The US Environmental Protection Agency (EPA) initiated the Small Engine Exposure Study (SEES) to evaluate potential exposures among users of small, gasoline-powered, non-road spark-ignition (SI) lawn and garden engines. Equipment tested included riding tractors, walk-behind lawn mowers, string trimmers, and chainsaws. Personal and background air quality measurements were collected on equipment operators for carbon monoxide (CO), particulate matter ⩽2.5 μm in aerodynamic diameter (PM2.5), volatile organic compounds (VOCs), and aldehydes. PM2.5 measurements included continuous and integrated mass, elemental and organic carbon (EC/OC), and trace metals. Aldehyde measurements included speciation for formaldehyde and acetaldehyde. The results demonstrated that equipment operators can experience elevated exposures to CO, PM2.5 and air toxics while operating these engines. Ten-second average CO personal exposures spanned over two orders of magnitude, with short-term concentrations exceeding 120 p.p.m. for some engine applications tested. PM2.5 concentrations averaged over each engine test period also spanned two orders of magnitude. The results also suggest that health standards, such as the CO and PM2.5 National Ambient Air Quality Standards (NAAQS), may be exceeded for certain equipment types under certain operating scenarios. Aldehyde measurements suggested exposures from primary engine emissions that exceed typical ambient concentrations, but do not exceed occupational health standards. Continuous exposure measurements illustrated the important role of the operator's activity and environmental conditions in affecting exposure levels.
A large segment of the population uses small, gasoline-powered spark-ignition (SI) lawn and garden equipment on a regular basis. This equipment includes lawn mowers, string trimmers, and chainsaws. Emissions from many of the small SI engines powering this equipment may lead to elevated air pollution exposures for a number of gaseous and particulate compounds, especially for individuals whose occupations require the use of these engines daily, such as landscapers. Emission inventory data in the United States, Canada, and Australia suggest that small engine emissions may account for over 10% of carbon monoxide (CO) and hydrocarbon (HC) emissions in certain regions (Yumlu, 1994; Priest et al., 2000; United States Environmental Protection Agency, 2004). The US Environmental Protection Agency (EPA) estimates that small gasoline-powered engines will account for a significant portion of the mobile source CO, HC, and particulate matter (PM) inventories within the next 20 years.
Emission studies with lawnmowers suggest a potential for high exposures during the equipment operation (Gabele, 1997; Priest et al., 2000). In addition, Bunger et al. (1997) reported elevated CO personal measurements related to chainsaw use, with short-term concentrations exceeding 400 p.p.m. for certain cutting activities. The EPA initiated the Small Engine Exposure Study (SEES) to evaluate the potential range of exposures from the use of small, SI engines. In addition, the study attempted to identify factors that lead to elevated exposures to air pollutants, and estimate the contribution of engine exhaust emissions on personal exposures.
The study consisted of measuring personal exposures to air contaminants while operating small, SI lawn and garden equipment. Table 1 lists the eight engine applications evaluated for the study: two commercial-grade lawnmowers, two residential-type push lawnmowers, two chainsaws, and two string trimmers. All engines evaluated in the study were certified to meet EPA's final Phase 2 standards for SI, non-road engines (United States Federal Register, 1999). The engines were purchased new in 2004, and aged for a minimum of 3 h as per manufacturer's specifications, if applicable, before any exposure monitoring. The engines chosen represent popular brands of low emission designs currently being phased in by manufacturers. Testing these engines provided exposure data on lawn and garden equipment that will likely represent the US market after full implementation of the Phase 2 emission standard program.
The lawn and garden equipment operator performed typical jobs appropriate for the equipment tested. Each test lasted for one complete tank of gas, with operating times ranging from 30 min for the chainsaws, to slightly over 2 h for the commercial lawnmowers. Conventional, unleaded summer-grade gasoline was used for each test. For the string trimmers and chainsaws, the gasoline was mixed with two-stroke engine oil (Pro-Mix p/n 54001) as per manufacturer's specifications (50:1 ratio of gasoline to oil). The amount of the fuel used for each test was determined by weight and volume before testing. The equipment operator did not perform any fuel-related activities during the study. To simulate the different types of operation represented in the small-engine emission certification duty cycles, the lawnmowers were operated using a cycle of 10 min of normal operation and 1 min of idling, while the string trimmers and chainsaws were operated using a cycle of 5 min of normal operation and 1 min of idling. A tachometer measured the engine operating speed (as revolutions per minute (r.p.m.)) for each test as a quality-assurance check to identify any changes in engine performance. Each engine was tested twice, once in the morning and once in the afternoon, to determine potential meteorological effects on exposures.
During field testing, the operator wore a vest containing personal samplers for continuous CO (Langan Model T15), continuous PM2.5 (TSI Inc. Sidepak Model AM510), integrated PM2.5 mass (Personal Exposure Monitor (PEM), MSP Corp. Model 200), integrated PM2.5 EC/OC (URG Inc. URG-2000-25A), volatile organic compounds (VOCs) (3 M Corp. Organic Vapor Monitor Model 3500), and Aldehydes (SKC Inc. DNPH-silica cartridge). The two integrated PM2.5 sampling devices allowed for the determination of gravimetric mass and metals from one filter, and EC/OC on the other filter.
Concurrent measurements were collected for background pollutant concentrations, using the same sampling instruments collecting personal samples. The background site was located approximately 50 m upwind of the study site. Study personnel rotated the personal and background continuous monitors throughout the study to reduce the potential for sampler bias. Co-located measurements for each sampler type were collected at the background site. A federal reference method (FRM) PM2.5 sampler (Partisol, Rupprecht and Patashnick, Inc.) and VOC summa canister were also located at the background site.
The background site also contained meteorological monitoring devices that collected wind speed, wind direction, temperature, pressure, and relative humidity data over 10 s averaging periods. A wind direction indicator (wind sock or ribbon, depending on the application) placed in the vicinity of the field-testing area insured that the operator remained downwind of the background site during sample collection. If a change in wind direction resulted in the operator being upwind of the background site, testing would have stopped. However, this did not occur during the study. The wind indicator also allowed an evaluation of any microscale variations in wind direction between the background site and the operator. Video recordings of the equipment operation, wind direction, and background site were collected to assist in data interpretation.
All applicable monitoring instruments were laboratory, and field, calibrated for the study. For the CO monitors, zero, 20 and 60 p.p.m. calibrations were performed using EPA Protocol II standard gas mixtures of CO in air (Scott Specialty Gases) before field deployment. All CO monitors were zero and span checked after each field test to ensure performance within the manufacturer's specifications. CO concentration results were corrected to temperatures measured at the background site. The continuous PM2.5 sampler was factory calibrated before use in the study. All sampling pumps were calibrated before and after each field test, using a NIST-traceable flow calibrator (BIOS International DryCal DC-2).
Integrated sample analysis included PM and VOC constituents. PM2.5 gravimetric mass samples were collected on 37 mm Teflon filters (Teflo™, 2 μm pore size, Pall Corp.), and measured on a balance, sensitive to 0.1 μg, located in a temperature- and humidity-controlled clean room. All filters were conditioned for a minimum of 24 h before analysis on the balance (pre- and postweight measurements). Two laboratory and two field filter blanks were analyzed for each test day. Field blanks were handled and installed in the personal and FRM samplers, respectively, at the test site, but the samplers were not operated.
After the gravimetric assessment, the integrated PM2.5 Teflon filters were analyzed by energy-dispersive X-ray fluorescence (EDXRF). First, the filters were analyzed in a helium atmosphere, and the EDXRF quantified 47 elements ranging from Al to Pb. For quality control purposes, 10% of the samples were reanalyzed in a vacuum environment, with a Kevex 771 EDXRF spectrometer. Comparison of the spectrometers resulted in r2 values greater than 0.98 for the abundant elements Al, Si, P, S, Cl, K, Ca, Ti, Mn, and Fe.
Separate prefired quartz fiber filters, followed by a polyurethane foam (PUF) cartridge, were used to collect PM for elemental carbon/organic carbon (EC/OC) content analysis. These filters were hole-punched, and EC/OC measurements were made by Sunset Laboratories (Hillsborough, NC, USA) on a Sunset Labs carbon analyzer using the NIOSH 5040 protocol (Birch and Cary, 1996).
The OVM badge samples were analyzed for benzene, toluene, ethylbenzene, and xylenes, using a gas chromatograph/mass spectrometer (GC/MS) as reported by Chung et al. (1999) for ambient outdoor and personal sampling conditions. Aldehyde samples were collected on DNPH-silica cartridges and analyzed using High Performance Liquid Chromatography (HPLC) for formaldehyde and acetaldehyde by using guidelines from EPA Compendium Method TO-11A. A potassium iodide (KI) scrubber was used to avoid potential interference from ozone.
Results and discussion
Personal monitor measurements provided information on integrated and continuous exposures to gaseous and particulate pollutants during the operation of lawn and garden equipment. The following sections present results of CO, PM, and air toxic measurements.
Figure 1 presents boxplots of the distribution of 10 s average CO concentration measurements for each engine application evaluated. The figure shows that short-term measurements can vary over two orders of magnitude. No consistent trends were identified indicating significant differences between morning and afternoon testing.
As shown in Figure 1, chainsaw operation resulted in the highest maximum and median CO exposure levels; whereas string trimmer operation resulted in the lowest median CO exposures. Although the chainsaw operations resulted in the highest CO concentrations, the levels recorded were lower than those reported by Bunger et al. (1997). These results suggest that the implementation of the Phase 2 standards resulted in the reduction of CO exposures for the chainsaw operations.
Continuous personal CO measurements suggested that concentrations varied depending on activity factors. Figure 2 provides continuous plots of the 10 s CO concentrations experienced by the operator of each equipment type during the morning or afternoon test that resulted in the highest average exposure. The measurements show the changes in concentrations resulting from activity. For example, the lawnmower plots (RLA, RLB, PLA, PLB) show that the personal concentrations vary as the operator conducts a regular row-cutting pattern. For one row, the operator faces the wind, and is exposed to direct engine exhaust. At the end of this row, the operator turns around with the wind now from the back, and exposures near zero. The differing concentration peak heights likely relate to changes in wind speed and direction in relation to the operator's orientation.
For this discussion, the 1 h CO National Ambient Air Quality Standard (NAAQS) of 35 p.p.m. is used as an estimate of potential health concerns related to CO exposures. Figures 1 and 2 indicate that the average and median concentrations measured for the lawnmowers and string trimmers did not exceed this NAAQS value, whereas the average and median concentrations for the chainsaws did exceed the NAAQS value during these tests. The plots also illustrate that operation of all the equipment can result in extremely high, short-term exposures. Presumably, if the combinations of engine load, activity and environmental conditions, leading to elevated exposures, persisted over long periods of time, the personal CO exposures could be much higher than that experienced during this study. The likelihood of this scenario varies among the equipment types. The operators experienced exposures above the NAAQS limit for chainsaw operation; however, the chainsaw tests took <30 min to complete owing to the capacity of the gas tanks in the equipment tested. If the operator had continued running the chainsaw for an additional 30 min, the personal exposures would likely have exceeded the CO NAAQS concentration value. If the operator discontinued working for the remaining 30 min and did not have any other significant CO exposure during that time, the 1-h NAAQS value would not be exceeded.
Over 90% of co-located CO concentration measurements at the background site were below the monitor's detection limit of 0.05 p.p.m. (results in a reading of 0 p.p.m.); thus a correlation coefficient, r2, could not be calculated for these samplers. Zero and span gas calibrations during the testing program indicated the CO samplers were performing within the manufacturer's specifications.
PM2.5 integrated measurements resulted in high exposures during the engine tests. Table 2 shows the average personal PM2.5 concentrations experienced by the operator during each test. The table also lists 8-h average background concentrations that encompassed both the morning and afternoon test periods. Uncertainty estimates listed in this table are based on field filter blank mass measurements for the applicable test day and the PEM and FRM sample flow rates, respectively. The table reveals that the operators experienced high exposures during the majority of the tests.
As the PM2.5 NAAQS value is based on a 24-h average exposure, the concentrations shown in Table 2 cannot be directly related to this health standard. Table 3 presents equivalent 24-hour average PM2.5 exposures based on the gravimetric measurements for the engine tests and the background measurements. The “personal use” exposure category assumes that the testing time periods used in the SEES study represent typical exposure lengths for the general population using the equipment for personal, residential use. During the remainder of the day, the operator is only exposed to PM2.5 concentrations at the background level. The “occupational” exposure category assumes that the equipment operator experiences the elevated exposures for an 8-h workday, and then experiences exposures at the background concentration level for the remainder of the day. This exposure scenario represents a likely worst-case occupational exposure situation owing to the long hours of equipment operation. The analysis showed that the high PM2.5 concentrations experienced by the operators, even for the short, “personal use” time period, could result in average 24-h exposures greater than the PM2.5 24-h NAAQS average concentration of 65 μg/m3.
A 10 s average PM2.5 optical concentration measurements suggested highly variable exposure levels that depended on activity and environmental factors. Figure 3 contains the complementary PM2.5 continuous measurements for comparison with the CO measurements in Figure 2. These plots indicate similar PM and CO exposure patterns for the lawnmowers, but differing patterns for the string trimmers and chainsaws. These results may be explained by the influence of the cutting activities on the PM measurements. The operator was closer to the cutting activities, especially for the chainsaws, than for the lawnmowing activities. In addition, these activities do not require the significant changes in the operator's orientation relative to prevailing winds, as do the lawnmowing activities.
The co-located PM2.5 measurements at the background site yielded correlation coefficients, r2, of 0.89 and 0.90 for the gravimetric and optical methods, respectively (n=8). Comparisons with the FRM sampler resulted in r2 values of 0.83 and 0.91 for the gravimetric and optical methods, respectively (n=8). For the background site FRM and Sidepak measurements, the ratio of gravimetric mass to average optical mass ranged from 0.25 to 0.47 (average=0.34; SD=0.05; n=8). For the background site PEM and Sidepak measurements, the ratio of gravimetric mass to average optical mass ranged from 0.16 to 0.59 (average=0.36; SD=0.09; n=16). The co-located gravimetric and optical measurements obtained during personal sampling indicated a highly varying ratio between the measurement methods. The ratio of gravimetric mass to average optical mass ranged from 0.45 (test STB) to 13.75 (test CSA) (average=4.34; SD=4.36; n=16). The variability in the gravimetric to optical mass ratio for the personal sampling was not surprising owing to a number of factors. The optical Sidepak method has a minimum size detection cutpoint of approximately 0.1 μm in aerodynamic diameter. Studies on gasoline-powered mobile source emissions indicate that a significant portion of mass emissions occur in the sub 0.1 μm size range (Cadle et al., 1999; Durbin et al., 1999; Kleeman et al., 2000); thus, the optical method would measure lower concentrations with a similar emission source containing a significant mass fraction of sub 0.1 μm particles. Another factor that may have contributed to differences in concentration values is the composition of the measured aerosol. Each engine activity may have produced varying mixtures of exhaust and mechanically generated aerosol, resulting in different optical properties for each test. Thus, each test may have a different calibration factor for the gravimetric to optical comparison for the personal exposure measurements. Based on the highly variable gravimetric to optical mass ratio for the personal samplers, the continuous data were used to estimate trends in exposures rather than quantifying the actual PM2.5 mass exposures.
The EDXRF results provided information on the relative contributions of trace elements to the PM2.5 mass measurements to identify the potential contribution of reentrained soil on the PM2.5 exposures. In this type of mass balance reconstruction, several conversions were performed to account for real-world occurrences of the analytes. As described by Malm et al. (1994), elemental sulfur was converted to atmospheric sulfate, several inorganic elements were summed and converted to represent the soil components (Al, Si, Ca, Fe, and Ti), and several trace elements were summed and converted to oxide forms, representing potential combustion-related particles (V, Mn, Ni, Cu, Zn, As, Pb, Se, Sr, P, Cr, and K). Concentration values less than two-times the corresponding uncertainty were removed from the analysis to avoid Type I false-positive errors.
Figure 4 presents a PM2.5 mass balance reconstruction based on the parameters previously described: a trace-element oxide variable (trace elements), atmospheric sulfate, soil, and the remaining mass not accounted for in the EDXRF analysis (category listed as “other”). Results indicated that reentrained soil material did not contribute a large portion of the PM mass, typically < 15% for all tests, except the commercial riding lawnmower (RLA). For the chainsaw tests, reentrained soil materials were negligible. Atmospheric sulfate concentrations suggest a contribution from regional background PM2.5 concentrations; although some sulfate emissions may result from the engine exhaust. The EDXRF results also indicated a measurable contribution of transition metal emissions from most of the test engines. The bulk of the mass occurred in the “other” category, with contributions ranging from 40% to 97% of the total mass.
The mass contained in the “other” category in Figure 4 likely contains a sizeable fraction of carbonaceous PM. Figure 5 presents results of the EC/OC analyses performed on the quartz fiber filters. As shown, the majority of carbon present on these filters was OC. These results are consistent with the findings from dynamometer emission tests conducted on similar type and technology engines (Snow et al., 2004).
Air Toxics Exposures
Measurements for benzene, toluene, ethylbenzene, and xylenes suggest that the method used for VOC sampling was not sufficient for this application. For the majority of samples collected, the mass identified on the exposed badges did not significantly exceed the mass identified on the field blank badge for the test period. These results suggest that detection limits for the badges were not sufficient for the concentrations experienced during the field study, or the badges were not properly handled during the study. Breath samples were collected from the equipment operators in 1-l summa canisters before and after each engine test, and these samples will be analyzed for VOC constituents. Reporting of this data is beyond the scope of this article; however, this data may be useful in evaluating the VOC personal sampling method used in this study.
Elevated personal aldehyde measurements relative to the background concentrations indicated the presence of primary formaldehyde and acetaldehyde emissions from the engines tested. Results included in Figures 6 and 7 for formaldehyde and acetaldehyde, respectively, identified exposure levels above background concentrations. The average concentration levels experienced by the engine operators (ranging from 30 min to slightly over 2 h) were below the 8-h Occupational Safety and Health (OSHA) permissible exposure limit (PEL) of 730 p.p.m. for formaldehyde and 200 p.p.m. for acetaldehyde, respectively. Note that these exposure limits represent potential adverse health effects for healthy workers. Individuals using this equipment for personal, residential use may be susceptible to adverse effects at concentrations lower than the OSHA standards.
A qualitative assessment of the data indicated that wind direction significantly affected concentration measurements for each engine. Wind direction factors into the concentration variability based on changes in the physical wind vector in the study area, as well as changes in the operator's orientation to the wind while conducting the applicable lawn and garden activity. A review of the video recordings for the engine tests suggested that the wind direction measured at the background site did not always represent the actual wind vectors experienced by the operator. The video revealed that the operator's orientation relative to the wind while conducting the applicable activity changed significantly during the tests, and these changes could not be quantified in the assessment. As expected, regressions evaluating CO and PM2.5 concentrations relative to environmental conditions indicated that wind direction values at the background site did not explain a significant portion of the data variability.
Results from the study indicate that emissions from some lawn and garden equipment meeting EPA's current Phase 2 standards may result in exposures to certain pollutants at levels of concern for adverse health effects. The results demonstrate that pollutant exposures significantly vary depending on operator activity and environmental factors. Under certain conditions, health standards may be exceeded. In order to fully characterize potential health effects from the operation of lawn and garden equipment, refined activity data will be needed to determine the frequency of events leading to potentially high exposures.
We acknowledge the efforts of others who contributed to this project. Carl Scarbro, Chris Lieske, and Phil Carlson of the US Environmental Protection Agency's (EPA) Office of Transportation and Air Quality (OTAQ) assisted in identifying appropriate equipment for testing, and procuring the actual equipment tested. Ron Williams and Don Whitaker of EPA's Office of Research and Development and Chad Bailey of OTAQ provided valuable assistance in designing and implementing the study.
The United States Environmental Protection Agency through its Office of Research and Development funded and managed the research described here under Contract 68-D-00-206 to ManTech Environmental Technology Incorporated. It has been subjected to Agency review and approved for publication.