On-Road Particulate Matter (PM2.5 and PM10) Emissions in the

The third most prominent species was Fe (18.5 ± 9.0%), which is greater than would .... Environmental Science and Pollution Research 2017 24 (2), 210...
2 downloads 0 Views 124KB Size
Environ. Sci. Technol. 2001, 35, 1054-1063

On-Road Particulate Matter (PM2.5 and PM10) Emissions in the Sepulveda Tunnel, Los Angeles, California J. A. GILLIES,* A. W. GERTLER, J. C. SAGEBIEL, AND W. A. DIPPEL Division of Atmospheric Sciences, Desert Research Institute, 2215 Raggio Parkway, Reno, Nevada 89512

Total and speciated particulate matter (PM2.5 and PM10) emission factors from in-use vehicles were measured for a mixed light- (97.4% LD) and heavy-duty fleet (2.6% HD) in the Sepulveda Tunnel, Los Angeles, CA. Seventeen 1-h test runs were performed between July 23, 1996, and July 27, 1996. Emission factors were calculated from mass concentration measurements taken at the tunnel entrance and exit, the volume of airflow through the tunnel, and the number of vehicles passing through the 582 m long tunnel. For the mixed LD and HD fleet, PM2.5 emission factors in the Sepulveda Tunnel ranged from 0.016 ((0.007) to 0.115 ((0.019) g/vehicle‚km traveled with an average of 0.052 ((0.027) g/vehicle‚km. PM10 emission factors ranged from 0.030 ((0.009) to 0.131 ((0.024) g/vehicle‚ km with an average of 0.069 ((0.030) g/vehicle‚km. The PM2.5 emission factor was ∼74% of the PM10 factor. Speciated emission rates and chemical profiles for use in receptor modeling were also developed. PM2.5 was dominated by organic carbon (OC) (31.0 ( 19.5%) and elemental carbon (EC) (48.5 ( 20.5%) that together account for 79% ((24%) of the total emissions. Crustal elements (Fe, Mg, Al, Si, Ca, and Mn) contribute ∼7.8%, and the ions Cl-, NO3-, NH3+, SO42-, and K+ together constitute another 9.8%. In the PM10 size fraction the particulate emissions were also dominated by OC (31 ( 12%) and EC (35 ( 13%). The third most prominent species was Fe (18.5 ( 9.0%), which is greater than would be expected from purely geological sources. Other geological components (Mg, Al, Si, K, Ca, and Mn) accounted for an additional 12.6%. PM10 emission factors showed some dependence on vehicle speed, whereas PM2.5 did not. For test runs in which the average vehicle speed was 42.6 km/h a 1.7 times increase in PM10 emission factor was observed compared to those runs with an average vehicle speed of 72.6 km/h. Speciated emissions were similar. However, there is significantly greater mass attributable to geological material in the PM10, indicative of an increased contribution from resuspended road dust. The PM2.5 shows relatively good correlation with NOx emissions, which indicates that even at the low percent of HD vehicles, which emit significantly more NOx than LD vehicles, they may also have a significant impact on the PM2.5 levels.

Introduction Atmospheric particulate matter (PM) has been implicated in human health effects. Recent studies have discussed the 1054

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001

epidemiology of this relationship (1-4), potential causal mechanisms (5), and the controversy that surrounds the PM and health effects debate (6). To mitigate the impacts of PM on health, a new standard has been proposed. The proposed national ambient air quality standard (NAAQS) for PM is for two size classes, PM10 (particles of aerodynamic diameter e10 µm) and PM2.5 (particles of aerodynamic diameter e2.5 µm) (7). The PM2.5 is a mass-based standard of 65 µg/m3 over a 24-h period and a 15 µg/m3 annual average. The PM10 standard is 150 µg/m3 over a 24-h period and 50 µg/m3 for a three-year average of the annual arithmetic mean PM10. Mobile sources emit particulate matter and contribute significantly to the ambient levels. Particulate emission sources from vehicles include their exhaust (8, 9), the mechanical wear of tires and brakes (10-12), and the ejection of particles from the pavement (13) and unpaved road shoulders (14) by resuspension processes. The products of tire and brake wear and the resuspended road dust are dominated by particles >10 µm, although a tail extends below this size (8). Particulate matter in the vehicle exhaust is dominated by particles smaller than PM10 (15, 16). Several studies have shown particulate emission rates from properly functioning late-model gasoline-fueled vehicles are low, typically on the order of 0.002-0.006 g/vehicle‚km (17, 18). Higher emission rates can be associated with older poorly performing vehicles. For example, Dickson et al. (19) measured PM10 emission factors for 31, 1964-1970 model year, in-use gasoline vehicles in the Unocal SCRAP Program that ranged from 0.06 to 10.4 g/vehicle‚km. Hildemann et al. (20) tested seven, 1977-1983 catalyst in-use vehicles with an average odometer reading of 122 360 km and found a PM2.0 emission rate of 0.011 g/vehicle‚km. In the Ft. McHenry Tunnel in 1993, Gertler et al. (21) reported the average PM10 emission factor for the light- (LD) and heavy-duty (HD) fleets to be 0.009 ((0.037) and 0.42 ((0.08) g/vehicle‚km, respectively. In a program designed to recruit high HC and/or CO emitters Sagebiel et al. (9) reported an average PM10 emission rate of 0.114 g/vehicle‚km. This was determined from a 23car sample with a model year range of 1976-1990, using roadside IM240 dynamometer testing. The vehicles from which smoke was visible (“smokers”) had an average emission rate of 0.346 g/vehicle‚km, whereas the “nonsmoker” rate was 0.032 g/vehicle‚km (9). Durbin et al. (22) carried out dynamometer testing on 129 gasoline vehicles as part of a program to characterize particulate mass emission rates for representative in-use vehicles in the South Coast Air Basin. The PM emissions ranged between 0.021 g/vehicle‚km for LD vehicles older than 1980 to 0.002 g/vehicle‚km for LD vehicles newer than 1991. The emission factors presented in all but one of the above-mentioned studies were developed from dynamometer data and hence represent small sample sizes drawn from a large population of vehicles. In addition, some of these studies used screening techniques to ensure they characterized only certain portions of the on-road fleet. Gillies and Gertler (23) raise concerns that the inherent variability in chemical composition among dynamometerderived profiles makes their use in chemical mass balance (CMB) receptor modeling questionable. Mixed-fleet profiles from thousands of in-use vehicles such as those developed in this study potentially offer a better characterization of mobile source emissions than profiles made by compositing highly variable individual vehicle profiles. Diesel vehicles emit significant amounts of particulate matter (24) that is predominantly less than PM2.5 (15, 25). Recent studies by Clark et al. (23), Wang et al. (26), and Graboski et al. (27) indicate diesel PM emission factors 10.1021/es991320p CCC: $20.00

 2001 American Chemical Society Published on Web 02/14/2001

ranging between 0.14 and 1.77 g/vehicle‚km. These values appear to be heavily influenced by the driving cycles chosen for the emissions testing. Emission factors from older studies such as Gertler et al.’s tunnel study (21) and Lowenthal et al.’s dynamometer study (28) report values of 0.42 and 0.6 g/vehicle‚km, respectively, for PM2.5. Mass and chemically speciated particulate emission rates from on-road, in-use vehicles remain relatively limited, and newer data are required for inventory modeling and source apportionment studies. Data on emission rates are also required periodically to develop a record of changes in fleet emissions with time. The purpose of this paper is to present PM2.5 and PM10 total and speciated emission factors (grams per vehicle per kilometer) for a mixed LD and HD fleet and compare these with other recent tunnel and dynamometerderived emission factor data. In addition, these data can be used to define trends in mobile source PM emissions in the South Coast Air Basin and elsewhere. The on-road emission factors were estimated from ambient PM samples collected in the Sepulveda Tunnel located near Los Angeles International Airport in Los Angeles, CA, in July 1996.

Site The Sepulveda Boulevard Tunnel is 582 m long, straight, and approximately flat in the covered portions, although there is a downgrade approaching the tunnel and an upgrade leaving it. There are two bores, three lanes each with a sidewalk on the right side of each bore. There is no breakdown lane, although there are three pullout areas in each direction. Each pullout is large enough to accommodate two vehicles. All of the openings between the two bores were sealed to prevent air transfer between the bores. The ventilation system within the tunnel was not in operation during the sampling periods. The test runs were conducted in the west bore of the tunnel that carries Sepulveda Boulevard southbound from the Los Angeles International Airport. Immediately after the tunnel there is a turn lane to allow access to the on ramps to Highway 105, which connects to Highway 405.

Methodology Instrumentation and Chemical Analyses. Particulate matter samples were collected using two PM10 and two PM2.5 medium-volume samplers designed to collect samples for chemical analyses (29). This type of sampler employs a SierraAndersen 254 PM10 inlet or Bendix PM2.5 cyclone to determine the size fractions collected. The ambient air is transmitted through the size-selective inlet and into a plenum. The flow rate is controlled by maintaining a constant pressure across a valve with a differential pressure regulator. For the sizeselective inlet to work properly, a flow rate of 113 liters per minute (lpm) must be maintained through the sampler. Two Savillex filter packs, one with a ringed 47-mm Teflon membrane filter (Gelman Scientific, Ann Arbor, MI) and one with a 47-mm quartz fiber filter (Pallflex Corp., Putnam, CT) draw air from the plenum with flow rates of 20 lpm to collect samples for gravimetric and chemical analyses. The remaining 73 lpm was drawn through a makeup airport. The flow rates were set with a calibrated rotometer and monitored with the same rotometer at each sample change. PM10 and PM2.5 samplers were positioned inside the tunnel, on the sidewalks, at both the inlet and the outlet, and programmed to sample for the same time interval. The volume of airflow through the tunnel, which is required for the emission factor calculation, was determined by a tracer method (30). A known amount of inert SF6 gas was released at the tunnel entrance, and measurements were made for the duration of each sampling period. The dilution ratio of SF6 at the tunnel outlet gives the tunnel volumetric

flow. The concentration of SF6 at the tunnel exit was determined with a Lagus model 215AUP portable trace gas monitor. Videotaped records of the vehicle traffic moving through the monitored tunnel bore were used to determine the number of vehicles, the mixture of LD and HD vehicles, and age characteristics of the LD fleet. A radar gun was used to determine the speed characteristics of the vehicles during a sampling interval. The Teflon membrane filters were weighed on a Cahn 31 electro-microbalance before and after sampling to determine mass concentrations. Chemical analyses were performed on both the Teflon membrane and quartz fiber filters following the methodology described by Watson and Chow (31). Briefly, the Teflon membrane filters were analyzed for elements by X-ray fluorescence. Half of the quartz filter was extracted with distilled-deionized water and the extract analyzed for chloride, nitrate, and sulfate ions by ion chromatography, for ammonium by automated colorimetry, and for sodium and potassium by atomic absorption spectrometry. Organic and elemental carbon were measured by thermal-optical reflectance on 0.5 cm2 punches taken from the remaining half of the quartz fiber filter (32). The individual chemical concentrations of PM10 and PM2.5 determined from the analytical techniques carried out on the Teflon membrane and quartz fiber filters were used to calculate the sum of species concentration.

Emission Factors The emission factors for PM10 and PM2.5 were calculated following the methodology of Pierson et al. (10, 30, 33). Briefly, samples are taken simultaneously of the air coming into and exiting the tunnel portals to determine the concentrations of the species of interest in the sampled air. The mass of PM produced by vehicles traveling through the tunnel can be determined from

∑(C

E)[

i

outVout)i

-

∑(C

inVin)j]/NL

(1)

i

where E ) emission factor in g/vehicle‚km traveled, Cout ) PM concentration in the outlet air (g/m3), Vout ) volume of air for each of the i exit channels (m3), Cin ) PM concentration in the inlet air (g/m3), Vin ) volume of air for each of the i inlet channels (m3), N ) traffic count, and L ) length of the tunnel (km).

Results Run Descriptions. The 17 runs were carried out over the period July 23-27, 1996. The exact times, environmental conditions, and vehicle characteristics observed for each run are given in Table 1. The average ambient temperature during the sampling period was 22 °C. The average speed for the 17 runs was 60.5 ((11.7) km/h and ranged between 30.2 ((14.0) and 80.9 ((9.1) km/h. For all runs the vehicles came in pulses due to a traffic light ∼0.4 km from the tunnel entrance. Overall, there were 11 runs with an average speed >64.4 km/h (average speed ) 72.6 km/h) and six runs with an average speed