack of (a)
respirable particle size samples, (b) background samples taken on same
day (c) pairs of samples suitable for organic and inorganic analysis.
2. Available Background Samples
Analysis of enough background samples to describe a range of potential
background RWC impact on days of expected high RWC impact at urban
sites.
3. Preference for Analysis of Respirable Particulate Samples
RWC emissions are assumed to be almost entirely in the respirable size
range (_<_ 2.5 microns). Accordingly, preference was given to inorganic
analysis of dichotomous sampler fine fraction filters and companion
carbon analysis of Sierra filters (< 2 urn). If, however, a SSI sample
(< 15 urn) was used for carbon analysis, companion elemental analysis
was completed on both dichotomous fine and coarse filters and summed
together to obtain comparable sets of data for an inhalable size fraction.
4. Consistency of Mass Loadings on Co-located Samplers
In some cases, (Boise) samples with possible high RWC impacts were not
selected for analysis because of apparent discrepancies in mass
loadings on filters from co-located samplers, e.g., (a) too large a
difference in dichotomous fine + coarse mass and hi-vol (SSI) mass, or;
(b) standard hi-vol TSP mass loadings significantly lower than
respirable or inhalable samples taken during the same 24 hour period at
the same site.
5. Availability of Recommended Filters for Analysis
Availability of both the urban and background samples for analysis was
a prerequisite to sample selection.
The process of selecting filters for analysis was proceeded by first
assembling available data on the mass loadings of respirable, inhalable and
total suspended particulate (TS?) samples collected during the October, 1980-
March, 1981 period. Since dichotomous samplers were operating in a number of
13
-------
cities in Oregon, Washington and Idaho, it was feasible to concentrate on
respirable particle size samples in the following cities:
Washington
Seattle
Spokane
Yakima
Longview
Idaho
Boise
Lack of dichotomous samplers, and other considerations, ruled out selection
of filters from some cities of potential interest (i.e., Olympia, Washington).
Second, samples associated with higher 24 hour average Bscat values were
assumed to correspond to higher RWC impacts. Peak Bscat values which occurred
in the evening and early morning hours also were assumed to indicate potential
RWC activity. The third consideration involved selecting samples associated
with maximum atmospheric stability and cold weather, as represented by low
surface wind speeds and low daily minimum temperatures, respectively. The
fourth and primary criteria for selection were the fine particle mass
concentrations.
Eighteen Sierra and SSI -samples from five Northwest cities were selected
for PAN analysis to (a) indicate potential human exposure to hazardous
chemicals and (b) to determine if PAN concentrations were correlated with RWC
impacts determined by CMS and Carbon-14 analysis. In addition, 13 samples were
selected for Carbon-14 analysis to provide a second, independent estimate of
RWC impact. Both sets of samples were selected in accordance with the criteria
described above.
3.0 SOURCE APPORTIONMENT
Identification of RWC particulate impacts requires the application of
methods capable of apportioning aerosol mass among possible sources. Chemical
receptor models have been selected for use in this study because these methods
do not rely on emission inventory or meteorological data as a basis ;or
14
-------
quantification of source impacts. Two independent, yet complementary
receptor model approaches have been included in this program design. Each
are discussed in the following sections.
3.1 Chemical Mass Balance
The chemical mass balance model matches source particle size and chemical
"fingerprints" to those measured at the receptor to back-calculate the impact
of specific source classes of similar chemical composition. Given
data on ambient concentrations of several chemical species and the percent by
weight of those species in the source emissions, a set of equations is
prepared and solved to determine source impacts.
3.1.1 Chemical Mass Balance Receptor Model
The starting point for the Chemical mass balance model is essentially
the same as the source (dispersion) model. The source model states that the
contribution of a source to a receptor is dependent upon the emission rate
and a dispersion factor. Limitations of dispersion modeling, however, arise
from difficulty in quantifying dispersion factors in areas of complex terrain
and/or meteorology, from difficulties in identifying emission rates, and other
operating parameters associated with emission sources. The advantage of the
CMS, and other receptor approaches, is that an exact knowledge of these
parameters is unnecessary.
The basic assumptions inherent in the CMS approach are the same as those
associated with other receptor and source (dispersion) models:
The particulate mass measured at the receptor is a direct, linear
sum of the mass contribution from a number of specific sources
P
m = £ M. . (1)
j-1 J
where
m = measured concentrations of particles from all "p" sources (ug/nT)
M. = the impact at the receptor of source "j" (yg/m")
15
-------
Similarly, the concentration of a given chemical species is assumed to be the
sum of the contributions from each individual source emitting that species.
C'. = E F. . ' X.. M. (2)
>• . 13 ^ 3
where
C'. = measured concentration of species "i" from all sources
"p" (yg/m3)
F.. = the fraction of species "i" in the emissions of source
"j" as measured at the source
X.. = the coefficient of fractionation of species "i" in the
emissions of source "j". Fractionation is a systematic
change in relative species abundances as the aerosol
ages due, for example, to gravitational settling or
condensation of volatile organic hydrocarbons
M. - the impact at the receptor of source "j" (yg/m3)
The F..' values associated with major emission sources are measured or
adapted from literature values, and the C. data are obtained through field
monitoring programs. One can ensure a fractionation coefficient which approaches
unity by (a) using stack, gas dilution techniques to ensure capture of condensible
•
aerosols and (b) obtaining size-resolved source composition and ambient aerosol
measurements. By conducting fine and coarse mode CMS calculations, source
composition (F..') errors due to gravitational settling of large particles can
be minimized. If no further fractionation or source emission composition
changes are assumed to occur,, then X.. is assigned a value of unity and F..'
can be considered to remain unchanged in time and distance downwind when
measured at the receptor (F..' = F..).
16
-------
By dividing both sides of equation (2) by the particulate mass
collected at the receptor, all of the variables in the equation are
expressed as a ratio to the aerosol mass:
p
C. - Z F.. S. (i=l to n) (3)
1 1 = 1 13 J
where
C. = concentration of species "i" expressed as a ratio to
P
mass (i.e., C± - C'±/l M.)
F.. " the fraction of species "i" in the emissions from source
"j" at the receptor, expressed as ratio to mass
S. = the impact at the receptor of source "j" expressed as a
ratio to mass emitted from all sources impacting the
P
receptor(i.e., S . = M./Z M.)
n = the number of chemical species considered
p = the number of sources impacting the receptor.
Since S. is the quantity to be determined, a set of simultaneous equations
can be written for all "p" sources of species "i", where "i" is defined as a
"fitting" element. Once the set of equations are formed, the value of S. can
be solved by standard matrix manipulation techniques using linear least squares
or effective variance solutions utilizing commonly available matrix mathematics.
The effective variance least squares fitting method used by NEA, INC. in
this study weights the solution in accordance with uncertainties in both the
source composition (F..) and the ambient aerosol species (C.). The procedure
ensures that those components with the greatest uncertainty receive less weight
in the fit and that both sources of uncertainty are propagated through to the
source impact (S.) determination.2
Uncertainties in the source impact estimates are calculated by an error
propagation algorithm incorporating (a) the imprecision in the ambient aerosol
chemical species (C.) analysis and (b) the variability in the source composition
species (F..) at the one sigma level.
17
-------
3.1.2 Source Characterization
The data sets referred to below and included in the Appendix have
been developed by comprehensive analysis of representative source test
samples, or adapted from the literature cited. In all cases, source
emissions have been collected using size segregated (fine, coarse) sampling
systems. Hogged fuel boiler and residential wood combustion samples were
taken with a size segregating source testing procedure for aerosols. Following
sample collection, the filters were analyzed for organic and elemental
-2 - -
carbon content, ions (SO, , NO , Cl , p ) and trace metals. Source emissions
are then characterized in terms of weight percent. Further detail on the
sampling protocol may be found in the references cited. The source signature
described in this section forms an essential element of the CMB source
apportionment analysis. References to the origin of the emission chemistry
used are noted below.
18
-------
Mneumonic
RDUST
MARIS
CDUST
KRAFT
Source Znission Chemistry Origin Reference
Geologic Local soil samples were collecced in the
vicinity of the Boise sice (5 samples),
Spokane-Country Hones (3 samples) and
Seattle-Newport Hay (4 sices) . Following
laboratory processing, resuspension and
preliminary XRF analysis, composite soil
chemistries vere formed for comprehensive
XRT analysis. City specific soil chemistry
was Chen used in association with samples
from each community. Although Che uneumoaic
RDUST is used in Che CMS calculations, Che
soil chemistry used cannot distinguish
between native soils, paved road dust,
unpaved road dust or other sources of
similar aerosol chemistry.
.Medford Aerosol Characterization Scudy i
Medford Aerosol Characterization Scudy
Bedford Aerosol Characterization Scudy
Portland Aerosol Characterization Scudy
Portland Aerosol Characterization Scudy
Based on excess sulfur calculated as sulfate
Portland Aerosol Characterization Scudy
Based on excess zinc
operacions (or
other zinc
rich source)
Marine Portland Aerosol Characterization Scudy
Aerosol
Continental Portland Aerosol Characterization Scudy
Ousc
SESWD
TRAKS
HOCFU
ALUMR
SDOIL
SECSO
GALVN
Residential
'flood
Comouscion
Transoor-
tation
Hog cuel
Boiler
Aluminum
Reduction
Residual
Oil
Combustion
Secondary
Sulfate
Galvanizing
Kraft
Recovery
Furnace
Portland Aerosol Characterization Study
19
-------
3.2 Carbon-14
Measurements of radioisotope carbon-14/carbon-12-ratios have been used by
Cooper and Curie to distinguish between "modern" and fossil fuel-derived
carbonaceous aerosol, providing an important new method of RWC impact assessment.
The method's ability to resolve these two classes of sources is based on the fact
that the isotopic ratio of lllC/12C in plant and animal tissue is in equilibrium
with the same ratio in atmospheric COa, whereas fossil carbon contains
essentially no radiocarbon (1J*C) because of its old age compared to the half
life of the 1I4C isotope (5730 years). Consequently, carbon 14 is a unique
tracer of carbonaceous aerosol from contemporary carbon sources (such as wood
combustion) and can be used to apportion the carbon content of the aerosol.
Application to RWC impact analysis requires knowledge of (a) ll*C/12C
ratio in the fuel being burned, (b) contemporary carbon source activities
impacting the receptor (i.e., structural fires, forest fires, open burning of
residential debris, incineration) and (c) the carbon content of RWC emissions.
Samples submitted for analysis must be size fractionated to minimize inter-
ference from large particle contemporary carbon sources, such as.leaf fragments.
Each of the three requirements are addressed below:
3.2.1 Fuel Ratio Measurements
Interpretation of ambient carbon 1UC/12C ratios requires a knowledge
of the isotope ratio in the fuel burned in the community. If the ratio is
significantly greater than 100%, corrections in the assumptions are required.
Four samples of typical wood fuel burned in Northwest communities were
analyzed by the ll(C/12C method. Fuel samples were formed following completion of
a telephone survey of wood suppliers, U.S. Forest Service and university
extension agents in Boise, Seattle, Portland and Spokane. Representative logs of
each of the species were obtained from each city and composite source samples
formed based on their relative use shown in Table 4. Cuts were made
perpendicular to the concentric growth rings to insure that sawdust samples
were integrated in correct proportion to the age of the "fixed" carbon contained
in the logs.
-------
Table 4
Relative Wood Type Use
Data Used to Form 1UC Fuel Samples
City
Percent in
Species Composite Sample
Seattle Alder 70%
Douglas Fir 30%
Portland Alder 20%
Douglas Fir . 60%
Hemlock 20%
Spokane Douglas Fir 30%
Western Larch 40%
Ponderosa Pine 15%
Lodge Pole Pine 15%
Boise Douglas Fir 40%
Ponderosa Pine 30%
Lodge Pole Pine 30%
The results of the 1 C analysis of these samples are listed below in
terms of percent of modern carbon and are in very close agreement to previously
7
published values (109%) measured relative to international radiocarbon standard.
The Seattle result (133 ± 2%) is substantially greater than the mean of the other
three samples (111 ± 4.6). No explanation for this discrepancy was found.
Table 5
Carbon-14 Analysis Results
For Typical Wood Fuels
Airshed % Modern Carbon
Portland . 115 t 1*
Boise 106 ± 1
Spokane 112 ± 1
Seattle 133 ± 2
Mean 116.5 ± 11.6%
* Analytical errors exclusive of sampling uncertainty
21
-------
As noted in section 1.0, the objective of this program was to quantify
the maximum 24 hour impact of RWC emissions in Pacific Northwest communities
during the 1980-81 heating season. Results presented in this document,
however, largely represent air quality impacts measured during the month
of February, 1981—a period during which the average heating load demand was
less than that which typically occurs during the period of peak residential
space heating demand (November, December and January). Data reported herein
therefore represents high impact days selected from samples collected during
a period of moderate weather conditions. Higher impacts may occur during
periods of greater space heating demand and poorer atmospheric dispersion.
Program results for Medford, Seattle (South Park and Georgetown),
Spokane (County Health) and other sites where 4 or less samples were analyzed,
cannot provide a comprehensive understanding of the actual magnitude of RWC
emission impacts at these sites. Instead, results for these communities
should be viewed as preliminary estimates of likely worst case impacts
representative of the sampling period.
Table 6 summarizes, by site, CMS estimates of RWC air quality impacts
measured (a) during the field monitoring program and (b) from selected
historical samples. Figure 1 illustrates average source contributions to the
fine particle mass in Spokane, Seattle and Portland, as well as inhalable
mass fraction in Boise. Table 7 includes results of Polynuclear Aromatic
Hydrocarbon (PNA) analysis for selected samples from five Northwest cities,
and Table 8 lists average TSP, fine mass and RWC impacts for concurrent
samples (only) by site. Results of the carbon-14 analysis are presented in
Table 9 and Figure 2.
5.0 DISCUSSION OF RESULTS
The Chemical Mass Balance results reported herein are often associated
with relatively high levels of uncertainty based on (a) imprecision in
analytical measurement of ambient samples and (b) uncertainty in the RWC
emission chemistry reported in Appendix 2. Since the RWC source composition
values in this analysis were developed during the Medford Aerosol Character-
ization Study, the impact estimates reported assume a fireplace/wood stove
ratio similar to Medford (0.63).
23
-------
Table 6
CMB Estimates of HWC Impacts
(October. 1980-Marcli, \98l)
Flue Particle Fraction
g,.i,l,,aio,.
Si>ukuiic
bt*uk.«ii«*
ii^tok U IIV *
b u U 1 I \ «
Scull U*
Su.ilt le«
TtM I.H..I
1 ilk l.i.a"
l.uiim/leu'
1 Jjl.l.
du 1 »•-••"
Ur«n"".
I'Ot 1 t dllj
Hcil lord*
Sice
Ho.
1
13
14
3
8
10
9
1 1
12
15
5
7
No. of
S a mp lea
15
1
d
10
2
3
<,
t,
i.
y
8
!•
Averagi HWC
Purctnt of
Site Untie' Vine Ha*t
Country
llouea (R)
County
llaalth (C)
Crown
Zellerbach (1)
Newport Way (»)
South Park (R)
Georgetown (I)
Fire
Station 112 (I)
Courthouse (C)
City Shops (C)
Folrvlew 4
Liberty St. (C)
81
64
45
8}
65
73
74
93
61
69
Whitman Sclioo 1(R) 6B
Courthouse (R)
.0
.7
.3
.B
.8
.4
.9
. 1
.4
.5
.5
-
Average
Flue Hitm
55.
53.
37.
36.
-
39.
47.
53.
4 1 .
121,
40
0
1
0
1
7
0
8
,8
,8
.0
t <|||>.|C t
44.5 i II .1
j<. J t 21 . J
16.7 t 14.9
30.2 i U.t.
-
29 . 1 : 20. 2
35.2 i 25. U
50.0 i 34. y
25.6 t 1 7. «
64 . 7 1 22. J1
il .'< i 10.3
-
.'•, II. Mil
H.ixlii.iim
6«. 1 1 10.0
19.1 1 1 1 . J
', I) li : 1 J . a
0(1.2 •- 19.9
J5 5 : 24 3
'. <. . i : 3 J . 2
j i . 1 : il .1,
-,(> •> '. 1:. 1
U'7 9 ! <"J.9' '
-------
Figure 1
Average Ik Hour Worst Case
Source Contributions
-Fine Particle Fraction-
(October, 1980-March, 1981)
RESIDENTIAL WOOD COHDU3T10H
ALUMINUM PROC. (0.4X)
RESIDUAL OIL <0.03X>
TRAHBPORTATIOH <3X>
GEOLOGICAL
SECONDARY 8ULFATE t3.6X>
UNEXPLAINED (9 .7K>
SPOKANE COUHTHY HOMES 3ITE I
*E310£NT.AL WOOO CUnUUST.O^ RESJOUftL 01L <„_„,
TRANSPORTATION (5 BX. )
8ECOIIOAKY SULFATE ( 7*. )
MARINE < i . i>. >
GEOLOGICAL ( I . y>. y
lLE lll;M|'l)K1 UfiY
(Ui:l 1 evuo )
KE6IOEMTIAL WOOD COKQUSTIOH
(GBX)
RESJDUftL OIL <0.6X>
HAR1HE (0.6K)
GEOLOGICAL ( I .8>:>
TRANSPORTATION (H. IX >
8ECOHO(\RY 8ULFATE (« 2X )
UHEXPLAIMEO < I 8 . 7 >i )
RESIDENT \M.
UOOO
COrtOUSTI OH
( 70>. )
\
< H
TRAHSPORTftI I Oil (3 ?/: )
SECOHOARY SULFnlt ( .! I
MARINE (0.0^.)
IJIIEXPLAIHCO (0 (,'<)
PORTLAND MAHCU9 WIUTHAH
I) 0 I 'J t I II I W >.' I I l<
I I 11 i k I
-------
Table 7
PNA Analysis Results
For Selected RUC Samples
(ng/m1)
City
Seiilt le
Se.ilt In
Seat I It
Sesit L le
SJP.JI 1 In
Sual 1 le
Spokane
Spok.me
Spokdite
Spok.me
Port 1 .in
Purl l,n,
I'ort l.iii
F'orl Ifin
llolse
llolse
Medlonl
Hcdtord
Sltet
Newport Way (R)
Newport Way (H)
1 ' Newport Way (R)
South I'ark (It)
South Park (R)
1 dke Samamlsh (fl)
Country Home (R)
Country Home (R)
Country Home (R)
Tvirnbull (B)
l Marcus Whitman Sell (R)
.1 Carus (B)
t'alrvlew & Liberty (C)
Falrvlew & Liberty (C)
Courthouse (C)
Courthouse (C)
* Sleira Model 235 linpactor
** Size Selective Inlet
Date
2/6
2/7
2/8
1/15
1/16
J/9
2/1
2/6
3/1
2/6
1/31
•ill
3/5
2/6
1/8
12/12
12/17
1/16
Minimum Detectlble Concent rat Ions
1 Slli
/ [
; key
i
Sampler
S*
S
S
111 Vol
111 Vol
S
S
S
S
S
S
S
S
S
SSI**
SSI
S
S
benz(a)
anthracene
8.9
5.2
0.6
48.2
7.B
--
38.6
50.7
8.6
0. 1
13.6
43.9
14.3
0.4
1 .3
0.9
27.6
22.5
: f luoranthene, pyre
benzof luoranthenes
benzo(ghl)perylene
benzo-
f luoranthenes
3.0
9.6
0.3
13.6
1.8
0.8
18. 1
6.0
13.2
0.3
3.2
1 1.6
5.4
1.5
3.6
1.6
12.4
8.3
ne, benz(a)anthrace
t benzo(a) pyrene
, dlbenzanthracene
benzo(a) fluor-
pyrene anthene
1.8
4.2
0.3
6.5
1.0
0.3
11.1
3.6
3.7
--
1.5
5.4
2.7
0.5
2.3
0.8
8.5
8.3
ne. )
1
3
0
11
3
0
42
39
5
0
2
23
2
0
0
0
12
10
0.05 ng/m
0.1 ng/m
.9
.0
.5
.9
.2
. 1
.2
.0
.4
.3
.8
.6
.9
.4
.7
.3
.4
.4
i
i
dlbenz-
pyrene anthracenes
2.
5.
1 .
16.
4.
0.
69.
45.
9.
0.
3.
27.
3.
0.
0.
0.
15.
9.
9
3 0.7
0
2 0.9
7
1
9 1 .0
0 0.2
3 1.4
4
5
9 0.2
9
6
8
3
9
0
benzo(ghl)
perylene
0.9
3.3
0.2
7.1
1.2
0.3
4.5
1 .8
6.6
—
1 .3
2.9
2.1
0.7
1.5
2.3
3.4
3.1
(It) Residential (1) Industrial
I l.or.i led In llellevue. Wash Ing ton
-------
Table 8
City/Site*
No. of
Obs.
TSP, Fine Particle and RWC
Impact Averages
(Concurrent Samples, Only)
TSP Mass
Mean ± Std. Dev.
(lag/m3)
Fine Mass
Mean ± Std. Dev.
(pg/m3)
RWC Impact
Mean i Std. Dev. % of TSP
(ug/m3) Mass
Spokane
Country Home(R) 13
Crown Zellerbach(I) 3
118.9 ± 50.9
232.0 ± 156.9
53.0 ± 17.7
16.4 ± 0.5
44.8 ± 14.9
37.6 ± 2.4
36.5
11.1
Seattle
Newport Way(R)
71.7 ± 22.8
33.8 ± 10.5
29.4 ± 11.0
42.0
Tacoma
Fire Station 12(1) 4 159.0 ± 56.7 47.0 t 5.5
Portland
Whitman School(R) 8 62.1 ± 19.7 43.3 ± 15.1
35.2 t 9.7
27.3 ± 10.3
29.7
43.9
Yakima
Courthouse(C) 3
221.6 ± 87.1
55.3 ± 12.4
50.6 ± 4.0
25.6
Boise
Fairview & Liberty(C)9 148.7 ± 59.3 121.6 ± 43.5** 84.6 ± 26.9*-'
60.4
* Other city sites have less than 3 concurrent observations
** Inhalable particulate mass (< 15pm)
C = Commerical I = Industrial R = Residential
-------
Table 9
Comparison of KWC Impact Estimates
Based on CMB and Radiocarbon Measurements
City/Site
Spokane
Count ry
Homes
Seattle
Newport
U.iy
South
Park
Portland
Marcus
Whi tman
Boise
Date
3-1-81
2-6-81
2-1-81
2-6-81
2-8-81
2-7-81
1-16-81
1-15-81
3-5-81
1-31-81
2-7-81
Z Modern
Carbon in
uuod (B)
112
112
112
133
133
133
133
133
115
115
115
I Modern
Carbon in
Aerosol (A)
97
107
118
96
100
63
53
96
97
90
Z RHC
(C-A/B)
86.
95.
(100)
72.
75.
47.
39.
83.
84.
78.
6
5
2
2
4
8
5
3
3
Total
Carbon
(l'g/m')(D)
40
34
48
24
35
53
52
32
35
44
.1
.8
.3
.2
.4
.7
.4
.5
.9
.9
RUC Carbon RUC 12)G
'"C/CMB
Ratio
0.99
0.96
1.18
0.75
0.91
0.62
0.53
0.92
0.92
0.95
Falrvlew
& l.lber
Medford
County
house
ty 1-8-81
12-12-80
Cour t-
12-17-80
1-16-81
106
106
111
111
64
24
104
114
60.
22.
93.
(100)
4
6
7
54
94
45
44
.5
.1
.8
.1
32
21
42
44
.9
.3
.9
.1
54.9
35.4
71.5
73.5
79
123
53
62
t 21
1 33
i 19
i 19
0.69
0.29
1.35
t .18
Notes
(B) Calculated In relation to International radiocarbon standard. Values are greater tlian IOOZ because of '""C generated during nuclear atmoRplier Ic
test Ing.
(A) X of modern carbon In atmospheric C02. adjusted to age of wood burned.
(C) 7. of modern carbon from RWC sources. 100* is an upper limit.
(I)) Measured total carbon concentration (sura of elemental, organic carbon).
(K) Concentration of TWC carbon (|ig/in ).
(K) RUC impact based on '""C, corrected lor associated II. 0. N.
(f;) ONI! 'estimates from filters of uncertain total sample volume.
-------
Figure 2
1'*C Validation of CHB Impact Estimates (yg/m3)
(with associated uncertainty error bars)
90
80
70
4 60 -
50
(i)
(2)
:o
_L
20 30 40 50 oO
IUC RWC Imoacc (ug/m3)
''CMS sstiaaces from filters of uncertain :ocal volume
(1) Seattle Soucn ?ark 16 Jan. 1981
(2) Seattle Soucn ?ark 15 Jan. 1981
70
30
90
29
-------
Although the fuel use survey (task 2B) results indicate different
fireplace/wood stove emission ratios for Portland, Seattle and Spokane
(0.45, 1.94 and 1.77, respectively), an insufficient body of source emission
chemistry for fireplaces and wood stoves is available to clearly differentiate
between these two sources. Apportioning impacts between fireplace and wood
stove emissions is further complicated by the high degeee of variability
in the composition of the emissions. Given the level of effort that had
previously been committed to development of the Medford RWC emission
composition data, the difficulties noted above and the presence of
independent validation of CMB-derived RWC impacts during the Medford Aerosol
Study (based on C-14 results), further attempt to modify the Medford emission
chemistry data were not attempted.
RWC impact estimates derived by the CMB methods have been compared to
independent measurements made by radiocarbon techniques. Impact estimates
obtained from these two methods (Table 9) show relatively good agreement
for samples collected at most sites. Discrepancies noted at the Seattle
South Park and Boise Fairview and Liberty locations may be caused by.a
number of factors including (a) deviations from the fuel use mix assumption,
(b) the presence of important contemporary carbon sources other than
residential wood combustion or (c) errors in the CMB calculations. Each are
discussed more fully below.
A. Wood Fuel Mix
Since the 1UC estimates are based on an average age (ll*C content)
of firewood in each community, deviations from the wood specie
weightings assumed for each city (Table 4) will result in an error
in the percent of modern carbon in the ambient aerosol. For example,
if residents in the near vicinity of the Seattle South Park site
burned a lesser proportion of alder to Douglas fir than estimated on
a city-wide basis, the radiocarbon estimates would be higher than
estimated. If the mean percent of modern carbon in Seattle fuels is
assumed to be nearer to the average value for other Pacific Northwest
(116%), RWC estimated by lkC would increase by about 13%.
B. Other Contemporary Carbon Sources
The radiocarbon method quantifies the total amount of contemporary
carbon in the ambient aerosol whereas the CMB estimates of RWC impact
30
-------
are calculated on the basis of -numerous aerosol components. An
exact match between estimates developed by both methods requires that
essentially all of the contemporary carbon measured by the '^C
technique be associated with RWC sources—an assumption that may not
be valid where (a) the upper size cut of the ambient sampling system
allows capture of carbonaceous particles of large aerodynamic diameter
(> 2.5 urn) such as plant fibers or (b) industrial sources utilizing
wood waste (e.g., Hogged fuel boilers) are major carbon emission
sources.
C. Errors in CM3 Calculations
Close examination of the CMS calculations for several of the
sampling sites indicates the potential for inclusion of distillate
oil combustion impacts in the fitting process. Calculations for the
South Park site (site 8, Appendix 5) indicates that, by reason of the
underestimate of sulfur, distillate oil may be an important component.
If so, it would contribute to a 30% overestimation of the RWC impact.
The rationale for excluding distillate oil from Che CMS fitting process
is explained below.
Site average RWC impacts ranged from 16.7 (Spokane-Crown Zellerbach) to
83.7 ug/m3 (Boise-Fairview & Liberty St.) and represented from 20% (Medford)
to 93% (Yakima) of the fine particle mass. The highest 24 hour impact
(128 ug/m3) was found at the Boise site. The lowest average impacts (16-30 ug/mi>)
were measured at the background and industrial sites (Tacoma, excepted) while
residential sites varied from 30 to 45 ug/m3.
RWC impact estimates obtained for Boise are significantly higher than those
obtained in other Northwest cities. The maximum impact measured in Boise
occurred on January 18, 1981 at the Fairview & Liberty site, 127.9 ± 29 .'9 ug/m3,
8 hour average. These source impact estimates are based on inhalable particulate
samples captured on Teflon and glass fiber filters ra.ther than fine particle
(< 2.5 um) samples. As a result, CMS estimated RWC impacts must be considered
to be upper limit values subject to considerable uncertainty. The uncertainty
is associated with (a) the RWC source composition which is based on measurements
of fine particle emissions (b) RWC impact estimates are highly dependent on
measurements of K, Cl, elemental and organic carbon, components which in the
size fraction greater than 2.5 urn, can become associated with sources other
than wood burning emissions.
31
-------
Discrepancies between SSI and dichotomous sampler mass measurements
reported for co-located, concurrent samples further complicates data
interpretation. For these reasons, further verification of the Boise
results is recommended.
The maximum industrial source impacts, while minimal, were noted at the
Longview site (Kraft recovery furnace and aluminum reduction), Tacoma Fire
Station 12 (aluminum reduction) and the Spokane-Crown Zellerbach site (Hog
fuel boiler and aluminum reduction emissions).
In addition, the chemistry of a large number of samples could be explained
by inclusion of a distillate oil source rather than a secondary sulfate source.
This would tend to reduce the impact of RWC sources by about 30%. However,
since the quality of the CMB fits were (a) quite similar in either case and
(b) distillate oil combustion impacts at the background sites were similar to
residential sites, distillate oil combustion was not routinely included as a
source. It is likely, however, that oil combustion associated with residential
space heating is an important contributing source to which a portion of the
measured sulfur should be assigned.
The Chemical Mass Balance calculation performed successfully apportioned
greater than 90% of the measured mass with reasonable agreement between measured
and model-predicted element concentrations. Occasional overpredictions of the
mass, however, occur due to uncertainty in the gravimetric analysis or in the
source assignments. In most cases, the calculated and measured mass values are
within the estimated mass uncertainty.
Analysis of selected samples for PNA compounds was completed because of
concern about risks to public health associated wich these potentially-
carcinogenic compounds. Review of RNA analysis results associated with this
study suggests that mean concentrations within residential areas are an order
of magnitude higher than similar measurements at background sites. Table 10
presents a summary of average PNA concentrations from residential sites during
the 1980-81 heating season. Since the samples selected for analysis were
intended to reflect worst case RWC impacts, the values reported do not indicate
typical conditions.
32
-------
Table 10
Average PNA Concentration
For Residential Sites (ng/m3)*
(1980-'81 Heating Season)
Arithmetic
Mean
19
7
4
10
14
0
2
.5
.4
.1
.6
.4
.3
,7
Standard
Deviation
17
5
3
13
18
0
1
.3
.2
.1
.2
.9
.4
.9
Maximum
Value
50
18
11
42
69
1
6
.7
.1
.1
.2
.9
.4
.6
Arithmetic
Minimum Mean
Value Background
0
0
0
0
0
0
0
.6
.3**
.3**
.3**
.3**
.05
.2
0
0
0
0
0
0
0
.2
.9
.4
.3
.4
.05
.3
PAH Group
Benzo(a)Anthracene
Benzofluoranthenes
Benzo(a)Pyrene
Fluoranthene
Pyrene
Dibenzanthracenes
Benzo(ghi)perylene
* Average of 15 selected 24 hour samples
** One-half minimum detectable concentration
PNA measurements made during this study can be compared to annual average
measurements of B(a)P during the 1960 "s in Portland, Eugene and Medford which
ranged from 2.3 to 4.8 ng/m3.6 Quarterly average maximum values of 8.2 ng/m3
were recorded in Medford in 1968. Although annual and quarterly average B(a)P
concentration data are generally not available after 1970, periodic data from
other Northwest locations (i.e., Seattle) suggests that concentrations may
7, a
have decreased during the 1970-76 period and then increased following 1977.
Routine PNA concentration measurements are no longer conducted by EPA.
The average B(a)P concentrations noted in Table 10 are similar to the annual
and quarterly mean values discussed above, but direct comparison is confounded
by differences in averaging times.
There are no published studies available which adequately compare the
carcinogenic activity of all ten PNA compounds under similar experimental
conditions. Similarly, there is no data available concerning human responses
33
-------
to individual PNA compounds, since environmental exposures to PNA invariably
involve contact with complex and usually undefined PNA mixtures. Therefore,
an assessment of the cancer risk related to PNA compounds emitted from RWC
sources can only be considered when placed in perspective with other sources
of exposure. Conversion, however, of ambient PNA concentrations to in-
halation intake requires consideration of the volume of air breathed per
day (15 m3 for a 10 year old child and 23 m3 for an adult male) and a 24
hour average assumed concentration of B(a)P.9 If an average daily volume
of 19 in3 and an average B(a)P concentration of 4.1 ng/m3 is assumed, a
daily dosage rate of 0.078 yg/day is obtained. This compares to an estimated
total B(a)P intake of 0.16-1.6 yg/day associated with food consumption. As
a further comparison, Bridbord10 estimates that smoking 20 cigarettes per day
provides an exposure of 0.4 ug/day of B(a)P, suggesting that the average
B(a)P concentration identified in this study is equivalent to dosage
received by smoking about 4 cigarettes/day. At the maximum B(a)P concen-
tration measured (11.1 ng/m3, 24 hour average), the dosage is equivalent to
11 cigarettes/day.
The impact of RWC emissions on total suspended particulate (TSP) air
quality was found to range from 11% in industrial areas (Spokane—Crown
Zellerbach site) to nearly 44% in residential areas (Portland—Marcus Whitman
School). In Boise, the RWC impact, calculated on the basis of the inhalable
particulate fraction, may contribute as much as 60% to the TSP mass during
worst case RWC impact conditions. Although this data clearly documents
the importance of wood combustion emissions to TSP air quality under the
conditions of this study, direct inference as to the importance of RWC
emissions on TSP air quality standard non-attainment is difficult because
(a) the RWC impacts cannot be calculated in terms of an annual geometric
mean averaging period and (b) the 24 hour worst case RWC impacts included
in this study may not represent the 24 hour worst cast TSP concentrations
upon which standard attainment is based.
34
-------
6.0 CONCLUSIONS
Ambient air quality studies conducted during the 1980-81 space heating
season in eight Pacific Northwest communities clearly indicate that
residential wood combustion (RWC) emissions, as estimated by validated
chemical mass balance analysis, are the most important contributor to the
fine particle mass less than 2 ym. Since the program design sought to
determine maximum RWC impacts, the following conclusions reflect reasonable
worst case impact conditions rather than, for example, annual average source
impacts representative of each community's airshed.
Key findings of Task 1, then, include the following:
• RWC emissions typically account for 66% to 75% of the fine particle
mass, while transportation sources contribute 5%, secondary sulfate,
5.6% and all industrial sources less than 0.5%.
Background RWC impacts were found to range from 3-12 yg/m , 24 hour
average—a factor of ten lower than the urban sites, suggesting
that 70-80% of the RWC impact is related to local sources.
• Maximum 24 hour impacts exceeded 60yg/m3 at residential sites
located in Seattle, Spokane, Portland and Medford. Impacts at
industrial sites in Longview, Seattle and Tacoma were significantly
lower (15-30 yg/m3).
The highest impacts measured in this study were in Boise, Idaho
(128 yg/m3, 8 hour average). These measurements must, however,
be considered as upper limit estimates requiring further verification.
• PNA concentrations measured at urban sites were a factor of ten
higher than chat measured at the rural background sites. Although
the measured PNA concentrations should be of concern, no direct
dose-response relationship are currently available in the literature
upon which to base a quantitative assessment of public health risk.
Further interpretation of the results of Task 1 are included in the
Task 2A report (Analysis of Air Quality Impacts).
35
-------
REFERENCES
x° J.A. Cooper, "Environmental Impact of Residential Wood Combustion
Emissions and its Implications", Journal Air Poll. Control Assoc. 30 (8),
1980.
2> J.A. Cooper, LiA. Currie, G.A. Klouda, "Application of Carbon-14
Measurements to Impact Assessment of Contemporary Carbon Sources on
Urban Air Quality", Environ. Science & Technology, September, 1981.
3 * "Performance Test for the Evaluation of Computerized GC/MS Equipment
and Laboratories", USEPA, EMSL, ORD. 1980.
"' R.T. DeCesar and J.A. Cooper, "Medford Aerosol Characterization Study",
Final Report to the State of Oregon Department of Environmental
Quality, February, 1981.
s' J.A. Cooper and J.G. Watson, "Portland Aerosol Characterization Study",
Final Report to the State of Oregon Department of Environmental
Quality, July, 1979.
"Scientific and Technical Assessment Report on Particulate Polycyclic
Organic Matter" USEPA # EPA-600/6-75-001, 1975.
7' "Health Assessment Document for Polycyclic Organic Matter" US EPA
// EPA-600/9-79-008- 1979.
,8 * "Coal Health Effects Review Committee; Summary Report to the Oregon
Department of Environmental Quality", State of Oregon Dept. of Env.
Quality. April, 1971.
9" International Commission of Radiological Protection (Report of the
Task Group on Reference Man), Pergamon Press, New York, 1974.
10' K. Bridbord, et. al., "Human Exposure co Polynuclear Aromatic Hydrocarbons'
Carcinogenesis, Volume I, Raven Press, NY, NT, 1976.
36
------- |