EPA-600 / 7-90-006
February 1990
COMPARISON OF MEASUREMENT TECHNIQUES
FOR QUANTIFYING SELECTED ORGANIC EMISSIONS
FROM KEROSENE SPACE HEATERS
by
Gregory W. Traynor, Michael G. Apte, and Harvey A. Sokol
Indoor Environment Program
Lawrence Berkeley Laboratory
University of California
Berkeley, CA 94720
Jane C. Chuang
Analytical and Structural Chemistry Center
Battelle Columbus Laboratories
Columbus, OH 43201
EPA Interagency Agreement DW89930753
DOE Contract DE-AC03-76SF00098
EPA Project Officer:
James B. White
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Prepared for:
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-600/7-90-006
3. RECIPIENT'S ACCESSION'NO.
0 ¦ v T c~- r- /- ^ r>
- -
4. TITLE AND SUBTITLE
Comparison of Measurement Techniques for Quantify-
ing Selected Organic Emissions from Kerosene
Space Heaters
5. REPORT DATE
February 1990
6T PERFORMING ORGANIZATION CODE
7. author(s)G< w# Traynor, M. G. Apte, H. A. Sokol (LBL);
and J. C. Chuang (Battelle)
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING OROANIZATION NAME AND ADDRESS
Lawrence Berkeley Laboratory, University of Califor-
nia, Berkeley, CA 94720, and
Battelle Columbus Laboratories, Columbus, OH 43201
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA DW89930753, and DOE
DE-AC03-76SF00098
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 7/86 - 3/88
14. SPONSORING AGENCY CODE
EPA/600/13
is. supplementary notes ^EERL project officer is James B. White, Mail Drop 54, 919/541-
1189.
16. abstract reporj. gives results of (l) a comparison of the hood and chamber tech-
niques for quantifying pollutant emission rates from unvented combustion appliances,
and (2) an assessment of the semivolatile and nonvolatile organic-compound emis-
sions from unvented kerosene space heaters. In general, the techniques yielded sim-
ilar emission-rate results for CO, NO, and N02. However, when differences were
observed, it was concluded that the chember-technique value was more realistic be-
cause it allows the oxygen level supplied to the appliance to decrease as it would in
residences. A well-tuned radiant heater and a maltuned convective heater were tes-
ted for semivolatile and nonvolatile organic pollutant emissions. Each heater was
operated in a 27 cu m chamber with a prescribed on/off pattern. Organic compounds
were collected on Teflon-impregnated glass filters backed by XAD-2 resin and anal-
yzed by gas chromatography/mass spectrometry. Pollutant source strengths were
calculated using a mass-balance equation. The results show that kerosene heaters
can emit polycyclic aromatic hydrocarbons (PAHs); nitrated PAHs; alkyl benzenes;
pentachlorphenol; phthalates; hydro naphthalenes; aliphatic hydrocarbons, alcohols,
and ketones; and other organic compounds, some of which are known mutagens.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. descriptors
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution Radiant Heating
Kerosene Convection
Space Heaters Heat Transfer
Emission Volatility
Organic Compounds
Residential Buildings
Pollution Control
Stationary Sources
Indoor Air
13B
21D 20M
13 A
14G
07 C
13 M
18. distribution statement
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
49
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
rm 2220-1 (9-73)
i
-------
ABSTRACT
A study was performed to compare the "hood" and "chamber" techniques for
quantifying pollutant emission rates from unvented combustion appliances and to assess
the semivolatile and nonvolatile organic-compound emissions from unvented kerosene
space heaters. In general, the hood and chamber techniques yielded similar emission-
rate results for CO, NO, and N02. However, when differences were observed, it was
concluded that the chamber-technique value was more realistic because this technique
allows the oxygen level supplied to the appliance to decrease as it would in residences.
A well-tuned radiant heater and a maltuned convective heater were tested for
semivolatile and nonvolatile organic pollutant emissions. Each heater was operated in a
27-m3 chamber with a prescribed on/off pattern. Organic compounds were collected
on Teflon-impregnated glass filters backed by XAD-2 resin and analyzed by gas
chromatography/mass spectrometry. Pollutant source strengths were calculated using a
mass-balance equation. The results show that kerosene heaters can emit polycyclic
aromatic hydrocarbons (PAHs); nitrated PAHs; alkyl benzenes; pentachlorphenol;
phthalates; hydro naphthalenes; aliphatic hydrocarbons, alcohols, and ketones; and other
organic compounds, some of which are known mutagens.
This report was submitted in fulfillment of Interagency Agreement DW89930753-01
under the sponsorship of the U.S. Environmental Protection Agency, through the U.S.
Department of Energy. This report covers the period from May 1, 1984, to September
30, 1986, and work was completed as of September 30, 1986
ACKNOWLEDGMENTS
The authors would like to thank J.C. Bare, D.B. Henschel, D.C. Sanchez, W.G. Tucker,
and J.B. White (AEERL), J. Lewtas and J.L. Mumford (HERL) of the U.S.
Environmental Protection Agency (Research Triangle Park, NC); W.F. Gutknecht of the
Research Triangle Institute (Research Triangle Park, NC); and J. McCann and A.
Carruthers of the Lawrence Berkeley Laboratory for their assistance during various
phases of this project.
i i i
-------
CONTENTS
Abstract "iii
Acknowledgments i i i
Figures VI
Tables vi
1. Summary 1
2. Introduction 4
3. Experimental Methods 6
Air pollution monitoring instrumentation 6
Special instrumentation for comparison of hood and chamber techniques 10
Special instrumentation for evaluating selected organic pollutant emissions 10
Sample extraction 13
Gravimetric analysis 14
Total chromatographable organic material analysis 14
Gas chromatography\mass spectrometry analysis 15
Negative chemical ionization, gas chromatography\mass spectrometry 15
Electron impact, gas chromatography\mass spectrometry 17
Environmental chamber 19
Test protocols 19
Hood test protocol 21
4. Results and Discussion 25
Comparison of hood and chamber techniques 25
Selected organic pollutant emissions from unvented kerosene heaters 28
5. Conclusions 39
References 40
Appendix. Quality Control Evaluation Report 42
v
-------
7
11
22
Pa;
8
9
16
18
26
27
29
30
31
33
34
35
FIGURES
Schematic diagram of environmental chamber and the Mobile
Atmospheric Research Laboratory (MARL)
Schematic diagram of the "hood" and ducting inside of the
Environmental Chamber
The pollutant profile of a chamber test as seen from a MARL
pollutant instrument
TABLES
Instrumentation for Appliance Pollutant Emission Testing
Instrument Calibration Equipment
GC and MS Operating Conditions
Standard Compounds for the Semiquantitative Analyses of Organic
Pollutants
NO, N02, and N (of NOx) Emission Rates for Hood and Chamber
Methods Using a Radiant Kerosene and an Infrared Natural Gas
Unvented Space Heater
CO, HCHO, and TSP Emission Rates for Hood and Chamber
Methods Using a Radiant Kerosene and an Infrared Natural-Gas
Unvented Space Heater
Carbon Monoxide, Nitric Oxide, and Nitrogen Dioxide Emission
Rates for Tests of Well-Tuned Radiant and Maltuned Convective
Heaters
Total Suspended Particulate Mass and GRAV Concentration
Results for Filter-Collected Samples for Well-Tuned Radiant and
Maltuned Convective Kerosene Heaters
Total Chromatographable Organic (TCO) and GRAV Concentration
Results for XAD-Collected Samples for Well-Tuned Radiant and
Maltuned Convective Heaters
TCO, GRAV, and TSP Mass Source Strengths for a Well-Tuned
Radiant and a Maltuned Convective Kerosene Heater
Nitrated-PAH Source Strengths from a Well-Tuned Radiant and a
Maltuned Convective Kerosene Space Heater
Selected Organic Pollutant Source Strengths from a Well-Tuned
Radiant and a Maltuned Convective Kerosene Space Heater
vi
-------
SECTION 1
SUMMARY
The sales and use of unvented kerosene space heaters over the past decade have
increased dramatically in the United States. Unvented kerosene space heaters have
been found to emit a wide variety of pollutants including carbon monoxide (CO),
carbon dioxide (C02), nitric oxide (NO), nitrogen dioxide (NOJ, sulfur dioxide (S02),
formaldehyde (HCHO), and suspended particles (Yamanaka et af.. 1979; Leaderer, 1983;
Ryan et al.. 1983; Traynor et al.. 1983; Lionel et al.. 1986). Several other studies using
a kerosene-fueled turbulent-diffusion continuous-flow combustor showed that many
polycyclic aromatic hydrocarbons (PAHs) are emitted during kerosene combustion
(Prado et al. 1973; Skopek et al.. 1973; Kaden et al.. 1979). Shopek et al. and Kaden
et al. also showed that kerosene soot is indirectly mutagenic. Kaden et al. showed that
essentially all of the indirect mutagenic activity of kerosene soot was due to unnitrated
PAH compounds. Tokiwa et al.. (1985) revealed kerosene heaters to emit
dinitropyrene. These researchers also showed kerosene soot to be directly mutagenic
and showed that most of the direct mutagenic activity could be attributed to
dinitropyrenes.
The above studies have shown the following: (1) kerosene combustion products can
be mutagenic, (2) kerosene combustion can produce PAHs and nitrated PAHs, and (3)
it is likely that much of the mutagenic activity of kerosene soot is due to the PAHs and
nitrated PAHs. However, it is not known whether the unvented portable kerosene
space heaters commonly used indoor in the U.S. produce emissions similar to those
emitted by the kerosene combustors used in several earlier studies or whether these
portable space heaters produce other potentially harmful organic pollutants. Of the two
major goals in this study, the first was to measure selected organic pollutant emissions
(including PAH and nitrated PAH emissions) from portable kerosene heaters commonly
used in the U.S.
The second, but chronologically first, goal of this study was to compare two
techniques for assessing pollution emissions from unvented combustion appliances. One
technique, described by Traynor et al.. 1982 and henceforth called the "chamber"
technique, involves placing the unvented combustion appliance in a room-size or large
chamber, operating the appliance for a representative period of time, and monitoring
the increase in the chamber pollutant concentrations. The pollutant emission rate,
expressed as mass of pollutant emitted per unit of fuel consumed often /ig/kJ, is then
calculated from the chamber and outside pollutant concentrations using a single-
equation, mass-balance model. The chamber technique has been used to quantify
pollutant emission rates from kerosene heaters (Leaderer, 1983 and Traynor et al..
1983). The other technique, described by Himmel and DeWerth (1974), Yamaka et al.
(1979), and ANSI (1982), henceforth to be called the "hood" technique, involves placing
the unvented combustion appliance under a hood large enough to capture all of the
pollutant emissions and measuring the ratio of the concentration of each pollutant
under investigation to the concentration of C02 in the hood exhaust flue. After
correcting for background dilution air, the theoretical C02 emission rate is used to
calculate the emission rate of the pollutant of concern. Both Yamanaka et al. (1979)
and Lionel et al. (1986) have used the hood technique to measure pollutant emission
rates form kerosene heaters.
For the evaluation of the hood and chamber measurement techniques, one unvented
radiant kerosene heater and one infrared unvented (natural) gas space heater (UVGSH)
1
-------
were used. Radiant and infrared combustion space heaters generally have more
repeatable emission-rate characteristics than do their convective counterparts (Apte and
Traynor, 1986); therefore, using such heaters allowed any hood vs. chamber technique
emission-rate differences to be more easily detected.
For the tests measuring organic pollutant emissions from kerosene heaters, two
heater/tuning conditions were chosen based, in part, on previously reported particulate
emission data (Traynor et al.. 1983). The previous study showed that particulate
emissions from a well-tuned properly operated convective kerosene heater were
negligible but that particulate emissions from a radiant heater were not. Therefore, it
was reasoned that significant organic emissions would be more likely to be observed
from a radiant heater rather than from a convective heater. A well-tuned radiant
heater was chosen as the first heater/tuning combination to be tested. The other
heater/tuning combination chosen for testing was a maltuned convective heater. This
choice was based, in part, on conversations with kerosene-heater users and testers, who
indicated that it was easier (more likely) for a convective heater to soot (i.e., emit a
visible stream of particles) than it was for a radiant heater. In fact, altering the burner
assembly itself was the only way the radiant heater tested in this study could be made
to "soot." The convective heater was maltuned by lifting the exterior shell of the heater
by approximately one cm, thereby providing excess air to the wick. Only two
heater/tuning combinations were tested because each test had to be conducted many
times to collect enough samples for mutagenicity testing.
All experiments were conducted at the Lawrence Berkeley Laboratory (LBL).
Battelle's Columbus Division prepared and analyzed filters and resins used by LBL to
collect selected organic pollutant emissions and provided sample extracts to the Health
Effects Research Laboratory of the U.S. Environmental Protection Agency (Chuang et
aLi 1986).
In general, the hood and chamber emission-rate measurement techniques yield
similar results for CO, NO and N02. However, when discrepancies were observed,
they were believed to be caused by differences in combustion-air oxygen levels. The
chamber method results were judged to be more accurate since this method allows the
oxygen content of the combustion air to drop, as would occur in actual residences. The
hood method appears to be adequate for quantifying CO, NO, and N02 emission rates
from appliances that are not oxygen sensitive or from appliances that marginally affect
a residence's oxygen level, such as a gas range. The chamber method was preferable
for measuring total suspended particulate emissions primarily because it was easier to
implement. No disadvantages to the chamber method were discovered.
With regard to organic pollutant emissions from kerosene heaters, this study has
confirmed the results of other studies, i.e., that the kerosene combustion process can
emit PAHs and nitrated PAHs. One-nitronaphathalene is clearly emitted by well-tuned
radiant and maltuned convective kerosene space heaters. One-nitronaphthalene was
found almost entirely in the semivolatile fraction for the radiant-heater tests. For the
maltuned-convective-heater tests, thirty percent of the nitronaphathalene was collected
on the filter. This is presumably due to the heavy loading of fresh soot on the filter
during the maltuned-convective test. Emissions of 9-nitroanthracene were observed in
the XAD fraction of one of the radiant-heater tests and in the filter fraction of the
maltuned-convective test. Emissions of 1-nitropyrene were also observed in the filter
fraction of both radiant test samples, whereas only trace amounts of 3-
nitrofluoranthene were observed in one of the two series of radiant-heater tests in the
filter-collected fraction.
2
-------
In addition, kerosene heaters were found to emit many other organic compounds,
including aliphatic hydrocarbons, alcohols, and ketones; phtalates; and alkyl benzenes.
Additional analysis is needed to correlate these results with health-effects data to
determine the risk associated with these organic emissions. PAH and nitrated-PAH
emissions are sufficiently important to justify additional quantitative studies;
furthermore, examinations of other organic compounds of toxicological significance and
of unvented combustion sources should be expanded.
One very important observation of this study was that some estimates of the indoor
reactivity of SVOCs were higher than 2 h"1. This implies that reactivity rates for some
SVOCs are more important than ventilation rates for determining indoor concentrations.
Clearly, this indicates that future studies must quantify the indoor reactivity process for
individual SVOCs in order to gain insight into potential indoor exposures to these
compounds.
3
-------
SECTION 2
INTRODUCTION
The sale and use of unvented kerosene space heaters have increased dramatically in
the United States during the past decade. These heaters have been found to emit a
wide variety of pollutants including carbon monoxide (CO), carbon dioxide (CO,),
nitric oxide (NO), nitrogen dioxide (N02), sulfur dioxide (S02), formaldehyde (HCHO),
and suspended particles (Yamanaka et al.. 1979; Leaderer, 1983; Ryan et al.. 1983;
Traynor et al.. 1983; Lionel et al.. 1986). Several studies using a kerosene-fueled
turbulent-diffusion continuous-flow combustor showed that many polycyclic aromatic
hydrocarbons (PAHs) are emitted during kerosene combustion (Prado et al.. 1973;
Skopek et al.. 1979; Kaden et al.. 1979). Skopek et al. and Kaden et al. also showed
that kerosene soot is indirectly mutagenic. Kaden et al. showed that essentially all of
the indirect mutagenic activity of kerosene soot was due to unnitrated PAH compounds.
Tokiwa et al. (1985) revealed that kerosene heaters emit dinitropyrene. These
researchers also showed kerosene soot to be directly mutagenic and showed that most of
the direct mutagenic activity could be attributed to dinitropyrenes.
The above studies have shown the following: 1) kerosene combustion products can
be mutagenic, 2) kerosene combustion can produce PAHs and nitrated PAHs, and 3) it
is likely that much of the mutagenic activity of kerosene soot is due to the PAHs and
nitrated PAHs. However, it is not known whether the unvented portable kerosene space
heaters commonly used indoors in the U.S. produce emissions similar to those emitted
by the kerosene combustors used (Prado et al.. 1973; Skopek et al.. 1979; Kaden et al..
1979) or whether these portable space heaters produce other potentially harmful organic
pollutants. Of the two major goals of this study, the most important was to measure
selected organic pollutant emissions (including PAH and nitro-PAH emissions) from
portable kerosene heaters commonly used in the U.S.
The secondary, goal of this study was to compare and contrast two different
techniques for assessing pollutant emissions from unvented combustion appliances. One
technique, described by Traynor et al.. 1982, which we will call the "chamber"
technique, involves placing the unvented combustion appliance in a large (i.e., room-
size or larger) chamber, operating the appliance for a representative period of time, and
monitoring the increase in the chamber pollutant concentrations. The pollutant
emission rate, expressed as mass of pollutant emitted per unit of fuel consumed—often
/xg/kJ, is then calculated from the chamber and outside pollutant concentrations using a
single-equation, mass-balance model. The chamber technique has been used to
quantify pollutant emission rates from kerosene heaters in at least two studies
(Leaderer, 1983 and Traynor et al.. 1983). The other technique, described by Himmel
and DeWerth (1974), Yamanaka et al. (1979), and ANSI (1982), which we will call the
"hood" technique, involves placing the unvented combustion appliance under a hood
large enough to capture all of the pollutant emissions and measuring the ratio of the
concentration of each pollutant under investigation to the concentration of C02 in the
hood exhaust flue. After correcting for background dilution air, the theoretical C02
emission rate is used to calculate the emission rate of the pollutant of concern. Both
Yamanaka et al. (1979) and Lionel et al. (1986) used the hood technique to measure
pollutant emission rates from kerosene heaters.
For the evaluation of the hood and chamber measurement techniques, we used one
unvented radiant kerosene heater and one infrared unvented (natural) gas space heater
4
-------
(UVGSH). Radiant and infrared combustion space heaters generally have more
repeatable emission-rate characteristics than do their convective counterparts (Apte and
Traynor, 1986); therefore, using such heaters allowed us to easily detect any hood-vs.-
chamber emission-rate differences.
For the tests investigating organic pollutant emissions from kerosene heaters, our
choices of two heater/tuning conditions were based, in part, on previously reported
particulate emission data (Traynor et al.. 1983). The previous study showed that
particulate emissions from a well-tuned, properly operated convective kerosene heater
were negligible but that particulate emissions from a radiant heater were not.
Therefore, we reasoned that significant organic emissions would be more likely to be
observed from a radiant heater than from a convective heater. A radiant heater
operating under well-tuned normal conditions was chosen as the first heater/tuning
combination to be tested. The other heater/tuning combination chosen for testing was
a convective heater operated under maltuned conditions. This choice was based, in
part, on conversations with kerosene-heater users and testers, who indicated that a
convective heater was more likely to "soot" (i.e., emit a visible stream of particles) than
was a radiant heater. In fact, altering the burner assembly itself was the only way the
radiant heater tested in this study could be made to soot. We maltuned the convective
heater by lifting the exterior shell of the heater by approximately one cm, thereby
providing excess air to the wick. Only two heater/tuning combinations were tested
because each test had to be conducted many times to collect enough samples for future
testing for mutagenic effects.
All experiments were conducted at the Lawrence Berkeley Laboratory (LBL).
Battelle's Columbus Division (BCD) prepared and analyzed filters and resins used by
LBL to collect selected organic pollutants and provided sample extracts to the Health
Effects Research Laboratory of the U.S. Environmental Protection Agency (EPA) for
future mutagenicity testing (Chuang et al.. 1986). The mutagenicity test results from
these sample extracts will be presented in another report (private communication, J.
Mumford, 1986).
5
-------
SECTION 3
EXPERIMENTAL METHODS
AIR POLLUTION MONITORING INSTRUMENTATION
Most of the air pollution monitoring instrumentation was located in the Lawrence
Berkeley Laboratory's Mobile Atmospheric Research Laboratory (MARL), which is
designed to monitor indoor or outdoor atmospheres and the dynamic changes in those
atmospheres associated with the operation of domestic combustion appliances. The
MARL is a mobile unit equipped with gas analyzers (for C02, 02, CO, NO, NO^ and
NO ) and temperature and humidity monitors sensitive in ranges normally associated
with indoor environments. Figure 1 gives a schematic layout of the MARL. Table 1
lists the instrumentation and gives accuracy and precision estimates. Table 2 lists the
special equipment used for instrument calibration.
The MARL continuously draws air samples through Teflon tubing from two
locations and uses a timing system to automatically switch from one site to the next at
preset intervals (see Fig. 1). Teflon prefilters are fitted at the inlets of the sampling
lines to protect the instruments from particulate matter. Although the MARL can
monitor gases only from a single location at a given time, sample air is drawn through
both lines continuously to minimize purge time when the unit switches between lines.
Lines that are not being monitored are vented to the outside by an exhaust pump. A
Teflon-lined pump supplies the sample from the site being monitored to the glass
mixing manifold and maintains the manifold pressure just above atmospheric. The gas
analyzers draw the sample from the manifold by means of individual pumps. (Only
nonreactive materials are used upstream of the gas analyzers to ensure minimum
degradation of the sample.) During typical testing the total sample flow is
approximately 0.5 m3/h.
The MARL calibration system was designed for rigorous calibration of the gas
analyzers. Certified primary standard gas mixtures are diluted with "ultrapure" air
using a mass-flow controlled mixing system to produce a large range of concentrations
suitable for calibration. To check for problems such as a bad pump diaphragm or
leaky sampling lines, we inject a primary standard gas of known concentration into the
upstream inlet of a sampling line after each daily calibration session.
Two data-acquisition systems connected to a central patchboard are used to monitor
pollutant concentrations (see Fig. 1). One, a microprocessor-based system, logs primary
data on magnetic tape at one-minute intervals. The second system provides a back-up
record by printing data onto paper tape. A chart recorder connected to the patchboard
is used for real-time graphic display of an experiment in progress. At the end of an
experiment, data from the magnetic tape are read into a mainframe computer for
subsequent analysis.
Natural-gas consumption is metered by a Singer AL-425 diaphragm gas meter
located outside of the chamber. Kerosene consumption is determined by pre- and post-
test mass measurements of the heater made with a Ohaus double-beam mechanical
balance.
6
-------
BUILDING 44
TEMPERATURE
DEWPOINT
I TEMPERATURE
DEWPOINT
ENVIRONMENTAL
CHAMBER
O O
O
SPACE HEATER
ELECTRICAL -
AEROSOL
SIZE ANALYSER
, ~ FORMALDEHYDE
1 ^ SAMPLER
PUMPS/VOLUME
METERS
TO COLLECT
SUSPENDED
PARTICLES
SAMPLE POINT
LOW VELOCITY
PERIMETER MIXING FANS
MARL
TEMPERATURE
DEWPOINT MtEMPERATURE
/DEWPOINT
TEMPERATURE
SAMPLING
SYSTEM
DEWPOINT
PRIMARY
DATA
ACQUISITION
SYSTEM
SWITCHING
MULTIPOINT
PANEL
TIMING
METER
SYSTEM
ANALYZER
BACKUP
DATA
ACQUISITION
SYSTEM
DATA
CO
ANALYZER
PATCH
MULTIPOINT
SAMPLING
SYSTEM
BOARD
TEFLON
CHART
RECORDER
CALIBRATION
TANKS
PUMP
ANALYZER
ANALYZER
CALIBRATION
SYSTEM
ZERO
TO EXHAUST
PUMP
BOX
ZERO AIR
GENERATION
SYSTEM
WATER
TRAP
*Ql a-to T378C
Figure 1. Schematic diagram of environmental chamber and the Mobile Atmospheric
Research Laboratory (MARL).
7
-------
TABLE 1. INSTRUMENTATION FOR APPLIANCE POLLUTANT EMISSION TESTING
PARAMETER METHOO MANUFACTURER/ MEASUREMENT ACCURACY ESTIMATE PRECISION ESTIMATE CALIBRATION LOCATION
MODEL RANGE MANU. LBL MANU. LBL FREQUENCY
CONTINUOUS MONITORS
co2
CO
NO, N02, NOx
°2
SO-
NDIR
NDIR
Chemiluminescence
Paramagnetism
Pulsed Fluorescence
Horiba PIR 2000
Bendix 8501-5SCA
Thermoelectron 14D
Beckman 755
Thermoelectron 43
0-25%
0-50 ppm
0-10 ppm
16-21 %
0-5 ppm
+ 1%
+ 5%
+ 5%
+ 5%
+ 5%
+ 5%
+ 0.5% FS
+ 1% FS
+ 0.5%
+ 5%
+ 1%
+ 3%
+ 3%
+ 3%
+ 3%
+ 3%
Dai ly
Dai ly
Dai ly
Dai ly
Dai ly
MARL
MARL
MARL
MARL
MARL
Temperature
Relative humidity
Submicron particles
Natural-gas flow rate
Thermistor
Hygrometri cani cal
Crystalite & strain
gauges
Yellowsprings Inc. 701
Hygrometrics 8501 A
Electric mobility Thermosystems Inc. 3030
Diaphragm gas meter Singer AL-425
0-100°C
0-100°C
0-1000 mg/m
5-425 L/min
NA
+ 4%
+ 1%
+ 1%
NA
NA
+ 2%
NA
+ 1%
+ 1%
NA
NA
+ 1%
6 mo.
6 mo.
6 mo.
MARL
MARL
Chamber
44
TIME-AVERAGED SAMPLERS
HCHO
Refrigerated bubblers,
colorimetry
LBL
NA
+ 30% NA
+ 25%
Daily
44
TSP MASS
Electrobalance
Cahn 21
0-1.5 g
+ 5%
+ 2%
Once per
sample
44
TSP sample volume
Kerosene metering
Diaphragm gas meter Rockwell R-200
Double beam balance Ohaus
0-5.7 m /h +1% +2% +1% +1% 6 mo. 44
0.1-21 kg --- + 2% --- +1% 6 mo. 44
TSP/XAD-2 sample
volume
Pressure-regulated LBL
flow controller
0-7.5 m /h
+ 5%
+ 3%
Dai ly
44
-------
TABLE 2. INSTRUMENT CALIBRATION EQUIPMENT
EQUIPMENT
PURPOSE
MANUFACTURER/
MODEL
RANGE
ACCURACY ESTIMATE PRECISION ESTIMATE CALIBRATION
MANU.
LBL MANU.
LBL FREQUENCY
Mass flow controller Metering dilution gas Tylan FC261
Metering calibration gas Tylan FC260
0.4-20 L/min
4-20 cm'/min
+ 1% FS + 3%
+ 1% FS + 3%
+ 0.2% FS + 2%
+ 0.2% FS + 2%
4 mo
4 mo
Bottled Gases for
co2, o2, CO, no2,
NO, S02
Matheson primary
standard
NA
1-2%
Wet test meter
Mass flow controller
calibration
Singer SF-8
0-8 L/min
+ 0.5% + 3%
+1% 6 mo
Bubble test meter
Mass flow controller
calibration
Varian 0-60 mL
0-150 cm /min
+ 2%
+ 1%
\o b
Dry test meter
Filter/XAD-2 sampling
system
Rockwell R-200
0-5.7 m /h
+1% +2% +1% +1%
aAlso represents quality objectives.
^Calibrated against primary standard at operating flow rate of approximately 7 iri^/h.
-------
SPECIAL INSTRUMENTATION FOR COMPARISON OF HOOD AND CHAMBER
TECHNIQUES
HCHO and total suspended particulate (TSP) concentrations were measured for the
tests designed to compare the hood and chamber pollutant emission-rate measurement
techniques.
The HCHO collection system is adjacent to the chamber and draws air at a constant
rate through two dedicated sample lines for subsequent wet-chemical analysis (Miksch
et al.. 1981). Each sample is collected by a pair of water-filled impingers in series.
Duplicate samples are provided by dividing each sample line in two just upstream of
the impinger trains. The HCHO collection system is purged with nitrogen for 15-20
minutes before each test to avoid HCHO accumulation in the sampling lines.
For the chamber tests, there were three identical TSP collection systems, one
located outside and two inside the chamber to give integrated particle concentration
measurements. For the hood tests, one TSP collection system was located in the
chamber, and two systems were connected to stainless-steel probes that sampled air
through the hood flue wall (see Fig. 2). Particles were collected by drawing air
through l-/im, 47-mm Teflon, Fluoropore filters. The sample volumes were monitored
with Rockwell R200 diaphragm meters, corrected for temperature and pressure. The
loading of the filters is gravimetrically measured by pre- and postsample mass
measurements of the filters. Filters are dessicated before each mass measurement. A
Cahn 21 Electrobalance is used for weighing the filters.
The Electrical Aerosol Sizer Analyzer (EASA), which uses the mobility of charged
particles to make continual measurements of submicron particles, is located inside of
the chamber and out of the direct path of hot exhaust gases (Whitby, 1976). The
control electronics and data-recording system for the EASA are located outside and
adjacent to the chamber. The real-time particulate concentration profiles collected
with the EASA were used to transform the average TSP data into a real-time TSP
concentration profile.
SPECIAL INSTRUMENTATION FOR EVALUATING SELECTED ORGANIC
POLLUTANT EMISSIONS
Using sampling and analytical techniques developed by the EPA (Mumford et al.,
1987), we measured semivolatile organic compounds (SVOCs) and TSP, including
nonvolatile and particle-bound organic pollutants, to assess selected organic pollutant
emission rates from unvented kerosene heaters. TSP concentrations were collected on
Teflon-impregnated, glass-fiber filters that were 102 mm in diameter. The semivolatile
organic compounds were collected on 100 g of precleaned XAD-2 resin.
BCD prepared and analyzed the Teflon-impregnated glass-fiber filter and the XAD-
2 resin. From the report by Chuang et al.. (1986) describing the analytic support for
this project, the following describes Battelle's sample preparation procedures.
The XAD-2 resin was purchased from Supelco Inc. as precleaned resin
and was further cleaned by Soxhlet extraction with distilled-in-glass
(DIG) methylene chloride for 16 hours. The clean XAD-2 resin was
dried using a nitrogen gas stream to evaporate the solvent. A
purification and background check was performed on the clean resin by
10
-------
MECHANICAL
~i FAN _
TO MARL
TEMPERATURE/
RELATIVE HUMIDITY
PITOT TUBE/
MANOMETER
TO MARL
GAS ANALYZERS
PARTICLE
SAMPLING
SYSTEM
MIXING
BAFFLE
CHAMBER
BUILDING 44
SPACE
HEATER
FORMALDEHYDE
SAMPLER
ELECTRICAL
AEROSOL
SIZE ANALYZER
L.
Figure 2.
XBL 8411-6287A
Schematic diagram of the "hood" and ducting inside of the Environmental Chamber.
-------
extracting and analyzing an aliquot of the resin using gas
chromatography with flame ionization detection (GC/FID).
The clean XAD-2 resin was sent to LBL for use in sampling. To ensure
resin integrity during transportation and handling, clean steel cans were
used as shipping containers. The cans were 16.5 cm in diameter and
17.8 cm high. The resin was first sealed in an amber bottle that had
been cleaned with laboratory detergent, rinsed with tap water, distilled
water, distilled-in-glass (DIG) methanol and DIG methylene chloride,
then put in the oven at 500°C overnight. The amber bottle had a Teflon-
lined cap. Each bottle contained 130 g of prepared resin. Three bottles
were placed in a steel can and were packed with clean cotton cloth. A
total of five steel cans containing 15 bottles of clean resin were prepared
and sent to LBL.
The pre-cut Teflon-impregnated, glass-fiber filters were received from
LBL. The filters were dessicated overnight, then weighed to a constant
weight by a Mettler HL 52 balance. After weighing, the filter was
placed between two watch glasses and sealed with Teflon tape. The
weight and filter number were recorded on a label on the watch glasses.
These filters and the XAD-2 resin were shipped to LBL by overnight
express mail.
Three XAD-2/filter sampling modules were used at LBL for organic pollutant and
TSP collection. Two modules were placed inside the chamber, and one module was
placed outside the chamber, but inside Building 44, which houses the chamber (see Fig.
1).
The XAD-2/filter sampling modules, primarily constructed of stainless steel, were
designed to collect particles and SVOC's when used with a medium-flow sampling
system. The sampling module consisted of an open-face, 102-mm, filter clamping ring
threaded onto a modified 15-cm-deep threaded stainless-steel pipe. Stainless-steel
screens and retainers are used to hold the filter in place and create a volume for
approximately 100 g of XAD-2 resin. A threaded cap fits on the back end of the pipe.
O-rings were used behind the filter clamping ring to prevent leaks.
Before each use, the XAD-2/filter sampling modules were cleaned. First, the
modules were checked to ensure that all XAD resin particles were removed. Second,
the modules were completely disassembled. Third, the O-rings and all parts of the
modules were held with stainless-steel forceps and thoroughly cleaned by a stream of
methylene chloride from a Teflon wash bottle. All rinsed parts were placed on a clean,
dust-free surface (such as lint-free laboratory wipes or glass). Finally, all parts (except
O-rings) were rinsed again with methylene chloride and dried in an oven at 100 °C for
at least 30 minutes.
After drying, the parts were removed from the oven; linen gloves were used, and
only the exterior of the parts were handled. The modules were partially reassembled,
and the inlet and outlet parts were covered with aluminum foil. The modules were
kept in a clean, safe place until ready for use.
Just prior to use, the XAD-2/filter modules were assembled and attached to a
medium-volume flow sampler (6.8 m3/h) by way of fittings on the filter holders. The
steps described below were followed:
12
-------
1. A clean location, lined with aluminum foil was prepared.
2. A bottle of approximately 130 g of XAD-2 resin was opened (linen gloves were
used), and approximately 100 g of resin was poured into the open XAD-2/filter
module. A retaining screen was placed over the XAD-2.
3. The remaining XAD-2 resin (approx. 30 g) was resealed in its original bottle,
labeled, and saved for analysis as a blank.
4. The filter-support screen assembly (with O-ring) was placed over the XAD-2's
retaining screen. The Teflon-impregnated glass-fiber filter was placed on the
support screen, and the open-face cap was threaded onto the rest of the module.
Once the filter and XAD-2 were placed in the module, sampling could begin.
After sampling was completed, a new clean workspace was made, and the following
steps were taken:
1. Clean stainless-steel tweezers were used to remove the filter from the module. The
filter was folded in half and then in half again so that all particulate-laden surfaces
faced inward. The folded filter was wrapped in aluminum foil and placed in a
glass vial sealed with Teflon tape and Teflon lid liners.
2. The XAD-2 resin was poured through a clean glass funnel into a clean amber glass
bottle with a Teflon-lined cap. Clumps of XAD-2 were broken up with a clean
stainless-steel spatula and glass rod.
3. The bottle was sealed and labeled.
4. The filters and XAD-2 resin were packaged in clean metal paint cans and shipped
on dry ice with a completed data sheet by an overnight express service to BCD for
extraction and analysis.
Only one of twenty-six XAD-2 resin bottles arrived at Battelle broken and was
eliminated from subsequent data analysis.
The following analytical procedures were used by BCD and are quoted from
Chuang et al.. (1986).
SAMPLE EXTRACTION
After sampling, the filter and XAD-2 samples were returned to BCD
from LBL. The container for sampled XAD #7 was broken when this
sample was received. However, approximately 90 percent of the XAD-2
resin was recovered from the steel can. The remaining samples were
received in good condition.
The filter samples were placed in a dessicator overnight and the filter
weights were obtained by using the same procedures described in the
previous section.
Soluble organic material was removed from the filter and XAD-2
13
-------
samples by Soxhlet extraction with DIG methylene chloride. Extractions
were carried out for a minimum of 16 hours and until the solutions in
the Soxhlet extractor's top chamber became colorless. After extraction,
an aliquot of each selected sample extract was removed for total
chromatographable organic matter (TCO) analysis. The remaining
extract was concentrated by Kuderna-Danish evaporation to
approximately 10 ml. The concentrated extract was then transferred to a
sample container for chemical or biological analyses.
GRAVIMETRIC (GRAV) ANALYSIS
The GRAV analyses were performed to quantify extracted organic
material with boiling points predominantly over 300°C. An aliquot of
known volume of the concentrated extract was placed on a tarred
aluminum pan and the solvent allowed to evaporate to dryness at room
temperature. The residue was weighed until weight change was less than
1-2 fig. A Mettler ME30 microbalance was used for these analyses.
The residue weight of the aliquot analyzed was then scaled to the total
quantity in the original sample extract.
TOTAL CHROMATOGRAPHABLE ORGANIC MATERIAL (TCO)
ANALYSIS
The TCO determination was performed to quantify organic materials
having boiling points in the range of 100 to 400°C. Because materials in
the TCO volatility range may be lost to varying degrees during solvent
evaporation, this analysis was performed prior to any concentration step.
For determination of TCO, 2 /xl of the extract was analyzed by gas
chromatography (GC) using a flame ionization detector. A 30-meter
with 0.25-mm I.D., DB-5 fused silica capillary column was used. The
GC was operated using the following program: injected at 40°C hold for
4 min., then program from 40°C to 300°C at 8°C/min. and hold at
300°C for 8 min. The injector temperature was 275°C and the detector
temperature was 300°C. A Hewlett-Packard Model 5730 GC, including
a Hewlett-Packard autosampler, was used for these analyses. Data
collection and processing were accomplished using a Computer
Automated Laboratory System (CALS) software program on a Hewlett-
Packard Model 1000 computer.
The quantitative calibration of the TCO procedure was accomplished by
using mixtures of known concentrations of the normal hydrocarbons Cg,
Ci2, C16, and C2Q. The quantitative calibration standards were prepared
to cover the concentration ranges of 2.5, 5, and 10 ppm levels.
Retention time limits corresponding to the TCO range of boiling points
were defined by the peak maxima for n-heptane (Cy, b.p. 98°C) and
hexacosane (C2_, b.p. 412°C). Therefore, integration of detector
response started at the retention time of C7 and terminated at the
retention time of C26. The integrated area covered organic material in
the boiling point range of 100 to 400°C. Data analyses involved
comparison of the total integrated peak area between the peak maxima
for Cy and C26 for the standard and unknown chromatograms. Data
14
-------
processing was accomplished using a CALS software program and the
data were reported as total weight (mg) per sample.
GAS CHROMATOGRAPHY/MASS SPECTROMETRY (GC/MS)
ANALYSIS
The combined extracts of filter and XAD-2 samples as well as filter and
XAD-2 field blanks were analyzed by on-column injection, negative
chemical ionization (NCI), GC/MS to identify and quantify selected
nitrated PAH. These samples were also analyzed by conventional
splitless injection, with electron impact mode (EI), GC/MS method in
full scan mode to semi-quantitatively determine the major components.
A Finnigan Model 4500 GC/MS with methane as the carrier gas was
used for these analyses. An Ultra #2 crosslinked, 5 percent phenyl
methyl silicon column was used as both the GC column and the transfer
line between the GC oven and the MS ionization source. Data
acquisition and processing were performed with a Finnigan INCOS
Model 2300 data system. The GC and MS conditions are summarized in
Table (3).
Prior to analysis, the system was calibrated by introducing a standard,
perfluorotributylamine (FC-43), and determining the mass assignment
for principal fragment ions. The mass calibration table was stored and
served to calibrate the ion masses over the scanning range.
NEGATIVE CHEMICAL IONIZATION, GAS CHROMATOGRAPHY/
MASS SPECTROMETRY (NCI GC/MS)
Each combined sample extract was spiked with the internal standard, dg-
nitropyrene, immediately prior to analysis. This NOz-PAH does not
occur naturally, and thus, will not pose a problem as an interferent.
The use of an internal standard from the same compound class as the
compounds to be quantified which has an average molecular weight and
elution time of those compounds enhances the analytical precision of the
quantification. This is due to the fact that an average correction factor
for injection technique, chromatographically active site effects and
variation in the daily mass spectrometer tuning are applied to the
quantification of all compounds.
One set of standards covering the calibration range was analyzed initially
to ensure linearity. A four-point standard curve was generated for each
of the target compounds. These standards were analyzed in triplicate
along with a blank sample to determine the baseline. The concentration
range of the standards was 0.1 /xg/mL, 0.3 /xg/mL, 1 /ig/mL and 3
/xg/mL, each containing the internal standard, dg-nitropyrene, at a
concentration of 1 /ig/mL. A volume of 1 pL of each solution was
injected and the response compared to that of the internal standard was
calculated. The calibration curve for each target compound was
generated using the following equation:
y = mx + b
15
-------
TABLE 3. GC AND MS OPERATING CONDITIONS®
Chromatopraphv
Columns Ultra #2 crosslinked, 5% phenyl methyl
silicon 50 m x 0.32 mm, 0.5-/*m film
thickness
Carrier
Carrier linear flow
velocity
Injection volume
Injection mode
50 cm/sec at 250°C
1 (iL for on-column injection
2 fiL for splitless injection
On-column modeb and splitless
mode0
Temperature
Initial column temperature
Initial hold time
Program rate
Final hold time
45°C
2 min
100°C (5 min) to 320°C at
10°C/minb, and 45°C to
320°C at 8°C/min
15 min
EI at 70 eV, NCI at 150 eV
0.35 mA
10~8 A/v
~105
aChuang et al. (1986)
bUsed for the NCI, GC/MS.
cUsed for the EI, GC/MS.
Mass spectrometer
Ionization
Filament emission current
Preamplifier
Electron multiplier gain
16
-------
where y =
A /A. , x = C /C.
s' 18 ' s' 18
m =
slope , b = intercept
A =
8
Area of the target compound
A. =
18
Area of the internal standard
C =
8
Concentration of the target compound
C. =
18
Concentration of the internal standard.
The computer performed a least squares analysis and calculated a
correlation coefficient and intercept. The correlation coefficient must
be at least 0.990 for an acceptable calibration curve. After the
calibration curve was established, the order of analysis was: standard,
sample, sample, sample, standard, sample, sample, sample, standard until
all analyses of a set were completed. In this way, the calibration curve
generated from all of the standard analyses accurately reflects the
condition and operation of the mass spectrometer throughout the
analyses.
The validity of the calibration curve was monitored by analyzing
successive calibration solutions using the curves, and comparing the
value obtained with the known value. Quantifications within 30 percent
of the true value are considered acceptable and do not require reanalysis
of the sample.
This project was not designed to determine the detection limit of the
NCI GC/MS method. However, based on the lowest standard analyses,
the estimated detection limit for the N02-PAH ranged from 0.01 ng to
0.001 ng on column.
ELECTRON IMPACT, GAS CHROMATOGRAPHY/MASS
SPECTROMETRY (EI GC/MS)
The combined sample extracts were also analyzed by EI, GC/MS in the
full scan mode. Four of the selected sample extracts were analyzed first.
The identifications of sample components were conducted by comparison
to reference spectra in the EPA/NIH mass spectral library which
contains over 30,000 reference mass spectra. Battelle reported the results
to the EPA Project Officer. The surrogate standards were then chosen
to represent the full range of organic classes that are present in the four
primary samples. The list of the surrogate standards is given in (Table
4). A partial calibration curve for each of these target compounds was
generated (2 standard concentration levels, triplicate analyses for each
concentration level as described in [Table 4]). The internal standard, 9-
phenyl anthracene, was added to the standard solution and the sample
extracts. The remaining sample extracts were also analyzed by the semi-
quantitative EI GC/MS technique. Identification of the components
from this semi-qualitative approach was performed by comparison to ref-
17
-------
TABLE 4. STANDARD COMPOUNDS FOR THE SEMIQUANTITATIVE
ANALYSES OF ORGANIC POLLUTANTS3
High Level, Low Level,
Compound Name Compound Class /ig/mL /ig/mL
Octane
Aliphatic hydrocarbons
20
10
Decane
Aliphatic hydrocarbons
20
10
Dodecane
Aliphatic hydrocarbons
20
10
Heptadecane
Aliphatic hydrocarbons
20
10
n-Eicosane
Aliphatic hydrocarbons
20
10
Docosane
Aliphatic hydrocarbons
20
10
Naphthalene
Polynuclear aromatic
compounds
10
5
Phenanthrene
Polynuclear aromatic
compounds
10
5
Pyrene
Polynuclear aromatic
compounds
10
5
Chrysene
Polynuclear aromatic
compounds
10
5
Benz o(a)pyrene
Polynuclear aromatic
compounds
10
5
Benzo(ghi)perylene
Polynuclear aromatic
compounds
10
5
Acridine
Nitrogen heterocyclic
compounds
10
5
Pentachlorophenol
Phenols
10
5
Benzoic acid
Organic Acids
20
10
1,2,4-Trimethyl-
benzene
Base-neutral organic
compounds
20
10
Di-n-ethyl phthalate
Phthalate
20
10
Di-n-butyl phthalate
Phthalate
20
10
Di-(2-ethyl hexyl)
Phthalate
20
10
phthalate
'Chuang et al. (1986)
18
-------
erence spectra in the EPA/NIH mass spectral library using an INCOS
software program. In addition, the manual search for PAH compounds
and the manual confirmation of the computer-search data, were also
carried out. In this semi-quantitative approach, the average response
factor of each target compound was generated from the standard
analyses. The response factor of each target standard was used for the
same class of organic compounds identified in the sample extracts (e.g.,
naphthalene as the standard for all 2-ring PAH and phenanthrene for 3-
ring PAH). If some compounds found in the sample extracts are not
represented by the selected target standards, the response factors for
these compounds were designated as 1. This analysis is called semi-
quantitative because results are estimated to be accurate only to within a
factor of 3 to 4.
ENVIRONMENTAL CHAMBER
All tests were conducted in a 27-m3 environmental chamber housed in Building 44 at LBL
(see Fig. 1). The chamber walls and ceiling are taped-and-sealed sheet rock, and the floor is
concrete. The air exchange rates of the chamber were in the range of 0.4 to 0.6 air changes
per hour for chamber tests conducted as part of the chamber- or hood-method comparisons.
The chamber air exchange rates for organic pollutant emissions tests averaged 1.1 ± 0.1 air
changes per hour because of high sampling flow rates needed for the XAD-2/filter collecting
systems. Hood tests were conducted in the chamber with the chamber door open; however, all
pollutants were exhausted outside of Building 44, and the chamber air exchange rate is not a
relevant parameter for such tests.
TEST PROTOCOLS
A total of 16 tests were conducted for the chamber vs. hood method comparison. Eight
tests were conducted using each method. Of the eight tests, four were conducted with a
radiant kerosene heater, and four were conducted with a natural-gas infrared heater. The fuel
consumption rates of the radiant kerosene heater averaged 7430 ± 100 kJ/h for the chamber
tests and 8080 ± 190 kJ/h for the hood tests, whereas the fuel consumption rates of the natural-
gas infrared heater were the same for both test types and averaged 20,600 ± 170 kJ/h. The 9%
higher fuel consumption rate for the radiant kerosene heater during the hood tests may be a
result of the higher oxygen concentrations of the combustion supply air for the hood tests.
A total of nine tests were conducted using the chamber method to assess the organic
pollutant emission rates from kerosene heaters. Five tests were conducted with a well-tuned
radiant heater, two tests were conducted with a maltuned convective heater, and two control
tests were conducted without a heater. The fuel consumption rates of the radiant heater
averaged 7000 ± 100 kJ/h, and the rates for the maltuned convective heater averaged 6900 ±
600 kJ/h. A single-equation mass-balance model was used to calculate pollutant emission rates
from unvented combustion space heaters based on laboratory data obtained by using the
chamber technique (Traynor, et al., 1982). This model has been used successfully to predict
indoor air pollution levels as well as to determine indoor air quality parameters that can affect
such levels. The model is repeated here.
The mathematical expression for a change in the average indoor gaseous pollutant
concentration of a whole house is as follows:
19
-------
dC = PaCo dt + S dt - (a + k) C dt, (1)
where C = indoor pollutant concentration (ppm),
CQ = outdoor pollutant concentration (ppm),
P = fraction of the outdoor pollutant level that penetrates the building shell
(unitless),
a = air exchange rate (h"1),
S = indoor pollutant source strength (cm3/h),
V = volume (m3), and
k = net rate of removal processes other than air exchange (h"1).
For particles, C and Co are usually expressed in units of Mg/m , and S is expressed in units of
/zg/h. Assuming CQ, P, a, S, and k are constant over the period of interest, Eq. 1 can be
solved for C(t), the chamber pollutant concentration at time t, to give
= PaCo+ S/y (1 _ e-(a+k)t} + C(Q) e-(a+k)t (2)
a + k
Equation 2 describes the spatial average concentration of a pollutant in an enclosed space of a
given volume.
Solving Eq. 2 for S, dividing it by the fuel consumption rate, R (kJ/h), and letting T equal
the duration of appliance operation, we can obtain the emission rate, E (/ig/kJ for particles
and cm3/kJ for gases):
-(a+k)T VPaC
E=l = F(a + k> " ? Ifr 2 (3)
R R j_e"(a+k)T R
For gases, E, in cm3/kJ, has been converted to /xg/kJ by using the ideal gas law and the time-
weighted average temperature and pressure in the space of concern. Note that Eq. 3 relies on
the final average indoor pollutant concentration, C(T). In laboratory tests, the use of mixing
fans increases the accuracy and precision of the C(T) measurement.
The chamber-technique testing protocol was used to assess both short-term and long-term
pollutant emission rates from UVGSHs. Each heater was placed on a movable cart before
testing, and a long flexible hose was used to supply the heaters with either natural gas or
propane. The heater was initially operated inside the chamber until 5500 kJ of fuel was
consumed (140 L of natural gas or 60 L of propane). Then, while the heater was still
operating, it was rolled out of the chamber and out of the building (Building 44) surrounding
the chamber to the outside. Heater operation continued in a partial enclosure for at least 90
minutes. The heater was then returned to the chamber, and another 5500 kJ of fuel was
consumed before the heater was shut off.
20
-------
Figure 3 shows the pollutant profile of a test as seen by the MARL pollutant instruments.
The pollutant concentration in Building 44 was monitored three times during each test. Twice
during each test, pollutants escaped from the chamber at a rate higher than the infiltration
rate; once when the door was opened to remove the heater from the chamber and again when
the door was opened to return the heater to the chamber. This loss of chamber pollutants
slightly alters the emission-rate calculations in this study as compared to those in our previous
studies. Previously we computed C(T) in Eq. 3 by "backtrack" from the decay curve after the
heater was shut off. This ensured that the C(T) value was determined from "well-mixed"
concentration data. This method cannot be used in the present case because we must open the
door and C(T) would be biased low during the short-term emission-rate portion of the test.
Also, during the long-term portion of these tests, C(0) in Eq. 3 is uncertain because the door
was open while we returned the heater to the chamber.
Under the protocol used in this study, C(T) for the first burn is taken directly from the
data, and C(T) for the second burn is determined by backtrack. Approximately 4% of the
chamber pollutants was lost after the peak of the first burn when the door was open, and the
measured peak is usually higher than the backtrack peak. However, in both burns the
emission-rate values were adjusted to compensate for these effects (see discussion below); C(0)
for both burns is taken directly from the actual data.
Because of the increased uncertainty in either C(T) and C(0) for each burn, the
uncertainty in our CO, emission rates was considerably greater than the 3% measured in our
earlier laboratory study (Traynor et al.. 1985). Assuming that this increased uncertainty
affected other pollutants as well as C02, we applied a normalizing factor, based on the
theoretical C02 emission rates, to our emission-rate calculations. Having previously
demonstrated that the measured COz emission rate was not discernible from the theoretical C02
emission rate (Traynor et al.. 1985), we corrected all of the data by multiplying each emission
rate by the ratio of the theoretical C02 emission rate to the measured C02 emission rate.
These corrections were on the order of 5 to 10%. The theoretical C02 emission-rate value
used for natural gas was 51,000 A»g/kJ (approximately 28 cc/kJ), and the value used for
kerosene was 71,300 /xg/kJ (approximately 39 cc/kJ).
The special chamber tests used to collect semivolatile organic pollutants and total
suspended particles varied slightly from the above protocol so we could get larger pollutant
samples to analyze. The five tests with the well-tuned radiant heater were eight hours long.
The heater was operated for one hour and then turned off for one hour. This cycle was
repeated four times. The heater was never removed from the chamber. A technician had to
enter the chamber to start and stop the heaters. The two maltuned-convective-heater tests
were four hours long, each consisting of two cycles of one hour on and one hour off. The two
control tests were each eight hours long.
HOOD TEST PROTOCOL
The hood method relies on the fact that the C02 emission rate from the combustion of a
particular fuel is very constant. The combustion products from the appliance being tested are
collected in a hood. Emission rates for any measured pollutant can be calculated from the
ratio of its concentration to the C02.
Inside the environmental chamber, a stainless-steel hood was placed over the heater to be
tested (see Fig. 2). The hood was connected, via sheet metal and metal flexhose ducting, to an
exhaust blower and regulator valve. The pollutants were vented outside of Building 44. All
21
-------
Door
opened
Door
opened
First
burn
Second
burn
Building 44 Sampling
0 12 3
Time (h)
XBL 847-8549
Figure 3. The pollutant profile of a chamber test as seen from a MARL pollutant instrument.
-------
parts of the flue system upstream of the sampling probe were constructed of nonreactive
materials such as stainless steel or Teflon. A baffle was placed in the inlet of the duct to
promote mixing. Ports for sampling probes were placed at points in the ductwork that allowed
for removal of a representative well-mixed sample of diluted combustion products from the
appliance being tested.
The exhaust flow rate in the hood was set high enough so that the hood collected the
entire mass of combustion products emitted from the device being tested. However, the
dilution of combustion products was not so great as to sacrifice the accuracy of pollutant
measurements. Finally, the pollutant concentrations were kept within the ranges of the gas
analyzers used.
The hood was placed directly above the heater. The pollutants were measured from the
hood ductwork at least 15 cm downstream of the mixing baffle. The concentrations of C02,
CO, NO, NOz, NOx, and 02 were measured every minute. Temperature and humidity were
also measured every minute both inside and outside the duct. Measurements of the
background concentrations of these parameters were made in the dilution air around the heater
(in the chamber) and in the makeup air entering the chamber. Integrated measurements began
when CO and C02 levels in the duct reached steady state, and measurements continued until a
sufficient sample was collected. Fuel consumption rates were also measured during each test.
The basic theory underlying the hood method is that combustion devices emit carbon
dioxide at a constant rate that is proportional to their fuel consumption rate. Thus, if a
representative instantaneous sample of emissions from a combustion appliance is taken, the
amount that the sample has been diluted by surrounding air can be calculated if the
concentration of CO^ is measured and the theoretical C02 emission rate for the fuel is known.
Furthermore, the emission rate of any other product of combustion can be found by the simple
relationship between the C02 concentration, the theoretical C02 emission rate, and the
concentration of the other product. The mathematical relationship is
E ¦ ec°2 <4>
where E = emission rate of the measured pollutant (/ig/kJ),
Eco = theoretical emission rate of C02 for fuel used (/xg/kJ),
2
o
AC = concentration of pollutant of interest minus background (/xg/m ), and
ACO2 = concentration of COz measured minus background (fig/m3).
The theoretical C02 emission rates are 51,000 ng/kJ for natural gas and 71,300 fig/kJ for
kerosene.
Source strengths in /xg/hr can be determined by the relationship:
S = E • R (5)
23
-------
where R = fuel consumption rate (kJ/hr), and
S = source strength (/ig/hr).
The amount of dilution air or the amount of ambient air that mixes with the combustion
pollutants is accounted for by measuring the ACO, concentration in the hood. However, if we
did want to know the dilution factor we would cuvide the measured AC02 by the theoretical
"air-free" CO, concentration (for natural gas the "air-free" concentration is approximately
12%).
When continuous data are available for the pollutant being measured, continuous pollutant
emission rates can be determined using this method. This is not the case for continual or
time-weighted average pollutant concentration measurements, which will yield average
emission rates over the period sampled.
Although the hood method is quite simple, it involves certain assumptions. The first and
probably the most important assumption is that the sample taken is representative of the
pollutant mixture being produced by the combustion device. Proper hood design, flow rates,
and mixing baffles should ensure this; in this study, tests were conducted to confirm good
mixing. A second assumption is that all the pollutants being measured are from the
combustion device. This is ensured by measuring the background air every 1/2 hour and
subtracting the average background pollutant concentrations from the concentrations measured
in the hood. A third assumption in the model is that the pollutants do not react with the
interior hood surfaces. Nonreactive materials such as teflon and stainless steel were used in
the hood to minimize any such reactions.
24
-------
SECTION 4
RESULTS AND DISCUSSION
COMPARISON OF HOOD AND CHAMBER TECHNIQUES
Tables 5 and 6 summarize the NO, N02, N(of NO ), CO, HCHO, and TSP emission rates
for a radiant kerosene heater and an infrared naturaf-gas heater tested using the hood and
chamber techniques. Both initial (Burn 1) and steady-state (Burn 2) emission rates were
reported.
The most striking difference between the two emission-rate measurement techniques was
observed for the NO emission rate during the second burn of the radiant kerosene heater test.
The chamber technique measured an average emission rate of 0.01 /Jg/kJ, whereas the hood
technique measured an average emission rate of 1.03 /xg/kJ. None of the other hood-versus -
chamber emission-rate differences were that dramatic; however, most of the N02 and N(of
NOx) emission-rate comparisons were significantly different at the 90% confidence level. For
NO,, the chamber method measured 10-30% higher emission rates than did the hood method.
Although statistically significant at the 90% confidence level, a difference in emission rates of
10-30% is not considered to be a major discrepancy.
The major difference observed for CO emission rate by each method was again during the
second burn of the radiant kerosene heater. The chamber method yielded a CO emission rate
of 172 /xg/kJ, whereas the hood method yielded 93 ng/kJ. No other major differences in CO
emissions were observed. The large test-to-test variation of HCHO and TSP emission-rate
measurements prevented any meaningful conclusion about differences being caused by the
measurement technique employed. However, it was more difficult to obtain a TSP emission
rate using the hood method, and we were only able to get a steady-state (Burn 2) TSP emission
rate using the hood method because of testing logistic difficulties.
In general, the differences in emission rates recorded by the hood and chamber techniques
were minor. The differences that were observed could be explained by the fact that the
chamber method allows the oxygen level of the heater's combustion air to drop, whereas the
hood method keeps the combustion-air oxygen concentration at ambient concentrations
(approximately 20.9%). This idea is reinforced by the difference between the hood and
chamber NO and CO emission rates for the first and second burns of the radiant kerosene
heater. In the chamber method, the combustion-air oxygen level is lower during the second
burn than during the first burn, so any difference in emission rates caused by differences in
combustion-air oxygen levels would logically be more dramatic during the second burn.
Another probable result of the difference in combustion-air oxygen levels was the 9% higher
fuel consumption rate of the radiant heater during hood tests, as compared to chamber tests.
Since oxygen levels do drop in real residences when unvented combustion space heaters are
used, there is reason to believe that the emission rates derived from the chamber technique are
more accurate. This would be especially true for unvented combustion appliances that are
very sensitive to combustion-air oxygen levels, such as some convective unvented gas space
heaters (Traynor et al.. 1985). It is not always possible to determine whether a particular
appliance is oxygen sensitive a priori. In such cases, the chamber method would be preferable.
However, if the appliance is not oxygen sensitive, or if the oxygen level does not decrease
substantially when the appliance is used, as is the case for gas cooking ranges that are used
sparingly compared with space heaters, then it appears that the hood and chamber techniques
will yield similar results for NO, N02, and CO. The results presented here are in general
25
-------
TABLE 5. NO, N02, AND N (OF NOx) EMISSION RATES FOR HOOD AND
CHAMBER METHODS USING A RADIANT KEROSENE AND AN INFRARED
NATURAL GAS UNVENTED SPACE HEATERS
NO
(fj. g/kJ)'
N°z
(Mg/w r
NO(of NO )
(P8AJ)'
Burn 1
RADIANT KEROSENE HEATER
Hood
Chamber
0.86 ± 0.31
0.67 ± 0.15
4.38 ± 0.35b
4.84 ± 0.33
1.73 ± 0.07
1.77 ± 0.12
Burn 2
Hood
Chamber
1.03 ± 0.42b
0.01 + 0.02
3.71 ± 0.59b
4.61 ± 0.29
1.61 ± 0.08b
1.36 ± 0.14
Burn 1
INFRARED NATURAL-GAS HEATER
Hood 0.73 ± 0.17 2.77 ± 0.13b 1.18 ± 0.06b
Chamber 0.92 ± 0.27 3.58 ± 0.56 1.51 ± 0.22
Burn 2
Hood 0.65 ± 0.48b 3.02 ± 0.25b 1.22 ± 0.18b
Chamber 1.08 ± 0.18 3.81 ± 0.46 1.66 ± 0.10
aPlus/minus values relate to 90% confidence interval, sample #4.
bEmission-rate results from chamber and hood differ significantly at
the 90% confidence level.
26
-------
TABLE 6. CO, HCHO, AND TSP EMISSION RATES FOR HOOD AND CHAMBER METHODS
USING A RADIANT KEROSENE AND AN INFRARED NATURAL-GAS UNVENTED SPACE HEATERS
CO HCHO TSP
(^g/kJ)a (a* g/kJ)a (MS AJ)3
Burn 1
RADIANT KEROSENE HEATER
Hood 109 ± 30 0.25 ± 0.33 b
Chamber 100 ± 12 0.22 ± 0.14 0.35 ± 0.33
Burn 2
Hood 93 ± 40° 0.42 ± 0.28 0.10 ± 0.06
Chamber 172 ± 33 0.37 ± 0.36 0.16 ± 0.27
Burn 1
INFRARED NATURAL-GAS HEATER
Hood 39 ±3° 0.19d b
Chamber 48 ± 9 0.40 ± 0.45 0.17 ± 0.14
Burn 2
Hood 39 ± 7 0.40 ± 0.25 < 0.05
Chamber 40 ± 6 0.67 ± 0.61 0.69 ± 1.5
aPlus/minus values relate to 90% confidence interval, sample #4.
b0nly one emission-rate measurement was made per test in the hood mode.
This rate reflects the emissions of the heaters after they have reached
steady- state operation during Burn 2.
cEmission rate results from chamber and hood differ significantly at the
90% confidence level.
d0nly one test.
27
-------
agreement with Moschandreas et al.. (1986a, 1986b).
SELECTED ORGANIC POLLUTANT EMISSIONS FROM UNVENTED KEROSENE
HEATERS
Carbon Monoxide. Nitric Oxide, and Nitrogen Dioxide
CO, NO, and NO^ emission-rate results for each test that employed organic pollutant
samples are presented in Table 7. As previously described, these tests are distinctly different
from the ones used to compare the hood and chamber techniques. The emission-rate results
for the well-tuned radiant heater are consistent with previously published data (Leaderer, 1983;
Traynor et al.. 1983). The results for the maltuned convective heater show that total nitrogen
oxide (NO and N02 = NOx) emissions are 27% lower than for a well-tuned convective heater.
A well-tuned heater will emit approximately 15 /ig/kJ of N (of NO ) (Traynor et al.. 1983);
the maltuned convective heater emitted approximately 11 ^g/kJ ofr N (of NOx). The CO
emissions from the maltuned-convective heater were similar to those from a well-tuned heater.
This is not expected to be a universal result but is probably unique to the method of
maltuning,i.e., supplying excess air to the combustion region, used in this study.
TCP. GRAY, and TSP Mass
Table 8 lists the TSP mass and GRAV concentration results for filter-collected samples. Both
TSP mass and GRAV concentrations were higher for the maltuned-convective-heater tests than
for the radiant-heater tests. However, the ratio of GRAV to TSP mass was much lower for
the maltuned-convective-heater tests than for the radiant-heater tests. This observation is
consistent with a previously reported observation that increased sooting does not cause a
proportionate increase in organic pollutants (Howard and Longwell, 1983).
Table 9 lists the GRAV and TCO results for XAD-collected samples. Again, notice that
the GRAV and TCO indoor concentrations for the well-tuned-radiant and the maltuned-
convective tests are not very different, despite the great difference in TSP concentrations.
Tables 8 and 9 clearly indicate that most kerosene heaters do emit some organic pollutants and,
based on GRAV results, that most of the organic pollutants were trapped by the XAD-2 resin.
TCO analysis was not performed on the filter-collected samples under the assumption that
compounds with boiling points lower than 300°C would pass through the filter and be collected
on the XAD. As will be discussed later in this report, the assumption was good for the well-
tuned-radiant-heater tests but was not as applicable for the maltuned-convective-heater tests.
For the latter tests, the heavy soot loading on the filters trapped a significant fraction of many
SVOCs before they reached the XAD.
Nitrated-PAH and Other Organic Compounds
Samples of similar types were combined before organic and nitrated-PAH compound analyses
were conducted. XAD samples were not mixed with filter samples. Samples for radiant tests
coded RAD-1, RAD-2, RAD-4, and RAD-5 were combined (note that the TCO results for
RAD-3 were much higher than for the other radiant tests so RAD-3 was analyzed separately);
samples from maltuned-convective tests MCON-1 and MCON-2 were combined; and samples
from CONTROL-1 and CONTROL-2 were combined. Indoor and outdoor samples were not
mixed but were combined individually with their counterparts from other tests using the above
scheme. The final result was eight XAD and eight filter samples: two indoor radiant, two
outdoor radiant, one indoor convective, one outdoor convective, one indoor control, and one
outdoor control.
28
-------
TABLE 7. CARBON MONOXIDE, NITRIC OXIDE, AND NITROGEN DIOXIDE EMISSION
RATES FOR TESTS OF WELL-TUNED RADIANT AND MALTUNED CONVECTIVE HEATERS
Test Average Emission Rate" Cug/kJ)
Code CO NO NO„
RAD-1 92 ± 16 0.69 ± 0.16 5.1 ± 0.4
RAD-2 88 ± 11 0.53 ± 0.19 5.0 ± 0.3
RAD-3 77 ± 10 0.69 ± 0.21 5.0 ± 0.3
RAD-4 85 ± 12 0.85 ± 0.24 4.7 ± 0.2
RAD-5 79 ± 4 0.71 ± 0.19 4.6 ± 0.2
MCON-1 22 ±7 21 ±3 7.5±1.8
MCON-2 18+4 22+0 5.6+1.6
aRadiant (RAD) test averages are from four 1-hour burns. Maltuned-convective
(MCON) test averages are from approximately two 1-hour burns. Plus/minus
values are standard deviations of the two or four emission rates calculated
from each 1-hour burn.
29
-------
TABLE 8. TOTAL SUSPENDED PARTICULATE MASS AND GRAVa CONCENTRATION
RESULTS FOR FILTER-COLLECTED SAMPLES FOR WELL-TUNED RADIANT AND
MALTUNED CONVECTIVE KEROSENE HEATERS
Test Mass (ug/rn^) GRAVa (ug/m3)
Code In Out In Out
RAD-1 28 18 nmb nm
RAD-2 23 9 nm nm
RAD-3 24 9 8.2 3.6
RAD-4 14 7 nm nm
RAD-5 13 2 nm nm
MCON-1 5300 62 nm nm
MCON-2 2300 40 100 13
CONTROL-1 5 13 1.6 5.6
CONTROL-2 4 13 nm nm
aGRAV analysis is designed to measure solvent-extractable organics, most
of which have boiling points over 300°C.
bNot measured.
30
-------
TABLE 9. TOTAL CHROMATOGRAPHABLE ORGANIC (TCO) AND GRAV CONCENTRATION
RESULTS FOR XAD-COLLECTED SAMPLES FOR WELL-TUNED RADIANT
AND MALTUNED CONVECTIVE HEATERS
Code
GRAVa (ug/m3
In
Out
TCOb (uz/m3)
Out
In
RAD-1
RAD-2
RAD-3
RAD-4
RAD-5
490
360
510
450
380
190
120
120
77
48
1400
930
4900
1000
1700
150
190
370
92
130
MCON-1
MCON-2
500
360
250
220
2100
950
400
240
CONTROL-1
CONTROL-2
99
50
94
29
700
100
290
110
aGRAV analysis is designed to measure solvent-extractable organics, most
of which have boiling points over 300°C.
bTCO analysis is designed to measure solvent-extractable organics with
boiling points between 100°C and 400°C.
31
-------
For convenience and ease of reporting, results of the contribution of a kerosene heater to
indoor air pollution concentrations will be expressed as pollutant source strengths while the
heater is on. Lower-limit indoor concentrations can be calculated using the steady-state form
of Eq. 2, if the outdoor concentration is assumed to be zero. Pollutant emission rates
(expressed in units of mass of pollutants per kJ of fuel consumed) can be calculated by
dividing the source-strength values by 7000 kJ/h. Rough error-propagation analyses have
been conducted on our calculations of pollutant source strengths using approximate precision
estimates for the variables in Eq. 2.
Except where noted, the omission of a source-strength value from the paper is the result
of one of three possibilities: we did not look for the pollutant, we looked for the pollutant but
did not find it, or the pollutant source strength was not significantly different from zero. In
the last circumstance, qualitative judgments were used in some cases, e.g., when the limit of
detection of a compound was only approximately known and no outdoor concentrations were
reported.
For comparison with other source-strength tables, the source strengths for TCO, GRAV,
and TSP mass are given in Table 10.
Table 11 lists the source strengths of several nitrated-PAHs. The nitrated-PAHs searched
for are listed in the "Experimental Methods" section. One-nitronaphthalene is clearly emitted
by the well-tuned radiant and maltuned convective kerosene space heaters. One-
nitronaphthalene was collected almost entirely on the XAD for the radiant-heater tests. For
the maltuned-convective-heater tests, thirty percent of the nitronaphthalene was collected on
the filter. This is presumably a result of collection by the heavy loading of fresh soot on the
filter during the maltuned-convective test.
Emissions of 9-nitroanthracene were observed in the XAD fraction of one of the radiant-
heater tests and in the filter fraction of the maltuned-convective test. (Again, the heavy soot
loading on the filters of the convective tests may have captured the 9-nitroanthracene before it
reached the XAD.) Emissions of 1-nitropyrene were also observed in the filter fraction of
both radiant-test samples, whereas only trace amounts of 3-nitrofluoranthene were observed in
one of the two series of radiant-heater tests in the filter-collected fraction.
Both 3-nitrofluoranthene and 1-nitropyrene have been observed to be somewhat mutagenic
(Tokiwa et al.. 1985; Rosenkranz and Mermelstein, 1985), but the mutagenic activities of 1-
nitronaphthalene and 9-nitroanthracene are low (Rosenkranz and Mermelstein, 1985). Notably
missing from Table 11 are the highly mutagenic dinitropyrenes (DNPs). The 1,3-DNP; 1,6-
DNP; and 1,8-DNP combined source strengths were measured by another research team to be
approximately 0.2 ng/h (Tokiwa et al.. 1985). The estimated limit of detection in terms of
source strengths for DNPs or other nitrated-PAHs investigated for this report is 1.0 ng/h.
Future studies could take advantage of various fractionation and cleanup techniques to improve
the detection sensitivity for this class of compounds; however, such elaborate techniques were
not appropriate for this exploratory study.
Table 12 presents pollutant source-strength results for selected organic pollutants emitted
from the well-tuned and maltuned convective heaters. Although the table contains more
information than can be discussed here, two topics are of particular interest: 1) the differences
in relative source strengths among the three test/sample categories and 2) the PAH emissions.
There is a striking difference in relative source strengths between the RAD-1,2,4, and 5
tests on the one hand and the RAD-3 test on the other. The alkyl benzene emissions from the
32
-------
TABLE 10. TCO, GRAV, AND TSP MASS SOURCE STRENGTHS FOR A
WELL-TUNED RADIANT AND A MALTUNED CONVECTIVE KEROSENE HEATER
Pollutant RAD-1,2,4,5 RAD-3 MCON-1,2
Group (mg/h) (mg/h) (mg/h)
TCO-XAD 140 540 160
GRAV-XAD 42 53 38
Filter nma 0.49 6.9
Total run 53 45
TSP 1.1 1.5 270
aNot measured.
33
-------
TABLE 11. NITRATED-PAH SOURCE STRENGTHS FROM A WELL-TUNED RADIANT AND
A MALTUNED CONVECTIVE KEROSENE SPACE HEATER
RAD-1,2,4,5 RAD-3 MCON-1,2
Compounds (ng/h) (ng/h) (ng/h)
1-nitronaphthalene
XAD 280 140 260
Filter -- 3 120
TOTAL 280 140 380
9-nitroanthracene
XAD -- 53
Filter -- 3 41
TOTAL -- 56 41
3-nitrofluoranthene 1.9
(filter only)
1-nitropyrene 4.4 8.2
(filter only)
34
-------
TABLE 12. SELECTED ORGANIC POLLUTANT SOURCE STRENGTHS FROM A WELL-TUNED
RADIANT AND A MALTUNED CONVECTIVE KEROSENE SPACE HEATER
COMPOUND CLASS
RAD-1,2,4,5
(Mg/h)
RAD-3
(jJg/h)
MCON-1,2
(Mg/h)
PAH
Naphthalene
XAD-2
Filter
56
230
18
140
C2, Naphthalene
(filter only)
C3, Naphthalene
(filter only)
Phenanthrene
(XAD only)
Fluoranthene
XAD-2
Filter
1.1
1.9
0.11
16
0.84
0.07
30
4.5
5.9
1.8
Anthracene
(filter only)
Chrysene
(filter only)
Indeno(c,d)pyrene
(filter only)
0.05
0.12
2.27
Total PAH in XAD-2
Total PAH on filter
58
1.3
250
0.2
24
180
Alkyl benzenes
XAD-2
Filter
89
1.8
61000
840
17
Pentachlorophenol
XAD-2
Filter
36
0.34
48
1.1
920
(Continued)
35
-------
TABLE 12. - Continued
COMPOUND CLASS RAD-1,2,4,5 RAD-3 MCON-1,2
(Atg/h) (A»g/h) (/ig /h)
Phthalates XAD-2 1200 3300 3500
Filter 7.8 13 1500
Hydro Naphthalenes
Decalin
(XAD-2 only)
C2, Decalin
(XAD-2 only)
CI, Tetralin XAD-2
Filter
Aliphatic Hydrocarbons XAD-2
Filter
Aliphatic Alcohols XAD-2
Filter
Aliphatic Ketones XAD-2
Filter
Benzoic Acids
(filter only)
Aromatic Acid
(XAD-2 only)
Fatty Acids
(filter only)
Esters
(filter only)
300 1000 20
1800 6500 1800
700 1600 1200
160
1500 -- 2900
9.4 6.4 1400
10000 4900 4900
5.5 32 590
670 -- 4500
1.1
2.3
630
14 18 220
6.4 15 200
(Continued)
36
-------
TABLE 12.
Continued
COMPOUND CLASS RAD-1,2,4,5 RAD-3 MCON-1,2
(Mg/h) (/ig/h) (Mg/h)
Miscellaneous
CI, Cyclohexane
(XAD-2 ONLY) -- 530
C2, Methoxy Benzene
(XAD-2 only) 3000
C2, Ethenyl Benzene
(XAD-2 only) -- 680
Chlorophenyl Ethanone
(XAD-2 only) -- -- 270
Acridene - - - - 1.3
(filter only)
Methyl Propoxy Benzene - - - - 57
(filter only)
Trichloropropene 0.66
(filter only)
Aliphatic Amine -- -- 200
(filter only)
37
-------
RAD-3 test were much greater than those from the other radiant-heater tests, whereas the
aliphatic hydrocarbon and aliphatic ketone emissions from the RAD-1,2,4, and 5 tests were
much greater than those from the RAD-3 test. From an experimental point of view, all five
radiant-heater tests were identical, yet the emission spectra of the RAD-3 test are dramatically
different from the other radiant tests.
The comparison of the convective-test emissions spectra with the radiant tests reveals both
similarities and differences. The PAH, phthalate, and aliphatic-alcohol emissions for the
radiant and convective tests are very similar, yet the aliphatic ketone and, particularly, the
pentachlorophenol emissions are much greater in the convective tests than in the radiant tests.
Since pentachlorophenol was used as a calibration standard for this analysis and some
pentachlorophenol was emitted during the radiant-heater tests, we must conclude that the
convective-heater pentachlorophenol source strength presented in Table 12 is valid, although
the authors do not understand how such a compound could be produced in a kerosene flame.
Higher levels of acidic compounds such as ketones, acids, and esters were emitted from the
maltuned convective heater than from the well-tuned radiant heater. This result is expected
since such acidic compounds are indicative of incomplete combustion. Also of interest is the
observation that many SVOCs were trapped by the soot-laden filter during the convective tests.
Relatively few PAHs were observed to be emitted by the kerosene heaters using the very-
broad GC/MS scanning technique employed in this study. Other PAHs would probably be
found if a more compound-specific technique were employed. Our analysis shows naphthalene
to be the primary PAH emission from kerosene heaters. Emissions of fluoranthene and
indeno(c,d)pyrene, two slightly mutagenic compounds (Kaden et al.. 1979; National Research
Council, 1983) were also found. Previous research from a turbulent-diffusion continuous-flow
kerosene combustor showed that 18 nonvolatile or particle-bound PAHs were emitted (Kaden
et al.. 1979); naphthalene accounted for only 3% of the particle-bound PAHs. The earlier
study also found that relatively few PAHs accounted for the mutagenic activity of the kerosene-
heater soot. Of those compounds, only fluoranthene was also observed in this study. A more
specific study of PAH emissions using more sensitive techniques is warranted.
38
-------
SECTION 5
CONCLUSIONS
In general, the hood and chamber emission-rate measurement techniques yield similar
results for CO, NO, and N02. However, when discrepancies were observed, they were, we
believe, caused by differences in combustion-air oxygen levels. The chamber method results
were judged to be more realistic since this method allows the oxygen content of the
combustion air to drop, as would occur in actual residences. The hood method appears to be
adequate for quantifying CO, NO, and NO, emission rates from appliances that are not oxygen
sensitive or from appliances that marginally affect a residence's oxygen level, such as a gas
range. The chamber method was preferable for measuring total suspended particulate
emissions, primarily because it was easier to implement. No disadvantages to the chamber
method were discovered.
With regard to organic pollutant emissions from kerosene heaters, this study has confirmed
the results of other studies, i.e., that the kerosene combustion process can emit PAHs and
nitrated-PAHs. In addition, kerosene heaters were found to emit many other organic
compounds, including aliphatic hydrocarbons, alcohols, and ketones; phthalates; alkyl benzenes;
and pentachlorophenol. These results need to be correlated with health-effects data to
determine the risk associated with these organic emissions. PAH and nitrated-PAH emissions
are sufficiently important to justify additional quantitative studies; furthermore, examinations
of other organic compounds of toxicological significance and of unvented combustion sources
should be expanded.
One very important observation of this study was that some approximate values of the
indoor reactivity of SVOCs were found to be higher than 2 h"1. This implies that reactivity
rates for some SVOCs are more important than ventilation rates for determining indoor
concentrations. Clearly, this indicates that future studies must quantify the indoor reactivity
process for individual SVOCs so we can better understand and quantify indoor exposures to
these compounds.
39
-------
REFERENCES
ANSI (1982). American National Standard for Household Cooking Appliances.
American National Standards Institute, Inc., New York (ANSI Z21.1-1982).
Apte M.G., and Traynor G. W. (1986). Comparison of pollutant emission rates from
unvented kerosene and gas space heaters. In IAQ *86: Managing Indoor Air for
Health and Energy Conservation. American Society of Heating, Refrigerating and
Air-Conditioning Engineers (ASHRAE) Publications, Atlanta, GA, pp. 405-416.
Chuang J.C., Hannon S.W., Danison T.H., and Cooke W.M. (1986). Final Report on
Analytical Support for the Combustion Appliance Characterization Study. Battelle,
Columbus Division, 505 King Ave., Columbus, OH 43201-2693.
Himmel R.L., and DeWerth D.W. (1974). Evaluation of the pollutant emissions from
gas-fired ranges. Report No. 1492, American Gas Association Laboratories,
Cleveland, OH.
Howard J.B., and Longwell J.P. (1983). Formation mechanism of PAH and soot in
flames. In Polynuclear Aromatic Hydrocarbons: Formation, Metabolism and
Measurement, M. Cooke and A. J. Dennis, Eds., Battelle Press, Columbus OH, pp.
27-62.
Kaden D.A., Hites R.A., and Thilly W. G. (1979). Mutagenicity of soot and associated
polycyclic aromatic hydrocarbons to Salmonella typhimurium. Cancer Research 39.
pp. 4152-4159.
Leaderer B.P. (1983). Air pollutant emissions from kerosene space heaters. Science
218. pp. 1113-1115.
Lionel T., Martin R.J., and Brown N.J. (1986). A comparative study of combustion in
kerosene heaters. Environ. Sci. Technol. 20. pp. 78-85.
Miksch R.R., Anthon D.W., Fanning L. Z., Hollowell C.D., Revzan K., and Glanville
J. (1981). Modified pararosaniline method for the determination of formaldehyde
in air. Analvt. Chem. 53. pp. 2118-2123.
Moschandreas D.J., Relwani S.M., Macriss R.A., and Cole J.T. (1986a). A comparison
of emission rates of unvented gas appliance measured by two different methods.
Environ. Intern. 12, pp. 241-246.
Moschandreas D.J., Relwani S.M., Johnson D., and Billick I. (1986b). Emission rates
from unvented gas appliances. Environ. Intern. 12. pp. 247-253.
Mumford J.L., Harris, D.B., Williams K„ Chuang, J.C., and Cooke, M. (1987). Indoor
air sampling and mutagenicity studies of emissions from unvented coal combustion.
Environ. Sci. Technol. 21. 21. pp. 308-311.
National Research Council (1983). Polycyclic Aromatic Hydrocarbons: Evaluation of
Sources and Effects. National Academy Press, Washington, D.C., p. A-14.
40
-------
Prado G.P., Lee M.L., Hites R.A., Hoult D.P., and Howard J.B. (1973). Soot and
hydrocarbon formation in a turbulent diffusion flame. Proceedings of the 16th
International Symposium on Combustion. The Combustion Institute, Cambridge, MA, pp.
649-661.
Rosenkranz H.S., and Mermelstein R. (1985). The genotoxicity, metabolism and
carcinogenicity of nitrated polycyclic aromatic hydrocarbons. J. Environ. Sci. Health C3.
pp. 221-272.
Ryan P.B., Spengler J.D., and Letz R. (1983). The effects of kerosene heaters on indoor
pollutant concentrations: a monitoring and modeling study. Atmos. Environ. 17. pp. 1339-
1345.
Skopek T.R., Liber H.L., Kaden D.A., Hites R.A., and Thilly W.G. (1979). Mutuation of
human cells by kerosene soot. JNCI 63. pp. 309-312.
Tokiwa T., Nakagawa R., and Horikawa K. (1985). Mutagenic/carcinogenic agents in indoor
pollutants; the dinitropyrenes generated by kerosene heaters and fuel gas and liquid
petroleum gas burners. Mutuation Research 157. pp. 39-47.
Traynor G.W., Anthon D.W., and Hollowell C.D. (1982). Technique for determining pollutant
emissions from a gas-fired range. Atmos. Environ. 16. pp. 2979-2987.
Traynor G.W., Allen J.R., Apte M.G., Girman J.R., and Hollowell C.D. (1983). Pollutant
emissions from portable kerosene-fired space heaters. Environ. Sci. Technol. 17. pp. 369-
371; Addendum: (1985) Environ. Sci. Technol. 19. p. 200.
Traynor G.W., Girman J.R., Apte M.G., Dillworth J.F., and White P.D. (1985). Indoor air
pollution due to emission from unvented gas-fired space heaters. JAPCA 35. pp. 231-237.
Whitby K.T. (1976). Electrical measurements of aerosol. In Fine Particles: Aerosol
Generation. Measurement. Sampling and Analysis. Academic Press, Inc., New York, NY.
Yamanaka S., Hirose H., and Takaoa S. (1979). Nitrogen oxide emissions from domestic
kerosene-fired and gas-fired appliances. Atmos. Environ. 13. pp. 407-412.
41
-------
APPENDIX
QUALITY CONTROL EVALUATION REPORT
The analytical techniques used to measure pollutant emissions from kerosene heaters in the
study can be classified in two groups--technique for the analysis of nonorganic emissions and
techniques for the analysis of organic pollutant emissions. The nonorganic emissions were
analyzed at LBL, whereas samples of the organic emissions were sent from LBL to BCD for
analysis. Quality control is described in detail in the report.
The nonorganic emissions were measured using well-established sampling and analytical
techniques. The tests conducted to compare results from two methods (the "hood" and the
"chamber" methods) used these techniques. Since the emission rates of the nonorganic
pollutants were well established, and the methods for their measurements were well developed,
the quality of the resulting data was expected to be, and in fact was, quite good. Discussion
of this, and some exceptions, will follow.
The organic emissions, on the other hand, were not at all well known, the experiments
were exploratory in nature, and therefore a semiquantitative approach to the analysis of the
organic samples was used. Extreme care was taken at every step of the procedure, from
sampling the heater emissions, through the EI, GC/MS analysis of the extracted samples.
Ensurance that the samples were not contaminated was provided by careful attention to lab
cleanliness and washing procedures and careful storage and shipment of samples, as well as
usage of several sets of blanks. The overall accuracies of the methods used were on the order
of 30% for the NCI GC/MS analyses and approximately a factor of 3 for the EI, GC/MS.
This semiquantitative approach did, however, satisfy the goals of the project, which were to
test for and identify a broad range of these compounds.
Table 1 of the report presents the instrumentation we used, with estimates of precision and
accuracy in the measurement of the nonorganic emissions. Table 2 presents calibration
equipment and standards used for calibrating the gas analyzers as well as the flow control for
the medium-volume samplers. The calibration schedules referred to in these tables were
rigidly adhered to. If any testing of samplers or analytical equipment identified a problem,
the problem was resolved before the experiments were continued.
In the comparison of the hood and chamber techniques, the overall accuracy goals of the
project were fulfilled. The emission-rate discrepancies that occurred were, with the exception
of the TSP measurements, due to physical differences in the two methods. As discussed in the
results section, the chamber method allows the oxygen concentration to change, changing the
burner's combustion characteristics, whereas the hood method does not cause this to happen.
The TSP emission rates measured by the two methods were very different: this resulted
because the temperatures in the hood were much higher than those in the chamber, and we
hypothesize that the particles did not have time, and/or were still too hot, to fully coagulate in
the hood before being sampled.
42
------- |