Subscriber access provided by SUNY DOWNSTATE
Article
Determining Particulate Matter and Black Carbon Exfiltration Estimates For Traditional Cookstove Use In Rural Nepalese Village Households Sutyajeet Inderjeet Soneja, James M. Tielsch, Frank C. Curriero, Benjamin Zaitchik, Subarna K. Khatry, Beizhan Yan, Steven N Chillrud, and Patrick N Breysse Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/es505565d • Publication Date (Web): 06 Apr 2015 Downloaded from http://pubs.acs.org on April 8, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 26
Environmental Science & Technology
1
Determining Particulate Matter and Black Carbon
2
Exfiltration Estimates For Traditional Cookstove
3
Use In Rural Nepalese Village Households
4
AUTHORS: Sutyajeet I. SonejaŦ, James M. Tielsch§, Frank C. CurrieroΤ, Benjamin ZaitchikŤ,
5
Subarna K. KhatryŢ, Beizhan YanĦ, Steven N. ChillrudĦ, Patrick N. BreysseŦ*
6 7
Ŧ
8 9
§
Department of Environmental Health Sciences, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205, USA
Department of Global Health, Milken School of Public Health and Health Services, George Washington University, Washington, DC 20037, USA Τ
10 11
Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205, USA
12 13
Ť
Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, MD 21218, USA
14
Ţ
15 16
Ħ
Nepal Nutrition Intervention Project Sarlahi, Kathmandu, Nepal
Division of Geochemistry, Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY 10964, USA
17 18
KEYWORDS:
19 20 21
Exfiltration fraction; air exchange rate; mixing factor; particulate matter; black carbon; cookstove; low resource environment; biomass burning; ventilation; village; household; Indo Gangetic Plain
ACS Paragon Plus Environment
1
Environmental Science & Technology
Page 2 of 26
22
ABSTRACT
23
A majority of Black Carbon (BC) emitted to the atmosphere in the Indo-Gangetic Plain (IGP)
24
region is from burning biomass fuel used in traditional, open-design cookstoves. However, BC
25
and particulate matter (PM) household emissions are not well characterized. Household emission
26
information is needed to develop emission profiles to validate regional climate change models
27
and serve as a baseline for assessing the impact of adopting improved stove technology. This
28
paper presents field-based household PM and BC exfiltration (amount exiting) estimates from
29
village homes in rural Nepal that utilize traditional, open-design cookstoves. Use of these stoves
30
resulted in a 26% mean PM exfiltration, ranging 6% to 58%. This is a significant departure from
31
an 80% estimate cited in previous literature. Furthermore, having a window/door resulted in an
32
11% increase in exfiltration when an opening was present, while fuel type had marginally
33
significant impact on emission. Air exchange rates (AER) were determined, with average (95%
34
CI) AER of 12 (10-14) per hour, consistent with previous studies. Also, BC to PM2.5 mass-ratio
35
composition during cooking was ascertained, with an average (95% CI) of 31% (24-39),
36
agreeing with previous biomass fuel emission composition literature.
37
ABSTRACT ART:
38 39 40 41 42 43
ACS Paragon Plus Environment
2
Page 3 of 26
Environmental Science & Technology
44
Introduction
45
The Himalayan glaciers provide fresh water via glacial melt to more than a billion people
46
across South and East Asia1. These glaciers have begun melting at an alarming rate over the past
47
decade, threatening the fresh water supply for this region1. Furthermore, alterations in the overall
48
hydrological cycle in South Asia is becoming more apparent2. Monsoons are deviating from
49
typical seasonal patterns, impacting a number of groups within this region including farmers who
50
are responsible for feeding a significant portion of the world’s population3,4. While it is believed
51
that glacial retreat is driven by global warming, the rapidity of the retreat indicate other factors
52
may be at play1. Black carbon emitted into the atmosphere and subsequent deposition on glaciers
53
and snow pack is thought to be a major contributor to the changing climate and accelerated
54
glacial retreat in this region5.
55
Black carbon (BC), a component of particulate matter (PM), is produced by incomplete
56
combustion of fossil fuels, burning of biomass from forest fires, and from households and
57
factories that use wood, dried animal manure, or crop residue for cooking or other energy needs6.
58
Worldwide, more than three-quarters of the black carbon produced is thought to come from
59
developing countries, generated from cookstoves, open burning, and older diesel engines4. It is
60
estimated that around 2.7 billion people worldwide cook using biomass fuel, most of which are
61
poor and in developing countries7. Twenty percent of the total global BC emissions has been
62
attributed to emissions of BC from biomass burning in cookstoves8, and biomass cooking is
63
believed to contribute about two-thirds of BC emissions within South Asia9.
64
Within the Himalayas, BC may be responsible for as much as 50% of the total glacial retreat
65
seen9. Furthermore, recent studies have estimated BC as a major contributor to global warming,
66
surpassed only by CO2 and possibly methane10. Since BC has an average residence time in the
ACS Paragon Plus Environment
3
Environmental Science & Technology
Page 4 of 26
67
atmosphere of a few days to weeks relative to greenhouse gases which range from years to
68
centuries, mitigating BC emissions to combat climate change can produce almost immediate
69
effects in terms of reduced radiative forcing, subsequently producing direct benefits for public
70
health10,11. Regional scale BC emission estimates, however, have a significant, 2-5 fold,
71
uncertainty5. The Indo-Gangetic Plain region (IGP) has been identified as a regional hotspot for
72
BC emissions10. The IGP is one of the most densely populated regions in the world and large
73
uncertainties exist for published BC emissions12,13.
74 75
Background. Cookstove emissions create indoor exposures and contribute to ambient air
76
pollution through passive exchanges between indoor and outdoor air (indirect venting) and
77
direct, active venting (i.e., chimney) to the outdoors. Open-design cookstoves, where combustion
78
byproducts are expelled directly into the indoor environment, result in high concentrations of
79
pollutants inside the home14–18. BC and PM emission factors for different stove designs have
80
been determined in laboratory settings19–21. These emission factors can vary by 5-6 fold
81
depending on stove design and fuel type19. In addition, the BC mass-ratio component of PM
82
under 2.5 micrometers in diameter (PM2.5) has been shown to vary from 5-52% using laboratory-
83
based test burns19. While laboratory-based studies are valuable for estimating stove emissions,
84
climate modeling requires estimates of household emissions. Characterization of cookstove
85
emissions emanating from a home requires an assessment of the PM exfiltrating from the house.
86
A fraction of PM emitted indoors during combustion will settle and deposit on indoor surfaces,
87
while the remaining will exfiltrate to the outdoor environment. Studies translating laboratory-
88
based cookstove emissions to ambient emissions from a house often use hypothetical air
89
exchange and particle deposition rates resulting in fractional exfiltration estimates. For example,
ACS Paragon Plus Environment
4
Page 5 of 26
Environmental Science & Technology
90
an 80% exfiltration estimate of PM emitted by biomass combustion19, a published value with no
91
documentation based upon hypothetical particle deposition rates has been cited without any field
92
validation. Furthermore, measurement of air exchange rates from housing in developing
93
countries is extremely limited22–24 compounding our poor understanding of how much PM may
94
be exfiltrating to the outdoor environment. Estimates of PM and BC exfiltration from homes are
95
needed to reduce uncertainty in emission inventories. Exfiltration estimates can be used to
96
evaluate the impact of cookstove interventions (i.e., use of a chimney), designed to improve
97
indoor air quality, on outdoor air quality. These estimates can further inform studies assessing
98
indoor air pollution in other homes, where exfiltration of biomass air pollution has been
99
identified as a significant contributor to indoor PM pollution in non-biomass using homes25.
100
This paper presents air exchange rates and PM exfiltration estimates from homes in rural Nepal
101
that utilize traditional, open-design cookstoves. Estimates of variability are provided for
102
exfiltration as a function of housing and fuel characteristics in a real-world setting. Furthermore,
103
this paper assesses the BC to PM2.5 mass-ratio produced by biomass cooking in order to estimate
104
BC exfiltration from homes. In combination with an assessment of indoor PM concentrations,
105
PM and BC exfiltration fractions can be used to estimate house emissions to ambient air in order
106
to better assess regional air quality and climate change impacts.
107 108
Methods
109
Study Overview. Located in Sarlahi District Nepal, this exfiltration assessment study is nested
110
under a broader parent cookstove intervention trial known as the Nepal Nutrition Intervention
111
Project - Sarlahi (NNIPS) setup by the Johns Hopkins Bloomberg School of Public Health
112
(JHSPH). This community-based, cluster randomized, step-wedge trial is characterizing indoor
ACS Paragon Plus Environment
5
Environmental Science & Technology
Page 6 of 26
113
PM exposure in ~3,000 homes from traditional and alternative cookstoves, assessing primary
114
health outcomes that include acute lower respiratory illness in children 1-36 months of age
115
(ALRI), as well as birth weight and preterm birth among newborn infants26. This cookstove
116
intervention trial is underway in 4 areas, referred to as Village Development Committees
117
(VDC’s), within the NNIPS site. Sarlahi is a rural area located in the Terai region of southern
118
Nepal (bordering Bihar State in India) and is representative of southern Nepal and most of
119
northern India with elevation approximately 200 meters above sea level26. Cooking with
120
traditional stoves, comprised of clay, mud, bricks, rice husk, and cow dung, are common in this
121
area26. The typical cooking style involves the burning of biomass fuels (wood, dried animal
122
manure, and crop residue) in traditional, open-design mud cookstoves without direct ventilation
123
to the exterior26.
124 125
Mock and Occupied Home Sampling. In order to assess exfiltration estimates of PM, a mock
126
house was built at the research site in Sarlahi District, Nepal. Based on parent trial study data, the
127
house (Figure 1) was representative of a typical household kitchen in this area with respect to
128
size, material, and number of openings. Homes in this region typically consist of mud, wood,
129
brick or cement walls with thatch roofs26. The mock house consisted of a 1-room floor plan with
Figure 1. Mock house representative of standard kitchen built in study area
ACS Paragon Plus Environment
6
Page 7 of 26
Environmental Science & Technology
130
the ability to close and open 1-window and door. Housing material consisted of bamboo with
131
mud, logs, and tree branches, while roofing was half tile and thatch/grass. House dimensions
132
were: length 3.85m, width 4.65m, ground to the lowest roof point 1.8m, ground to apex of the
133
roof 2.7m, window 0.6m by 0.6m (located on the back wall), and door frame 1.28m width by
134
1.64m height (located on the front wall). Both the window and door had a hinged, wood-framed
135
metal panel that allowed for opening/closing. In addition, a traditional mud-based cookstove with
136
two openings was built inside according to typical practices for stove construction, placed on the
137
floor of the back wall.
138
Air quality data in the mock house included PM concentration collected passively with a
139
DataRAM pDR-1000AN (Thermo Scientific, Franklin, MA), carbon monoxide concentration
140
with an EasyLog USB CO Monitor (Lascar Electronics, Eerie, PA), and relative
141
humidity/temperature using the HOBO Data Logger (Onset Corp., Bourne, MA). All devices
142
were co-located at the center of the house before, during, and after simulated cooking events
143
(discussed later), and recorded data in 10-second intervals. The pDR-1000 was zeroed in the
144
field prior to each deployment using procedures recommended by the manufacturer. The CO
145
monitor was calibrated before deployment at JHSPH using standard procedures27.
146
In addition, gravimetric PM2.5 was collected with a PM2.5 cyclone inlet (BGI, Waltham, MA)
147
on Teflon filters (37mm 2.0µm pore PTFE Membrane Filter w/ PMP ring Pel Life Sciences, Ann
148
Arbor, MI). A personal sampling pump (5400 BGI Inc., Waltham, MA) was utilized at a flow
149
rate of 4 L/min. Flow rates were recorded before and after sampling using a Drycal Flowmeter
150
(DC Light BIOS Intl., Butler, NJ). Filters were pre and post-weighed in a temperature and
151
humidity controlled room using an XP2U Microbalance (Mettler Toledo, Columbus, OH)
152
located at JHSPH. Pre-weighed Teflon filters were loaded into polypropylene filter cassettes
ACS Paragon Plus Environment
7
Environmental Science & Technology
Page 8 of 26
153
(SKC Inc., Eighty Four, PA), along with filter pads (Pall Life Sciences, Ann Arbor, MI) and
154
37mm drain discs (Model #230800 Air Diagnostics and Engineering, Harrison, ME) at the field-
155
site in a field-developed clean box to minimize contamination during assembly while in Nepal.
156
After sampling, filter cassettes were sealed in plastic bags until returned to the United States,
157
where they were disassembled and post-weighed. Ten percent duplicate samples were collected
158
and all filter weights were blank corrected.
159
Fuels utilized during simulated cooking events in the mock house were derived from survey
160
data from the parent cookstove trial. The most common types of fuel were wood and a mixture of
161
wood/dung/crop waste28. Fuel sources were obtained from the same area, by the same person,
162
and subsequently sun dried to a consistent moisture content. The same person performed all of
163
the simulated cooking events in the mock house. Each fuel combination was tested three times.
164
In addition, two boundary conditions were tested, where both the window and door were either
165
open or closed during mock house testing.
166
The same test protocol was performed in 50 occupied homes randomly selected from the 2,854
167
households in the parent cookstove trial. Homes chosen reflected typical housing based on data
168
from the NNIPS cookstove intervention trial study (i.e., have 1 window/door, similar in kitchen
169
size and composition to the mock house). Eligible homes met the criteria of only using
170
cookstoves for personal food production, agreeing to not have other sources of combustion
171
(including tobacco) ongoing in the house during measurement. Testing was performed midday to
172
minimize interference to participants’ daily routines as well as smoke infiltration from
173
surrounding homes’ cooking activities. Fuel type utilized and window/door status were observed
174
in order to allow cooking preferences of the home to remain intact. The individual in the home
175
that did the most cooking performed the simulated cook session and was supervised by local
ACS Paragon Plus Environment
8
Page 9 of 26
Environmental Science & Technology
176
staff for protocol compliance. Kitchen volume and number of windows/doors were also
177
recorded.
178 179
Exfiltration Fraction Determination. The air exchange rate (AER) and subsequent exfiltration 180
fraction (EF), or the amount of PM exiting a home
181
via natural ventilation, were determined via the
182
concentration decay method29. This method involves
183
measuring pollutant decay over time once the source
184
of the contaminant is terminated or removed in order
185 to determine passive air exchange rates between Figure 2. Hypothetical decay curves for indoor to outdoor29,30. Hypothetical decay curves for 186 PM and a non-reactive gas. 187
a non-reactive gas and PM illustrating the concentration decay method are displayed in Figure 2.
188
Assuming that the house is a simple one-room structure (i.e., no tight internal partitions), it can
189
be treated as a single zone and a mass-balance continuity equation can be used to estimate the
190
exchange of air between indoors and outdoors. For non-reactive gases (i.e., no other mechanism
191
of removal) this model assumes that the air exchange carries the pollutant from inside to out, or
192
visa versa if one is interested in infiltration. The air exchange rate (AER) is calculated assuming
193
perfect mixing conditions29 using Equation (1):
194
∗
(1)
195
Where Q is ventilation rate in m3/minutes, V is room volume (m3), C2 and C1 are concentrations
196
at various points in time, t2 and t1 are the corresponding time points (minutes), and k is a factor
197
used to account for imperfect mixing conditions. The AER, , is determined by regression of
ACS Paragon Plus Environment
9
Environmental Science & Technology
198
Page 10 of 26
Ln( ) of the non-reactive gas against ∆ time. Equation (2) illustrates the simple linear regression
199
model used to obtain this value. Ln(C) = β0 + β1×t + ε, where ε ~ N(0,σ2)
200
(2)
201
Where C is the concentration of the tracer gas, t represents time (minutes), and β1 represents the
202
AER. Ventilation studies typically utilize a tracer gas not commonly found in the environment,
203
such as sulfur hexafluoride (SF6), to measure AER’s31. Given logistical limitations regarding
204
acquisition and transportation of SF6 to the study site, carbon monoxide (CO), deemed a viable
205
substitute32, was utilized as the tracer gas. CO was suitable because the concentrations produced
206
inside the home during cooking were more than two orders of magnitude higher than
207
background.
208
Figure
3
summarizes
passive
209
exfiltration pathways for PM from a
210
house. PM exfiltrates through open
211
windows/doors as well as through
212
cracks in the roof and walls (PME in
213
Figure 3). Exfiltration for PM is
214
complicated by the fact that the decay
215 216
over time will be governed by the Figure 3. Pathways for particulate matter exfiltration from a house losses due to settling, surface
217
deposition, as well as exfiltration. Calculating the AER based on the PM decay curve will
218
therefore overestimate the real passive air exchange. To address this problem, two decay curves
219
were measured as indicated in Figure 2. The slope of the PM decay curve was calculated in the
220
same manner as the CO decay curve, utilizing Equations (1) and (2). The fraction of PM decay
ACS Paragon Plus Environment
10
Page 11 of 26
Environmental Science & Technology
221
associated with exfiltration was calculated by comparing the slopes of the decay curves (i.e., the
222
AERs) for CO and PM. Since CO is a non-reactive gas, its AER will not be affected by surface
223
deposition. As shown in Figure 2, the decay rate for PM will be faster than that for a gas. The
224
ratio of the AERs, Equation (3), provides an estimate of the fraction of PM that exfiltrates to the
225
ambient air:
EF
226
(3)
227
where SCO and SPM are the slopes of the respective decay curves, or the AERs, when calculated
228
according to Equation (2). Conceptually, if both agents had the same pathways for exfiltration
229
then the decay curves would overlap resulting in unity. Deviations from unity are accounted for
230
by the faster decay of PM due to particle losses on surfaces.
231 232
Exfiltration Fraction Test Protocol. A simulated cooking session was performed to create
233
elevated concentrations of PM and CO within the home. Prior to ignition, air-sampling
234
equipment was deployed in the mock house at a height of 1m in front from the edge of the stove
235
and 1.8m off the floor to maintain consistency between readings. Placement of sampling
236
equipment in occupied homes relative to the mock house was slightly altered due to logistical
237
limitations.
238
By generating PM and CO concentrations that were orders of magnitude higher than outdoors,
239
PM and CO infiltration from outdoor air into the home (assumed to be low relative to the indoor
240
generated contaminants) when using Equation (2) could be ignored. To determine PM and CO
241
decay curves, a modified version of the Water Boiling Test (WBT) 3.0 was used33. A cooking
242
session was simulated by bringing 2 pots of water (5 Liters each) to a rolling boil from ambient
243
temperature. The fire was then extinguished and remnant fuel removed from the house to halt
ACS Paragon Plus Environment
11
Environmental Science & Technology
Page 12 of 26
244
further emissions. Passive PM, CO, and RH sampling were initiated 30 minutes prior to the
245
cooking session and continued for an additional two hours post-fire extinguishment. Gravimetric
246
sampling for PM2.5 took place only during the active flame period. Passive PM and CO
247
concentration decays were examined only after the flame was extinguished. Equations (2) and
248
(3) were applied to the decay curves to determine the AERs for CO and associated exfiltration
249
fraction. All tests were performed during the Spring of 2012, considered the dry season as the
250
outdoor temperature is moderate to high and precipitation is minimal.
251 252
Determining the Black Carbon Fraction of PM2.5. PM2.5 filter samples were collected during
253
the active flame period of the simulated cooking session in both the mock house and occupied
254
homes in order to assess BC content of the PM using a validated optical method34. It is assumed
255
that BC is mostly contained on smaller particles (i.e., PM< 2.5 m) and it is these particles that
256
drive the impacts to climate35, therefore measuring PM over a broader size distribution is not
257
necessary.
258 259
Mixing Factor Assessment. An additional analysis examined if further adjustments were
260
necessary when determining air exchange rates. Methods, data analysis, and results are presented
261
in the Supporting Information section.
262 263
Data Analysis. Exploratory analyses included generating summary statistics for AERs, PM and
264
CO ventilation rates, exfiltration fraction by fuel type and window/door status, and BC to PM2.5
265
mass-ratios. A linear mixed effects model with a random intercept and slope was utilized to
ACS Paragon Plus Environment
12
Page 13 of 26
Environmental Science & Technology
266
examine if household factors are predictive of PM exfiltration. This regression model took the
267
form
268
EFijk = (β0+ b0k) + β1×Fi + (β2+ b2k)WDj + εijk
(4)
269
Where i, j, k index fuel type, window/door status, and house type respectively, and {b0k} are
270
independent Normal (0,σ02), {b2k} are independent Normal (0,σ22), errors {εijk} are independent
271
Normal (0,σe2), and {b0k}, {b2k}, and {εijk} are mutually independent.
272 273 274 275
The effect of fuel type (Fi) is fixed while house type is random. The effect of status of window/door during cooking (WDj) varies by house type to attain the random slope model. All statistical analyses were performed in the R Statistical Computing Environment (version 3.0.2; R Project for Statistical Computing, Vienna, Austria).
276 Results and Discussion
278
Example of decay curves for PM and CO
279
post-fire extinguishment determined in an
280
occupied house is presented in Figure 4. Both
281
curves
282
presented
283
relationship is seen for all tests performed.
284
The more rapid decay of PM indicates that
285
surface losses are significant.
286 287
mimic in
the Figure
hypothetical 2
and
curves
a similar
Figure 4. Example of PM (red) and CO (blue) decay curves post-fire extinguishment in an occupied house
277
ACS Paragon Plus Environment
13
Environmental Science & Technology
Page 14 of 26
Table 1. Summary Statistics for CO Air Exchange and Ventilation Rates, and Particulate Matter Exfiltration Fraction (%)
Carbon Monoxide Air Exchange Rate (per hour)
Carbon Monoxide Ventilation Rate (m3/min)
Particulate Matter Exfiltration Fraction (%)
Data Type
N
All Data
50
Mock House
Mean (95% CI)
Range
10th%
25th%
Median
75th%
90th%
11.9 (9.7, 14.1)
2.3-41.8
4.2
6.8
10.6
15.5
20.5
20
9.6 (5.4, 13.8)
2.3-41.8
3.8
4.7
6.8
9.7
17.9
Occupied Homes
30
13.4 (11, 15.6)
4.2-35.2
7.5
9.1
11.7
15.8
21.7
All Data
50
6 (4, 7.9)
0.6-31.5
1
2.3
4
5.6
13.8
Mock House
20
6.4 (3.6, 9.3)
1.6-28.1
2.5
3.2
4.6
6.5
12
Occupied Homes
30
5.7 (2.9, 8.4)
0.6-31.5
0.9
1.7
3.2
5.1
15.3
All Data
50
26.4 (22.4, 30.5)
5.8-57.6
10.3
14.8
22.1
37.4
45.4
Mock House
20
23 (15.9, 30.1)
6.3-57.6
8.9
13.6
19
28.2
45.3
Occupied Homes
30
28.7 (23.7, 33.6)
5.8-53.6
10.7
16.6
30.9
38.7
45.4
288 289 290 291 292 293 294
ACS Paragon Plus Environment
14
Page 15 of 26
Environmental Science & Technology
295
Air Exchange, Ventilation Rate, and Exfiltration Fraction. Summary statistics of AERs,
296
ventilation rates, and exfiltration fraction, with AER and ventilation rates calculated from the
297
decay curves, are provided in Table 1.
298
Air exchange rates based on the CO decay curves are reported in Table 1, with a mean (95%
299
CI) for all tests equivalent to 11.9 h-1 (9.7, 14.1). Examining CO AER by window/door status
300
yielded an average of 5.9 h-1 (4.4, 7.3) for the closed category (n=10) and 13.4 h-1 (10.9, 16) for
301
having an opening present (n=40). This trend follows expectations for window/door status, with
302
an opening present allowing for increased airflow and hence a higher average AER. Limited
303
field studies examining AER in other village housing have been conducted. Measured AER for
304
Indian village homes have been recorded between 7 to 30 per hour22. A more recent study in
305
village homes located in Peru reported a range of AER’s between 0.5 h-1 to 20 h-1 depending
306
upon partition presence23. These ranges of values are in agreement with results found in rural
307
Nepal village housing, which has similar housing style to much of the Indo-Gangetic Plain
308
region. In addition to demonstrating similarities across village housing, these findings contribute
309
to the expansion of field-based residential AER data. Accurate estimation of AERs is critical to
310
understanding cookstove contributions to ambient air pollution as well as the broader climate
311
change impacts that may be associated with traditional vs. alternative stoves. Ventilation rates,
312
which are a function of the home volume and air exchange rate, are also presented for all tests in
313
Table 1. The ventilation rates calculated from CO decay for the mock house under varying
314
conditions averaged 6.4 m3/min and ranged from 1.6 to 28.1 m3/min. Occupied homes’ CO
315
ventilation rates similarly averaged 5.7 m3/min and ranged from 0.6 to 31.5 m3/min. As
316
expected, mock house ventilation rates were similar to the occupied homes.
ACS Paragon Plus Environment
15
Environmental Science & Technology
Page 16 of 26
317
The average (95% CI) PM exfiltration fraction for all tests (n=50) was 26.4% (22.4, 30.5),
318
with values ranging from 5.8% to 57.6%. Mock and occupied home EF averages (95% CI) were
319
23% (15.9, 30.1) and 28.7% (23.7, 33.6), respectively. EFs for the mock and occupied houses
320
were not statistically different, thus conforming to the expectation that the designed mock house
321
kitchen is representative of a typical kitchen in this area. The average (95% CI) EF for closed
322
window/door status (n=10 tests) was 18.2% (13.1, 23.2), while having an opening present (n=40
323
tests) resulted in an increase to 28.5% (23.8, 33.2). Furthermore, the range of EF data is less
324
variable when the window/door are both closed relative to having an opening present.
325
Comparing open vs. closed conditions yielded statistically different results. When partitions are
326
closed, airflow through the kitchen is reduced, resulting in less exfiltration and a range of values
327
with a relatively smaller spread in contrast to having openings.
328
EF distributions broken down by fuel type are presented in Figure 5. A wide range of fuel
329
types were tested in the occupied homes because these were the fuel types used by the families
330
during the testing period. They included wood, crop residue, mixtures of wood/dung/crop
331
residue, dung/wood, and crop residue/wood. Only wood and mixture of wood/dung/crop residue
332
were evaluated in the mock house because these were the two fuel types most prevalent in the
333
parent cookstove trial. Fuel categories consisting of a mixture still had wood as the primary fuel
334
source. EF for fuel types of wood and crop residue by themselves had similar averages (95% CI)
335
of 30.2% (25.6, 34.8) and 28.1% (13.9, 42.4), respectively.
ACS Paragon Plus Environment
16
Page 17 of 26
Environmental Science & Technology
Figure 5. Particulate matter exfiltration fraction presented by fuel type 336
Crop residue is a mixture of materials with the potential to produce a wider range of particle
337
sizes when burned, hence the potential for particle size distribution to be broader relative to
338
wood. Subsequently, one expects that EF variability for crop residue should be larger than that
339
seen for wood, which is shown in Figure 5.
340
Figure 5 has a line indicating where 80% exfiltration is located, which is a hypothetical value
341
with no field validation that has been previously cited but not independently published19. The
342
80% exfiltration is significantly greater than the averages and upper limits of the EF data,
343
presented in the overall EF data (Table 1) and in Figure 5. It is difficult to account for the
344
difference between the EF presented in this paper and the 80% estimate previously used19
345
because the assumptions made in the derivation of the theoretical 80% EF value have not been
346
published. One possible reason why the hypothetical estimate is greater than our measured
347
estimates may be that the actual mass median size of particulate matter generated from cooking
348
is larger than previously assumed, which can impact the assumed deposition rate. It should be
349
noted that temperature for all tests were measured, with no significant difference found and thus
350
considered not a factor.
ACS Paragon Plus Environment
17
Environmental Science & Technology
Page 18 of 26
351
Utilizing a linear mixed effects model, factors that may impact EFs were evaluated. Pooling of
352
categories for fuel type was necessary due to the small sample sizes of certain fuel categories.
353
With wood being the dominant fuel source for the categories of dung/wood, wood/dung/crop
354
residue, and crop residue/wood, these three sources were combined into one category and called
355
wood/other. The regression model examined window/door status (everything closed vs. opening
356
present) and fuel type (wood only and wood/other), with crop residue excluded due to the small
357
sample size. Reference levels for the covariates were set as wood for fuel type and everything
358
closed for window/door status. Regression results indicate that having an opening present
359
resulted in an 11.37% (SE=4.51) increase in exfiltration (p=0.022), while switching fuel type
360
from wood to wood/other resulted in a 7.64% (SE=3.83) decrease in exfiltration (p=0.0621).
361
Thus, window/door status was deemed a significant covariate while pooled fuel type was
362
marginally significant.
363 364
Ratio of Black Carbon to PM2.5 Analysis. Of the 50 PM2.5 filters collected during exfiltration
365
fraction tests, 43 were overloaded with PM thus too saturated to estimate BC using the optical
366
analysis method. The excluded filters exceeded the optical method’s upper threshold detection
367
limit which results in under prediction of BC values34,36. Fuel categories for the 7 samples with
368
acceptable loadings for BC analysis were broken down into wood/other (n=6) and crop residue
369
(n=1).
370 371 372 373
ACS Paragon Plus Environment
18
Page 19 of 26
Environmental Science & Technology
Table 2. Summary Statistics for Black Carbon to PM2.5 Mass-Ratio N
Mean (95% CI)
Median
Range
PM2.5 Concentration (ug/m3)
7
4206.6 (2342.1, 6071)
3427
2341.2-8382.3
Black Carbon Concentration (ug/m3)
7
1317.5 (671.9, 1963.2)
1197.5
630.8-2470.7
Black Carbon to PM2.5 Ratio (%)
7
31.2 (23.8, 38.6)
29.5
17.6-41.3
374 375
Concentration summary statistics of BC and PM2.5, along with ratio estimates are provided in
376
Table 2. Measured only over the active cooking period of the simulated cooking test, the mean
377
(95% CI) ratio was 31.2% (23.8, 38.6), ranging from 17.6% to 41.3%. For the oversaturated
378
filters that were excluded, the most that can be reported is that all the mean BC to PM2.5 ratios
379
were larger than 9%.
380
Multiple studies have reported BC to PM ratios. For example, Rau (2007) reported elemental
381
carbon values from wood smoke particles ranging from 3 to 38%37, while other studies of wood
382
smoke have reported ratios that range from 8 to 53%38. Our estimates agree well with these
383
published data, further validating estimates determined in laboratory and field settings.
384
This research has major implications with regards to epidemiological and climate change
385
models that utilize emission inventories of BC generated by cooking activities. Specifically, this
386
research contributes to developing more robust estimates of the amount of BC that exits the
387
home and into the atmosphere as a result of cooking indoors with biomass fuels in traditional
388
stoves. To better understand the black carbon exfiltration from a cooking event, simple
389
calculations can be conducted. To present upper and lower bounds, the range of average PM2.5
390
concentration values during cooking (1.8 mg/m3 to 17.3 mg/m3) in the mock house were
ACS Paragon Plus Environment
19
Environmental Science & Technology
Page 20 of 26
391
extracted. These values were multiplied by the volume of the mock house (40 m3) and BC/PM2.5
392
ratio (31.2%), producing a range of 23 to 216 milligrams of black carbon generated indoors over
393
the cooking event (mg of BC/cooking event). Subsequently, the mass of black carbon generated
394
indoors over the cooking event was then multiplied by the average exfiltration fraction (26.4%),
395
yielding the mass of black carbon that exfiltrates to the atmosphere over the course of a cooking
396
event (6 to 57 mg/cooking event). This range illustrates a rough estimate of the amount of black
397
carbon that can be emitted into the atmosphere from a household per cooking event after
398
accounting for deposition indoors.
399 400
Limitations. There are a number of limitations for this work. While we tried to keep combustion
401
characteristics constant in the mock house, burn rate, air-fuel ratio, flame turbulence, and
402
combustion temperature were not measured. These factors can influence particle size
403
distribution, which was also not measured. Differences in particle size will impact the settling
404
and fraction of PM that may deposit in cracks and on other surfaces. Furthermore, BC particles
405
may deposit at higher or lower rates than PM2.5 overall, but this issue was not measured. Future
406
work should also seek to address the incorporation of activity patterns in real-world homes to
407
assess if multiple family members within a home create changes in air exchange rate estimates.
408
Seasonality may also play a role with certain fuel types being more prevalent and is an area
409
requiring further exploration as sampling took place during one season (Spring). Due to
410
extremely high PM concentrations, oversaturation for the BC analysis occurred, thus future
411
studies should consider utilizing methods with pumps set to lower flow rates (i.e., 1.5 LPM). In
412
addition, all data collected for BC analysis was conducted utilizing a simulated cooking event
413
(water boil test) and further assessment over the course of actual cooking events is needed to
ACS Paragon Plus Environment
20
Page 21 of 26
Environmental Science & Technology
414
assess the generalizability of the mock house testing. A limitation of using CO as the tracer gas is
415
that the fire had to be extinguished so that impact of the cooking fire on AER could not be
416
determined. Finally, our testing was limited to one area and one stove type, thus results may not
417
extrapolate to other areas and stoves. Data are needed on different housing stock, stoves
418
(traditional and improved), and analysis is also needed to assess the impact chimneys may have
419
on exfiltration.
420 421
Author Contributions
422
The manuscript was written through contributions of all authors. All authors have given approval
423
to the final version of the manuscript. Specific contributions were as follows: SonejaŦ designed
424
the study, conducted primary data collection and sample processing, performed data analysis,
425
and was the main author of the paper; KhatryŢ assisted with primary data collection; YanĦ and
426
ChillrudĦ assisted with black carbon sample processing and analysis; CurrieroΤ and ZaitchikŤ
427
assisted with data analysis; BreysseŦ and Tielsch§ assisted in the design of the study, provided
428
oversight to data collection, and assisted with data analysis. All coauthorsŢĦΤŤŦ§ assisted with
429
writing of this paper.
430 431
Corresponding Author
432
Patrick N. Breysse, Ph.D.
433
Department of Environmental Health Sciences
434
Bloomberg School of Public Health
435
Johns Hopkins University
ACS Paragon Plus Environment
21
Environmental Science & Technology
436
615 North Wolfe Street, Room E6628
437
Baltimore, MD 21205
438
Phone: 1-410-955-3608
439
Fax: 1-410-955-9334
440
[email protected] Page 22 of 26
441 442
Funding Sources
443
Funding sources included a grant from The Environment, Energy, and Sustainability Institute at
444
Johns Hopkins, grant support from the National Institute of Environmental Health Sciences
445
(ES015558; ES003819), and the Thrasher Research Fund (02830-4).
446 447
Acknowledgements
448
We would like to thank NNIPS field research supervisors and staff, Steven LeClerq, Pramod
449
Sah, Samjhana Magar, Ana Rule, D’ann Williams, Jana Mihalic, Lance Wallace, Andrew
450
Persilly, and Chen Chen. This manuscript is dedicated to Dr. Alison Geyh, a mentor and
451
inspiration.
452
Supporting Information
453
An additional analysis examining mixing factors assessed if further adjustment was necessary
454
when determining air exchange rates. This information is available free of charge via the Internet
455
at http://pubs.acs.org/.
ACS Paragon Plus Environment
22
Page 23 of 26
Environmental Science & Technology
456
Abbreviations
457
BC, black carbon; PM, particulate matter; PM2.5, particulate matter with aerodynamic diameter
458
of 2.5 micrometers; EF, exfiltration fraction; AER, air exchange rate; IGP, Indo Gangetic Plain
459 460
References
461 462
1.
Xu B, Cao J, Hansen J, Yao T, Joswia DR, Wang N, Wu G, Wang M, Zhao H, Yang W. Black soot and the survival of Tibetan glaciers. Proc. Natl. Acad. Sci. 2009;106(52):22114.
463 464
2.
Ramanathan V. Aerosols, Climate, and the Hydrological Cycle. Science 2001;294:21192124. doi:10.1126/science.1064034.
465 466
3.
Ramanathan V, et al. Atmospheric Brown Clouds Regional Assessment Report with Focus on Asia. 2008. Available at: http://tinyurl.com/68r7mpd.
467 468
4.
Schmidt CW. Black Carbon: The Dark Horse of Climate Change Drivers. Environ. Health Perspect. 2011;119(4):A172.
469 470
5.
Ramanathan V, Carmichael G. Global and regional climate changes due to black carbon. Nat. Geosci. 2008;1(4):221–227.
471 472
6.
Wallack JS, Ramanathan V. The Other Climate Changers: Why Black Carbon and Ozone Also Matter. Foreign Aff. 2009;88:105.
473 474
7.
WHO. WHO | Indoor air pollution and health. 2011. Available http://www.who.int/mediacentre/factsheets/fs292/en/. Accessed October 14, 2011.
475 476 477
8.
Bond TC, Streets DG, Yarber KF, Nelson SM, Woo JH, Klimont Z. A technology-based global inventory of black and organic carbon emissions from combustion. J Geophys Res 2004;109(D14):D14203.
478 479
9.
Adler T. Better Burning, Better Breathing: Improving Health with Cleaner Cook Stoves. Environ Health Perspect 2010;118(3):A124-A129.
480 481 482 483
10. USAID/ECO-Asia CDCP. Black Carbon Emissions in Asia: Sources, Impacts, and Abatement Opportunities. 2010. Available at: http://www.cleanenergyasia.net/library/black-carbon-emissions-asia-sources-impacts-andabatement-opportunities.
484 485 486 487
11. Anenberg SC, Schwartz J, Shindell D, Amann M, Faluvegi G, Klimont Z, JanssensMaenhout G, Pozzoli L, Van Dingenen R, Vignati E, Emberson L, Muller NZ, West JJ, Williams M, Demkine V, Hicks WK, Kuylenstierna J, Raes F, Ramanathan V. Global Air Quality and Health Co-benefits of Mitigating Near-Term Climate Change through Methane
at:
ACS Paragon Plus Environment
23
Environmental Science & Technology
Page 24 of 26
488 489
and Black Carbon Emission Controls. Environ. Health Perspect. 2012;120(6):831-839. doi:10.1289/ehp.1104301.
490 491 492
12. Sasser E, et al. US EPA Report to Congress on Black Carbon: Department of the Interior, Environment, and Related Agencies Appropriations Act, 2010. US EPA; 2012. Available at: http://www.epa.gov/blackcarbon/2012report/fullreport.pdf.
493 494
13. Rehman IH, Ahmed T, Praveen PS, Kar A, Ramanathan V. Black carbon emissions from biomass and fossil fuels in rural India. Atmos Chem Phys Discuss 2011;11:10845–10874.
495 496 497 498
14. Balakrishnan K, Sankar S, Parikh J, Padmavathi R, Srividya K, Venugopal V, Prasad S, Pandey VL. Daily average exposures to respirable particulate matter from combustion of biomass fuels in rural households of southern India. Environ. Health Perspect. 2002;110(11):1069.
499 500 501
15. Bruce N, Perez-Padilla R, Albalak R. Indoor air pollution in developing countries: a major environmental and public health challenge. Bull. World Health Organ. 2000;78(9):1078– 1092.
502 503
16. Reid H, Smith K. Indoor Smoke Exposures from Traditional and Improved Cookstoves: Comparisons among Rural Nepali Women. Mt. Res. Dev. 1986:293-303.
504 505 506
17. Begum BA, Paul SK, Dildar Hossain M, Biswas SK, Hopke PK. Indoor air pollution from particulate matter emissions in different households in rural areas of Bangladesh. Build. Environ. 2009;44(5):898-903. doi:10.1016/j.buildenv.2008.06.005.
507 508
18. Smith KR. Air Pollution and Rural Biomass Fuels in Developing Countries: a Pilot Village Study in India and Implications for Research and Policy.
509 510
19. Venkataraman C. Residential Biofuels in South Asia: Carbonaceous Aerosol Emissions and Climate Impacts. Science 2005;307(5714):1454-1456. doi:10.1126/science.1104359.
511 512 513
20. National Risk Management Research Laboratory RTP, NC. Greenhouse Gases From SmallScale Combustion Devices In Developing Countries: Phase IIA Household Stoves in India. 2000.
514 515 516
21. MacCarty N, Ogle D, Still D, Bond T, Roden C. A laboratory comparison of the global warming impact of five major types of biomass cooking stoves. Energy Sustain. Dev. 2008;12(2):56–65.
517 518
22. Smith KR. Biofuels, Air Pollution, and Health: A Global Review. New York: Plenum Press; 1987.
519 520
23. Williams PRD, Unice K. Field Study of Air Exchange Rates in Northern Highlands of Peru. Environ. Forensics 2013;14(3):215-229. doi:10.1080/15275922.2013.814182.
521 522
24. McCracken JP, Smith KR. Emissions and efficiency of improved woodburning cookstoves in Highland Guatemala. Environ. Int. 1998;24(7):739–747.
ACS Paragon Plus Environment
24
Page 25 of 26
Environmental Science & Technology
523 524 525 526
25. Salje H, Gurley ES, Homaira N, Ram PK, Haque R, Petri W, Moss WJ, Luby SP, Breysse P, Azziz-Baumgartner E. Impact of neighborhood biomass cooking patterns on episodic high indoor particulate matter concentrations in clean fuel homes in Dhaka, Bangladesh. Indoor Air 2014;24(2):213-220. doi:10.1111/ina.12065.
527 528 529 530
26. Tielsch JM, Katz J, Zeger SL, Khatry SK, Shrestha L, Breysse P, Checkley W, Mullany LC, LeClerq SC. Designs of Two randomized, community-based trials to assess the impact of alternative cookstove installation on respiratory illness among young children and reproductive outcomes in rural Nepal. BMC Public Health 2014;14(1):1271.
531 532 533
27. Torrey CM, Moon KA, Williams DAL, Green T, Cohen JE, Navas-Acien A, Breysse PN. Waterpipe cafes in Baltimore, Maryland: Carbon monoxide, particulate matter, and nicotine exposure. J. Expo. Sci. Environ. Epidemiol. 2014. doi:10.1038/jes.2014.19.
534 535
28. Soneja S. Preliminary Data Analysis on Housing and Fuel Characteristics for Sarlahi District, Nepal. 2011.
536 537
29. Sherman MH. Tracer-gas techniques for measuring ventilation in a single zone. Build. Environ. 1990;25(4):365–374.
538 539
30. ACGIH. Industrial Ventilation: A Manual of Recommended Practice. 2010th ed. Cincinnati, OH: ACGIH; 2010.
540 541 542 543
31. Moore R. Tracer Gas Testing Applications for Industrial Hygiene Evaluations. EHS Today 2004. Available at: http://ehstoday.com/industrial_hygiene/instrumentation/ehs_imp_36992/. Accessed October 27, 2011.
544 545 546 547 548
32. Samfield M. Project Summary: Air Infiltration Measurements Using Tracer Gases: A Literature Review. Research Triangle Park, NC: United States Environmental Protection Agency; 1995. Available at: http://nepis.epa.gov/Exe/ZyPDF.cgi/30003X2I.PDF?Dockey=30003X2I.PDF. Accessed April 3, 2014.
549
33. Bailis R, Ogle D, MacCarty N, Still D. The Water Boiling Test (WBT) Version 3.0. 2007.
550 551 552 553
34. Yan B, Kennedy D, Miller RL, Cowin JP, Jung K, Perzanowski M, Balletta M, Perera FP, Kinney PL, Chillrud SN. Validating a nondestructive optical method for apportioning colored particulate matter into black carbon and additional components. Atmos. Environ. 2011;45(39):7478-7486. doi:10.1016/j.atmosenv.2011.01.044.
554 555 556
35. Shrestha G, Traina SJ, Swanston CW. Black Carbon’s Properties and Role in the Environment: A Comprehensive Review. Sustainability 2010;2(1):294-320. doi:10.3390/su2010294.
557
36. Yan B. Personal Communication: Validity of Optical Method. 2014.
ACS Paragon Plus Environment
25
Environmental Science & Technology
Page 26 of 26
558 559
37. Rau JA. Composition and Size Distribution of Residential Wood Smoke Particles. Aerosol Sci. Technol. 1989;10(1):181-192. doi:10.1080/02786828908959233.
560 561 562
38. Just B, Rogak S, Kandlikar M. Characterization of Ultrafine Particulate Matter from Traditional and Improved Biomass Cookstoves. Environ. Sci. Technol. 2013:130319140434009. doi:10.1021/es304351p.
563
ACS Paragon Plus Environment
26