ARTICLE pubs.acs.org/est
Evaluation of Mass and Surface Area Concentration of Particle Emissions and Development of Emissions Indices for Cookstoves in Rural India Manoranjan Sahu,†,‡ John Peipert,‡,§,^ Vidhi Singhal,† Gautam N. Yadama,§ and Pratim Biswas*,† †
Aerosol and Air Quality Research Laboratory, Department of Energy, Environmental and Chemical Engineering, Campus Box 1180 Washington University in St. Louis, St. Louis, Missouri 63130, United Staytes § George Warren Brown School of Social Work, Campus Box 1196, Washington University in St. Louis, St. Louis, Missouri 63130, United States ^ Foundation of Ecological Security, Gujarat 38801, India
bS Supporting Information ABSTRACT: Mass-based dose parameters (for example, PM2.5) are most often used to characterize cookstove particulate matter emissions. Particle surface area deposition in the tracheobronchial (TB) and alveolar (A) regions of the human lung is also an important metric with respect to health effects, though very little research has investigated this dose parameter for cookstove emissions. Field sampling of cookstove emissions was performed in two regions of rural India, wherein PM2.5, particulate surface area concentration in both TB and A regions, and carbon monoxide (CO) were measured in 120 households and two roadside restaurants. Novel indices were developed and used to compare the emissions and efficiency of several types of household and commercial cookstoves, as well as to compare mass-based (PM2.5) and surface area-based measurements of particle concentration. The correlation between PM2.5 and surface area concentration was low to moderate: Pearson’s correlation coefficient (R) for PM2.5 vs surface area concentration in TB region is 0.38 and for PM2.5 vs surface area concentration in A region is 0.47, indicating that PM2.5 is not a sufficient proxy for particle surface area concentration. The indices will also help communicate results of cookstove studies to decision makers more easily.
’ INTRODUCTION Over 2 billion people use biomass to cook, boil water, for heating, and other household needs,1,2 and this number is expected to increase over the next decades.3 This enormous scale of household biomass energy use demands attention from the public health and development communities given the known adverse impacts of the emissions from biomass cookstoves.4-15 For decades, governments and NGOs have attempted to disseminate improved household stoves in developing nations in order to facilitate more efficient and less harmful household cooking practices.6 In this effort, alterations have been made to traditional biomass cookstoves commonly used in rural areas, and new, nonbiomass using stoves have also been offered. In the case of India, it is recognized that the disseminated improved stoves have not met their efficiency or emissions objectives,7 and a very low fraction has been adopted for use.8 A recent evaluation by the Ministry of New and Renewable Energy (MNRE), Government of India has resulted in the launch of a new national biomass cookstove initiative.9 Extant research on cookstoves has analyzed several emissions types to investigate their health effects. Mass-based dose parameters, especially PM2.5, are used most often to characterize cookstove particle emissions.10,11 Small particles usually have low mass compared to large particles but have large surface areas and, therefore, their concentrations are not accurately reflected in the mass concentration of the particles. Studies of particle emissions from noncookstove sources indicate that the surface area (SA) of r 2011 American Chemical Society
smaller particles deposited in the lungs is an important dose parameter in determining health effects.12-14 Park et al.15 evaluated the exposure metrics by classifying them into ranks based on the different aerosol concentrations measured during cooking in kerosene or liquefied petroleum gas in residences in India. The analysis indicated that exposure ranking by mass and surface area was similar but was different when number concentrations were used, thus illustrating the importance of selecting the exposure metric most relevant in evaluating adverse health impacts.15 Ultrafine particles with diameters between 10 and 400 nm, where the deposition curves (International Commission on Radiological Protection (ICRP) model) exhibit their maxima, are of particular interest in studying health effects.16 To the best of our knowledge, surface area has not been studied as a dose metric for biomass cookstove particle emissions in the past due to the unavailability of appropriate instruments. Recent developments in design of an instrument to measure the surface area concentration of particles deposited in regions of the lung allow more health relevant measurements to be taken.17 Field measurements of cookstove efficiency and emissions often produce a large set of disparately parametrized data. Indices can standardize these disparate factors or quantities for easier Received: August 26, 2010 Accepted: January 20, 2011 Revised: December 14, 2010 Published: February 18, 2011 2428
dx.doi.org/10.1021/es1029415 | Environ. Sci. Technol. 2011, 45, 2428–2434
Environmental Science & Technology
ARTICLE
Table 1. Overall Experimental Test Plan stove operating condition test no. 1
stoves testeda
study area Andhra Pradesh & Karnataka 3-stone stove
parameters measured
measurement locationa
operation
air/fuel
PM2.5, surface area
L1, L2
steady state
normal
PM2.5, surface area, CO
L1 L1, L2, L3
start up steady state
normal normal
traditional biomass chulha traditional biomass chulha with chimney traditional chulha (outdoor) improved chulha kerosene biogas 2 3
Orissa
traditional biomass chulha traditional coal chulha improved
high
kerosene LPG biogas 4
biomass bhati
L1, L2
steady state
normal
coal bhati 5 a
water boiling test during steady-state cooking conditions of the stove
Details of the stoves and measurement locations are in Table S1 in the Supporting Information.
comprehension or comparison. To the best of our knowledge, there are presently no standard indices for the efficiency and emissions of cookstoves. McCraken and Smith18 evaluated cookstove emissions and thermal performance per standard task, which functions similarly to an index. This method provides a useful way of simplifying cookstove comparisons, though it focuses on performance across several parameters per event (cooking task), and does not summarize these different parameters in a single value. The development of new cookstove indices is useful in two ways. First, raw data can be standardized in order to create a clear “score” that will easily communicate a high or low efficiency or emissions levels. Second, indices will help to compare the total emissions of a cookstove by standardizing and combining the levels of particulate matter, CO, and other types of emissions within a single “score.” The availability of such integrated indices will also help perform correlations with other indices, such as those based on socio-economic factors, and this is the subject of a future paper by the authors. In this paper, several types of rural cookstoves are compared by analyzing the surface area concentration of their particulate emissions and examining this against a more common measure, PM2.5 concentration. Efficiency and emissions indices are proposed and employed to give holistic comparisons between 95 several types of rural cookstoves used in India.
’ METHODS AND ANALYSIS Study Area. Emissions sampling was conducted in two geographical regions of India. The state of Orissa (eastern India), and a contiguous region of Andhra Pradesh and Karnataka (southern India) were chosen to compensate for the dearth of evidence from these regions,19 and to capture variation in geographical and socio-economic contexts. A small number of households were purposively chosen in Orissa to cover the range of cookstoves used there, whereas in Andhra Pradesh and Karnataka, a stratified, random sample of 110 households were selected to get an accurate representation of the household cooking technology used.
Rural households in India in these areas predominantly use biomass stoves because of the inexpensive, readily available fuel wood and ease of maintenance. Many tree and shrub species are used by households as fuel, usually several at a time. A detailed description of the stoves tested in this study is provided in the Supporting Information section (see Figure S1 and Table S1). Emissions Measurements and Characterization. Realtime PM2.5 emissions concentrations were measured under different operating scenarios using a personal aerosol monitor (TSI SidePak AM510, St. Paul, MN, USA). The personal aerosol monitor works on the principle of light scattering in which the measured scattered intensity is linearly proportional to the mass concentration of aerosol. A UCB monitor designed at the University of California, Berkeley was also used to measure the PM2.5 concentrations.20 The real-time surface area concentration of particles deposited in the tracheobronchial (TB) and alveolar (A) regions of the lung was measured by a nanoparticle surface area monitor (NSAM; TSI AEROTRAK 9000, St. Paul, MN, USA). The nanoparticle surface area monitor (NSAM) measures surface area concentration based on diffusion charging of aerosols followed by the detection of the charged particles by an electrometer. Additional details about the principles of measurement of these devices are reported in previously published papers.21,22 Carbon monoxide was measured using a Langan Model T15n monitor (Langan products, San Francisco, USA). For evaluating the stove energy efficiency, a water-boiling test (WBT) was conducted by taking a fixed amount of water (∼2500 g) in an aluminum pot that is used in the rural areas for cooking food. The time required for boiling water (i.e., time to reach 100 C from the room temperature) when the stove is in steady-state burning conditions was recorded to calculate the efficiency of the stoves. Experimental Test Plan. Field tests were conducted in Orissa in June 2008, and in Andhra Pradesh and Karnataka between June and August 2008. The experimental plan and test conditions are summarized in Table 1. The emissions sampling was conducted for 10-15 min continuously during the different operating scenarios considered in this study. Emissions sampling 2429
dx.doi.org/10.1021/es1029415 |Environ. Sci. Technol. 2011, 45, 2428–2434
Environmental Science & Technology was conducted at three locations within the household (see Figure S1B in the Supporting Information). Location 1 (L1), or the breathing zone, approximated the position of a person using the stove (within 0.5 m of the stove). Location 2 (L2) represents the emissions experienced by people working on other household tasks while the stove is burning and for children who are near their mother while she is operating the stove. In Orissa, L2 was between 1 and 3 m away from the stove, and in Andhra Pradesh and Karnataka, L2 was between 1 and 5 m away from the stove. L2 was chosen based on the location in the household where the aforementioned activities took place as opposed to a precise distance from the stove, and it varies by household accordingly. Location 3 (L3) was between 3 and 5 m away from the stove, depending on the particular household, and represents the emissions experienced by other household members sitting or sleeping in another part of the house. In Orissa (Test 2) each stove was sampled for emissions at L1, L2, and L3. In Andhra Pradesh and Karnataka (Test 1), only L1 and L2 were sampled. Several cookstove operating conditions were tested in Orissa. Two stove operating phases were considered: start-up and steady-state (Tests 2 and 3). Start-up refers to the time directly after the stove is ignited, and steady state refers to the normal operating phase, after the start-up phase has finished (approximately 20 min after start up). These two phases of cookstove use are important to measure separately because the nonideal combustion process at start-up (e.g., moisture in the wood) impacts emissions levels, and emissions levels may change after this phase. All of the household cookstoves in the Orissa sample were tested at start-up phase in L1, and in steady-state phase at L1, L2, and L3. The objective of the tests in Orissa was to compare different types of stoves, rather than make large number of repeated measurements on the same stove. The impact of air-fuel ratio on cookstove efficiency and emissions levels was qualitatively tested for traditional stoves. Measurements were taken at the breathing zone; and values compared when using a battery operated fan to improve air-fuel mixing during combustion to that without the use of the fan (Test 3). In addition to the household stoves, commercial stoves, or bhatis, were also sampled for emissions (Test 4). This test included both biomass and coal bhatis from roadside restaurants, which were sampled for PM2.5, particulate matter surface area, and CO emissions at L1 and L2 in the steady-state phase for normal air-fuel ratio (i.e., air/fuel ratio in the actual cooking scenario). The definition of L1 is consistent between the household and commercial stove tests, but L2 in the commercial stove tests is defined as 3-5 m from the stove, the location where restaurant patrons would sit and be impacted by emissions. Finally, cookstove efficiency was tested for household cookstoves in Orissa using the water-boiling test (WBT) (Test 5). Relationship between PM2.5 (Mass) and Surface Area Concentrations. To analyze the relationship between PM2.5 and surface area concentration associated with cookstove emissions, both experimental measurements and theoretically (based on assumed size distribution parameters) calculated PM2.5 and surface area concentration values were used. The real time PM2.5 mass concentration was measured using the personal aerosol monitor as discussed earlier. As surface area characterization methods are at an early stage of development, surface area was estimated based on the number and mass concentration measurements by assuming a log-normal size distribution of the emitted aerosols with a fixed standard deviation value obtained from previous studies.13,14 Park et al.15 also measured the aerosol
ARTICLE
Figure 1. Real time PM2.5 emissions as a function of time for a traditional wood biomass stove (wood fired) compared to that of an improved biomass stove (wood fired). Initial times are during start up (from a cold condition), and times after 21 min are during steady-state operation.
active surface area by using a surface area monitor and compared with the estimated surface area concentrations. They found that the estimated surface area concentrations are 2-6 times higher than the actual surface area concentration measurements. However, in our study, a nanoparticle aerosol monitor was used for the first time for real time surface area concentration measurement of the emitted particles from different cookstoves. Details of the comparative analysis procedures are outlined in the Supporting Information section. Indices for Comparison of Stove Performance. Normalized indices to illustrate integrated performance of the cookstoves are proposed (details in the Supporting Information). The values of these indices range from 0 (best performing-high efficiency and/or low emissions) to 1 (worst performing). As 1 is considered to be worst performing, an inverse energy efficiency index is used in the discussion of effectivenss of combustion processes. The indices could be based on a single metric that is normalized (typically to a safe level for an environmental parameter, such as PM2.5). Or, simple linear combinations of emissions indices are used to determine an integrated emission index (IEI). When only PM2.5 and SA indices are considered in IEI, it is defined as overall particulate index (OPI). It should be noted that more comprehensive indices could be created with additional data (for example, measurement of additional pollutant emissions such as NOx). This is merely a demonstration of the index concept applied to cookstove comparisons, and more detailed, weighted methodologies based on health effects could be used.
’ RESULTS AND DISCUSSION Real Time PM2.5 Concentrations for Steady and Unsteady Operation of a Traditional and Improved Biomass Stove.
A comparison of the UCB and personal aerosol monitor indicates that the average value of the measurements compare well (see the Supporting Information). As the personal aerosol monitor provides more robust real time variations, the data from this instrument is discussed primarily in this paper. The PM2.5 concentration of a traditional and improved stove (Test 2) is shown in Figure 1 as a function of time. Very high mass 2430
dx.doi.org/10.1021/es1029415 |Environ. Sci. Technol. 2011, 45, 2428–2434
Environmental Science & Technology concentrations were observed during the start up conditions, with peak levels higher for the traditional stove. At steady state (after 20 min of operation), the traditional stove showed lower emissions (lower index score) than the improved stove. The improved stove is designed to use small pieces of wood that, under ideal conditions, would promote better air-fuel (small wood pieces) mixing and thus more efficient combustion. The user has to constantly feed in fresh pieces of wood, in contrast to the traditional stove, which can take a large log or branch. The study was done in the monsoon season, so the fuel wood was moist. Thus, the improved stove was constantly fed with moist wood, creating a disturbed combustion scenario (analogous to the start up conditions) for the improved stove. For the traditional stove, as a large log of wood is introduced only at stove start-up, the wood dries during the start up period, and steadystate combustion conditions are more efficient than for the improved stove (which constantly feeds in moist wood chips). These field measurements indicate the complexities involved in actual cooking conditions, and that designs and characterization generated in idealized laboratory conditions may not be representative of actual operating conditions. Comparison of PM2.5 and Surface Area Concentrations for Various Stoves. Variation in PM2.5 concentrations for several types of stoves (biogas, coal, kerosene, and LPG) is illustrated in Figure 2A. The mass concentrations of all the stoves are more or less similar after a short initial start up period (LPG being the lowest). The PM2.5 concentrations for these stoves are much lower than the biomass cookstoves (shown in Figure 1). The surface area concentration is plotted in Figure 2B for the same stoves. Clear differences in the surface area concentrations are observed, with the kerosene stove emissions having the highest values. Recent results have shown that kerosene stove emissions have a greater adverse impact on health,23 most likely due to the emission of large number concentrations of fine particles (as measured in this study). These fine particles do not contribute much to PM2.5 levels but have a large surface area that exacerbates health effects. The importance of the measurement of the appropriate metric is therefore very evident. PM2.5 and Surface Area Concentrations as Exposure Metrics. The correlation between PM2.5 concentration and surface area concentration, in either the TB or A regions, is not strong enough to justify using PM2.5 as a proxy for surface area when selecting dose parameters for cookstove particulate emissions. Detailed analysis and results are provided in the Supporting Information section. The moderate correlation between these parameters is not sufficient for information about one dose parameter to be communicated reliably by the other. Although these parameters covary at low concentrations, the surface area values at high, smaller size particle concentrations are relevant for understanding the health impact of particulate emissions of household cookstoves, and this information is missed in the PM2.5 measures. This analysis validates the use of an independent measurement of the surface area concentration, especially if this is an important metric that impacts health outcomes. Inverse Energy Efficiency Index. An inverse energy efficiency index (lower value indicating that it is more energy efficient) was calculated for the various stoves. The inverse energy efficiency index (EI) was 0.86 for the improved stove compared to 0.84 for the traditional stove. The calculated index for traditional and improved stoves studied by Zhang et al.24 was 0.87 and 0.76, respectively. The EI for traditional coal fired stove was 0.57. The EI calculated for kerosene stoves was 0.5724 and
ARTICLE
Figure 2. (A) PM2.5 and (B) surface area concentration emissions measurements from various stoves as a function of time: biogas (b), coal (3), LPG (9), and kerosene ()).
0.51,25 whereas in this study the index obtained for kerosene stove is 0.55. The discrepancies in stove efficiency obtained in this study compared to the values reported in other studies are due to variations in specific stove designs and operating conditions. Differences between stoves in design, feeding of the fuels, and fuel moisture content and control of air-flow to the combustion chamber affected stove performance. Cookstove Emissions Indices. Andhra Pradesh/Karnataka Site. The PM2.5 and surface area indices (see the Supporting Information for methodology) in the breathing zone were calculated for several types of stoves (Test 1) and are shown in Figure S4 in the Supporting Information. The PM2.5 index was highest for improved chulhas (0.56) followed by traditional chulhas without chimney (0.15) and traditional chulhas with chimney (0.12). The PM2.5 index values were low for kerosene (0.09) and biogas stoves (0.02). The surface area index was highest for 3-stone stoves (TB-0.56, alveolar-0.42) followed by outdoor stoves (TB-0.49, alveolar-0.35) and kerosene stoves (TB-0.29), but lower surface area index values were observed for traditional chulhas with (TB-0.14) and without (TB-0.21) chimneys. Other studies have reported improved stoves to have lower emissions levels than those in the present study. Naeher et al26 compared PM2.5 concentrations from gas stoves, improved biomass stoves, and open fires (biomass) measured at breathing 2431
dx.doi.org/10.1021/es1029415 |Environ. Sci. Technol. 2011, 45, 2428–2434
Environmental Science & Technology zone and in the kitchen, and these locations are roughly equivalent to L1 and L2, respectively, from the present study. These authors found PM2.5 concentrations to be highest from the open fire (481.2 μg/m3 averaged for stove users/breathing zone and 527.9 μg/m3 averaged for kitchen measurements) and lowest for gas stoves (135.6 μg/m3 averaged for stove users at breathing zone and 56.8 μg/m3 averaged for kitchen measurements), with the improved biomass stoves falling between these measurements (257.2 μg/m3 averaged for stove users at breathing zone and 96.5 μg/m3 averaged for kitchen measurements). The high PM2.5 index values for improved chulhas in the present study is due to improper combustion in the combustion chamber, which leads to the emission of significant quantities of unburnt, large-size soot particles, contributing to higher PM2.5 concentration. Several of the improved chulhas were broken, often because of a malfunctioning or missing flue. For traditional chulhas with chimneys, steady-state particulate emissions are low, indicating that the chimney is effective in removing emitted particles, which concurs with previous studies.27,28 It is notable that the PM2.5 index values are low for 3-stone, kerosene, and outdoor stoves, whereas surface area indices values are comparatively high for these stoves. The high surface area index values for 3-stone and outdoor stoves can be attributed to emissions of large quantities of ultrafine particles, which is consistent with low PM2.5 concentrations. Emissions at 1 m (L2) from the stoves were also investigated (Test 1). The PM2.5 indices for emissions at this distance for traditional chulhas without/with chimney were 0.11 and 0.12 respectively, compared to the PM2.5 index at breathing zone (L1) index values of 0.15 and 0.13. The decreasing index values at distances further from the cookstove suggest that particulate emissions disperse through windows or to other rooms. The calculated overall particulate index (OPI), a combination of the PM and Surface Area indices, is shown in Figure 3A and compared with the PM2.5 index. CO could not be measured because of the malfunction of the instrument during the field sampling and is not included in the integrated index calculation for this site. The OPI value was highest for 3-stone stoves (OPI (TB)-0.34, OPI (TB)-0.27) followed by outdoor traditional chulhas (OPI (TB)-0.31, OPI (A)-0.24). However, the PM2.5 index follows a different trend than the combined index wherein the PM2.5 index is highest for traditional chulhas without chimney followed by outdoor traditional chulhas (improved stove is excluded from this comparison as SA area index is not available for integrated index calculation). It is to be noted that the OPI (TB) for kerosene stoves was higher than for traditional chulhas without chimney, whereas the opposite obtains in the case of OPI (A). However, the PM2.5 index was lower for kerosene (0.09). The higher OPI is due to the dominant particle concentration of ultrafine particles from the kerosene stove. The higher OPI (TB) for kerosene compared to traditional chulhas without chimney is due to emissions of smaller particles from kerosene stoves, and most these particle fraction deposits in the TB region according to the ICRP model compared to the larger particles that are mostly deposited in alveolar region of lungs. Biogas is the cleanest stove according to the PM2.5 and integrated indices. Ordinary Least Squares (OLS) multiple regression was used to quantitatively compare each type of stove’s particulate emissions measured by several dose parameters.29 Our models assume that the emissions indices are linear function of the cookstoves used in households, and they are described by the following regression
ARTICLE
Figure 3. Emissions indices at breathing zone for different types of stoves tested in this study (A) Comparison of PM2.5 and overall particulate indices in Andhra Pradesh and Karnataka (B) Comparison of PM2.5 and Integrated emissions indices in Orissa (TB, index calculated for particle deposited in Tracheobronchial region; A, index calculated for particle deposited in Alveolar region).
equation EI ¼ R þ β1 TCC þ β2 TSS þ β3 OS þ β4 BS þ β5 IC þ ε
ð1Þ
where EI refers to the emissions index (PM2.5 and surface area (A and TB) indices), TCC refers to a traditional chulha with a chimney, TSS refers to a 3-stone stove, OS refers to an outdoor stove, BS refers to a biogas stove, IC refers to an improved chulha, R refers to the constant term, and ε refers to a random error term. A detailed description of the regression analysis is provided in the Supporting Information. The results are summarized in Table S2 in the Supporting Information. The F-statistic for all regression models is significant, indicating that each model explains variance in the dependent variable and that the regression tests are reliable. Regarding the PM2.5 index, cooking with a biogas stove 2432
dx.doi.org/10.1021/es1029415 |Environ. Sci. Technol. 2011, 45, 2428–2434
Environmental Science & Technology compared to a traditional chulha without a chimney is significantly associated with a 0.12 index unit (“unit” in the following interpretations for brevity) decrease in household PM2.5 concentration, and cooking with an improved chulha compared to a traditional chulha without a chimney is significantly associated with a 0.42 unit increase in the household PM2.5 concentration. Regarding the surface area indices, cooking with a 3-stone stove and outdoor stove compared to a traditional chulha without a chimney are significantly associated with a 0.35 unit and 0.28 increases in the concentration of surface area of particles deposited in the TB region respectively. Cooking with a biogas stove compared to cooking with a traditional chulha without a chimney is significantly associated with a 0.18 unit decrease in the surface area of particles deposited in the tracheobronchial region and a 0.18 unit decrease in the surface area concentration of particles deposited in the alveolar region. Interestingly, improved chulhas represent the highest PM2.5 concentration of any of the stoves. The households in our sample that owned these stoves discussed the heavy burden associated with proper maintenance of the improved chulhas and were aware that improper maintenance led to these stoves’ malfunction. Given other demands on time, stove users expressed inability to keep-up with the stoves’ maintenance needs. Improved biomass stoves for future dissemination should be constructed to require less maintenance. The increase in surface area concentration of particulate emissions associated with 3-stone and outdoor stoves indicates that the poorest households are most vulnerable to negative health impacts from cookstove emissions, as these stoves are very basic, constructed with found materials, and are typically only used among households with no capacity to obtain newer, alternative cooking technology. More research investigating the social and economic drivers of household cooking technology use should be conducted to identify the most vulnerable populations that should be targeted for future improved stove interventions. Biogas stoves are associated with lower concentrations of all particulate dose parameters, indicating that this is a relatively clean household cookstove. However, given the biogas stove’s high CO emissions from the Orissa field sample, it is important to conduct a more holistic evaluation of emissions from such stoves in the field. Orissa Site: Various Stove Operations. The emissions index values obtained for all the stove types at the breathing zone (L1) in the Orissa study site (Test 3) were estimated. The calculated indices suggest that PM2.5 emissions are highest for the improved chulha (0.61) followed by the traditional biomass chulha (0.48). For all other stoves, lower PM2.5 emissions were observed (