William R. Pierson

THE DRI THERMAL/OPTICAL REFLECTANCE CARBON ANALYSIS SYSTEM: DESCRIPTION, EVALUATION AND APPLICATIONS IN U.S. AIR QUALITY STUDIES

Abstract The thermal/optical reflectance method of carbon analysis developed by Huntzicker et al. (in Particulate Carbon, Atmospheric Life Cycle, edited by Wolff G. T. and Klimisch R. L., pp. 79–88, Plenum Press, New York, 1982) has been adapted by several laboratories for the quantification of organic and elemental carbon on quartz-fiber filter deposits. While the principle used by these laboratories is identical to that of Huntzicker et al., the details differ with respect to calibration standards, analysis time, temperature ramping and volatilization/combustion temperatures. This paper reports a variation on this method which has been applied to over 27,000 samples taken in more than a dozen urban and regional air quality studies in the U.S.A. In this variation, a 0.5 cm2 punch from a dozen urban and regional air quality studies in 120, 250, 450 and 550°C in a pure helium atmosphere, then to combustion at temperatures of 550, 700 and 800°C in a 2% oxygen and 98% helium atmosphere. The carbon which evolves at each temperature is converted to methane and quantified with a flame ionization detector. The seven separate carbon fractions facilitate evaluation of the method and increase the information content concerning the samples. The reflectance from the deposit side of the filter punch is monitored throughout the analysis. This reflectance usually decreases during volatilization in the helium atmosphere owing to the pyrolysis of organic material. When oxygen is added, the reflectance increases as the light-absorbing carbon is combusted and removed. Organic carbon is defined as that which evolves prior to re-attainment of the original reflectance, and elemental carbon is defined as that which evolves after the original reflectance has been attained. By this definition, “organic carbon” is actually organic carbon that does not absorb light at the wavelength used (632.8 nm) and “elemental carbon” is light-absorbing organic and elemental carbon. Assumptions underlying the procedure are discussed, as well as comparisons with other methods and recommendations for further work.

Real-world automotive emissions : Summary of studies in the Fort McHenry and Tuscarora Mountain Tunnels

Al~ract--Motor vehicle emission rates of CO, NO, NOx, and gas-phase speciated nonmethane hydrocarbons (NMHC) and carbonyl compounds were measured in 1992 in the Fort McHenry Tunnel under Baltimore Harbor and in the Tuscarora Mountain Tunnel of the Pennsylvania Turnpike, for comparison with emission-model predictions and for calculation of the reactivity of vehicle emissions with respect to 03 formation. Both tunnels represent a high-speed setting at relatively steady speed. The cars at both sites tended to be newer than elsewhere (median age was < 4 yr), and much better maintained as judged by low CO/CO2 ratios and other emissions characteristics. The Tuscarora Mountain Tunnel is flat, making it advantageous for testing automotive emission models, while in the underwater Fort McHenry Tunnel the impact of roadway grade can be evaluated. MOBILE4.1 and MOBILE5 gave predictions within + 50% of observation most of the time. Tbere was a tendency to overpredict, especially with MOBILE5 and especially at Tuscarora. However, fight-dutyvehicle CO, NMHC, and NOx all were underpredicted by MOBILE4.1 at Fort McHenry. Light-dutyvehicle CO/NO~ ratios and NMHC/NO~ ratios were generally a little higher than predicted. The comparability of the predictions to the observations contrasts with a 1987 experiment in an urban tunnel (Van Nuys) where CO and HC, as well as CO/NO~ and NMHC/NO~ ratios, were grossly underpredicted. The effect of roadway grade on gram per mile (g mi- 1) emissions was substantial. Fuel-specific emissions (g gal-1), however, were almost independent of roadway grade, which suggests a potential virtue in emissions models based on fuel-specific emissions rather than g mi- 1 emissions. Some 200 NMHC and carbonyl emissions species were quantified as to their light- and heavy-dutyvehicle emission rates. The heavy-duty-vehicle NMHC emissions were calculated to possess more reactivity, per vehicle-mile, with respect to 03 formation (g 03 per vehicle-mile) than did the light-duty-vehicle NMHC emissions. Per gallon of fuel consumed, the light-duty vehicles had the greater reactivity. Much of the NMHC, and much of their reactivity with respect to O3 formation, resided in compounds heavier than Cto, mostly from beavy-duty diesels, implying that atmospheric NMHC sampling with canisters alone is inadequate in at least some situations since canisters were found not to be quantitative beyond ~ C1o. The contrasting lack of compounds heavier than C1o from light-duty vehicles suggests a way to separate light- and heavy-duty-vehicle contributions in receptor modeling source apportionment. The division between light-duty-vehicle tailpipe and nontaiipipe NMHC emissions was ~ 85% tailpipe and ~ 15% nontailpipe (evaporative running losses, etc.). Measured CO/CO2 ratios agreed well with concurrent roadside infrared remote sensing measurements on light-duty vehicles, although remote sensing HC/CO2 ratio measurements were not successful at the low HC levels prevailing. Remote sensing measurements on heavy-duty diesels were obtained for the first time, and were roughly in agreement with the regular (bag sampling) tunnel measurements in both CO/CO2 and HC/CO2 ratios. A number of recommendations for further experiments, measurement methodology development, and emissions model development and evaluation are offered. Copyright

Comparison of the SCAQS Tunnel Study with Other On Road Vehicle Emission Data

The Van Nuys Tunnel experiment conducted in 1987 by Ingalls et al. (see A&WMA Paper 89-137.3), to verify automotive emission inventories as part of the Southern California Air Quality Study (SCAQS), gave higher CO and HC emission-rate values than expected on the basis of automotive-emission models—by factors of approximately 3 and 4, respectively. The CO/NOX and HC/NOX emission-rate ratios moreover were higher than expected—by similar factors (NOX emission rates were about as expected). The purpose of the present paper is to review the literature on dynamometer and on-road (in tunnels and along roadways) testing of in-use vehicles, and on urban-air CO/HC/NOX concentration ratios, to see whether the Van Nuys Tunnel results are reasonable in terms of previous experience. The conclusions are that (1) on-road CO and HC emissions higher than expected have been reported before, (2) on-road CO and HC emissions consistent with the Van Nuys Tunnel results have been reported before, and (3) on-road CO/NOX and HC/NO...

Volatile organic compounds up to C20 emitted from motor vehicles; measurement methods

To understand better the sources of observed differences between on-road vehicle emissions and model estimates, and to evaluate the emission of ozone precursors from motor vehicles, a series of experiments was conducted in the Fort McHenry Tunnel, Baltimore, Maryland (18–24 June 1992), and in the Tuscarora Mountain Tunnel, Pennsylvania (2–8 September 1992). Samples were collected using stainless steel canisters (whole air samples, analyzed for C2ue5f8C12 hydrocarbons), Tenax-TA solid adsorbent cartridges (for semi-volatile hydrocarbons, in the C8ue5f8C20 range), and 2,4-dinitrophenylhydrazine (DNPH) impregnated cartridges (for carbonyl compounds). The samples were analyzed using high resolution gas chromatographic separation with Fourier transform infrared/mass spectrometric detection (GC/IRD/ MSD) for qualitative identification and with flame ionization detection (GC/FID) for quantitation of hydrocarbons, and high performance liquid chromatography (HPLC) for identification and quantitation of carbonyl compounds. A custom-designed database management system was used to handle the large data sets generated by these analyses. From the evaluation of canister and Tenax sample stability upon storage, it was found that hydrocarbons in the C8ue5f8C12 range seemed to be more stable in the Tenax cartridge than in the canister. The effect of the Nafion® dryer (frequently used for moisture removal prior to cryogenic concentration of the canister samples) was also assessed and it was found to lower the measured concentrations of hydrocarbons collected in the canisters. Comparison of hydrocarbon concentrations found in the Tenax and canister samples allows an assessment of the contribution of semi-volatile hydrocarbons (C10ue5f8C20 range derived from Tenax data) to the total non-methane hydrocarbons (C2ue5f8C20, derived from canisters and Tenax data). The results of this study show that hydrocarbons in the range of C10ue5f8C20 are important components of gas-phase hydrocarbons emitted from heavy-duty diesel vehicles (they account for approximately half of the total gas-phase non-methane hydrocarbon emission rates) and hence that solid adsorbent sampling should be used in addition to canister sampling in measurements of motor vehicle emissions.

Real-world emissions and calculated reactivities of organic species from motor vehicles

To obtain real-world motor vehicle emission rates for the hydrocarbon ozone precursors, a series of experiments was conducted in the Fort McHenry Tunnel, Baltimore, Maryland and in the Tuscarora Mountain Tunnel, Pennsylvania. Air samples collected in the tunnels were analyzed for approximately 200 non-methane hydrocarbon (NMHC) species up to C20, and formaldehyde. Emission rates were determined from tunnel inlet and outlet fluxes. Traffic composition analysis allowed emissions to be split into light-duty (LD; mostly spark-ignition) and heavy-duty (HD; mostly diesel) contributions. LD emissions of NMHC at Tuscarora were 293 mg/veh-mile, with paraflins constituting 35%, olefins 23%, aromatics 42%, and 6 mg/veh-mile of formaldehyde. At Fort McHenry, LD hydrocarbon emissions were 615 mg/veh-mile, with 38% paraffins, 18% olefins, and 44% aromatics, and 7 mg/veh-mile of formaldehyde. In both tunnels, HD emissions were approximately double LD emissions, but with higher percent paraffins, lower percent olefins, and an order of magnitude more formaldehyde. Through use of reactivity adjustment factors, the reactivity of the NMHC emissions with respect to ozone formation was assessed. Reactivity followed emissions, with HD emissions approximately twice the reactivity of LD emissions (on a per vehicle-mile basis). The mass specific reactivity (g-O3/g-emission) was nearly constant among all vehicles. The effect of grade (assessed at Fort McHenry) was approximately a factor of 2 for both emissions and reactivity. However, since fuel-specific emissions (g-emission/gallon fuel consumed for LD and HD vehicles were nearly independent of grade at Fort McHenry, the fuel-specific ozone reactivity (g-O3/gallon fuel consumed) was also nearly constant over the down- and up-grades.

An Assessment of the Mobile Source Contribution to PM10 and PM2.5 in the United States

Mobile sources are significant contributors to ambient particulate matter (PM) in the United States. As the emphasis shifts from PM10 to PM2.5, it becomes particularly important to account for the mobile source contribution to observed particulate levels since these sources may be the major contributor to the fine particle fraction. This is due to the fact that most mobile source mass emissions have an aerodynamic diameter less than 2.5 μn, while the particles of geological origin that tend to dominate the PM10 fraction generally have an aerodynamic diameter greater than 2.5 μm. A common approach to assess the relative contributions of sources to observed particulate mass concentrations is the application of source apportionment methods. These methods include material balance, chemical mass balance (CMB), and multivariate receptor models. This paper describes a number recent source attribution studies performed in the United States in order to evaluate the range of the mobile source contribution to observed PM. In addition, a review of the methods used to apportion source contributions to ambient particulate loadings is presented.

Exhaust Particle Size Distribution Measurements at the Tuscarora Mountain Tunnel

On-road particle size distributions were measured at the Tuscarora Mountain tunnel on the Pennsylvania Turnpike in May 1999. The data were obtained using a scanning mobility particle sizer. The nucleation modes of the size distributions contained most of the particles on a number concentration basis and exhibited peak diameters ranging from 11 to 17 nm. This observation is consistent with previous calculations and measurements, indicating that significant numbers of ultrafine aerosol particles can be expected in close proximity to busy motorways. The experiment provided 4 case studies for which the tunnel inlet data could be used to correct data obtained at the outlet, allowing for estimates of particle production within the tunnel. Exhaust particle production rates per vehicle kilometer were estimated; the results are presented with the caveat that the measurements were affected by ambient dilution. The 4 case study nucleation mode sizes varied inversely with ambient temperature. The light-duty vehicle contributions to the ultrafine particle distributions were apparently dominated by the heavy-duty vehicle contributions.

Apportionment of NMHC tailpipe vs non-tailpipe emissions in the Fort McHenry and Tuscarora mountain tunnels

Measurements of on-road emissions of non-methane hydrocarbons (NMHCs) were made in the Fort McHenry Tunnel (Baltimore) and Tuscarora Mountain Tunnel (Pennsylvania) during the summer of 1992. Measurements were made during 11 one-hour periods in the Fort McHenry Tunnel and during 11 one-hour periods in the Tuscarora Mountain Tunnel. The observed light-duty fleets were quite new, with a median model year of approximately 1989. Speciated NMHC values were obtained from analyses of canister and Tenax samples, and light-duty speciated emission factors were calculated for the two tunnels. Fuel samples were collected in the area around the tunnels for use in constructing headspace and liquid fuel profiles for the chemical mass balance (CMB) receptor model. Profiles of tailpipe emissions were obtained from the literature. The CMB was used to apportion tailpipe from non-tailpipe emissions. Non-tailpipe sources were found to constitute approximately 15% of the light-duty NMHC emissions. The Federal automotive emission-rate models, MOBILE4.1 and MOBILE5, underpredicted non-tailpipe emissions, assigning approximately 9% and 6.5%, respectively, to non-tailpipe sources. In terms of total absolute emissions, MOBILE5 predictions were approximately a factor of 2 greater than MOBILE4.1 predictions. Both MOBILE4.1 and MOBILE5 underestimated the NMHC emissions in the Fort McHenry Tunnel and overpredicted the NMHC emissions in the Tuscarora Mountain Tunnel. In all cases, the MOBILE models underestimated the absolute value of the non-tailpipe emissions. The ability of the MOBILE models to account for observed emissions when conditions are more variable than those studies in the Fort McHenry and Tuscarora Mountain tunnels is still an open question.

Recent measurements of mobile source emission factors in North American tunnels

Results of recent tunnel studies in the Cassiar, Tuscarora, Fort McHenry, and Caldecott tunnels are described, along with the methodology for calculating vehicle emissions in these studies. Results for Cassiar, Fort McHenry and Tuscarora generally agree within ±50% with the model predictions and the emissions ratios are also in good agreement with the models. These results contrast with the 1987 Van Nuys experiment, wherein the CO and HC were greatly underpredicted, as were the CO/NO x and HC/NO x ratios by factors of two and more. The 1994 Caldecott CO/NO x results deviate significantly from model predictions and are similar to those seen in the 1987 Van Nuys study, although the HC/NO x ratio is in agreement with the model prediction. Possible explanations for differences or lack of differences between the observed tunnel emissions and model predictions may include the evolution of the models and the increase in the basic emission factors incorporated in the models, differences in the tunnel fleet, manner of driving (lack of speed variability in some of the tunnels), and the impact of grade. None of these explanations, however, can adequately account for the range of deviations between on-road observations and model predictions.

Method comparisons of vehicle emissions measurements in the Fort McHenry and Tuscarora Mountain Tunnels

Abstract Experiments were conducted in the Fort McHenry Tunnel in Baltimore, MD, and in the Tuscarora Mountain Tunnel in Pennsylvania, during the summer of 1992 to evaluate real-world automotive emissions. Included in these experiments were the first reported measurements of individual vehicle exhaust in tunnels by a remote sensing device (RSD). Results are compared to integrated emission measurements carried out by analysis of concurrent collections of tunnel air into bags, canisters, and adsorbent traps and by conventional Fourier transform infrared (FTIR) spectroscopy. The vehicles using these highway tunnels proved to be lower emitting than vehicles usually measured by remote sensing in urban areas. At Fort McHenry the RSD-measured CO CO 2 ratios were, on average, high compared to either the bag or FTIR measurements (by a factor of 1.4 ± 0.2) for the four runs monitored. RSD hydrocarbon data were obtained only at the uphill location ( + 3.76% grade). RSD HC CO 2 ratios were lower on average, but statistically indistinguishable when compared with either the FTIR or the integrated uphill measurements. At Tuscarora, the RSD-measured CO CO 2 ratios were in agreement with the CO CO 2 ratios in the tunnel bag measurements and FTIR measurements (within a factor of 1.00 ± 0.16 by one method and 0.82 ± 0.32 by a second, when traffic was dominated by light-duty spark-ignition vehicles). The RSD HC CO 2 ratios were, however, higher than the light-duty vehicle estimates from the integrated (bag/canister/Tenax) tunnel measurements by a factor of 3, and higher than the FTIR Δ HC Δ CO 2 ratios by an even higher factor, mostly owing to water vapor interferences in the low average RSD measurements. For the first time RSD measurements were collected from a small sample of heavy-duty diesels; comparisons to the heavy-duty emissions contributions for CO and HC were favorable. Analysis of emissions data for vehicle variability at Fort McHenry revealed that low CO emitting vehicles tended to be consistently low but that the minority that were high emitters ( > 2.5% CO) were more likely to be high only at the uphill location. Vehicle mileage information was collected at a toll booth in the case of Fort McHenry and at a service plaza in the case of Tuscarora for comparison against the RSD emissions measurements. This comparison showed little conventional deterioration of CO or HC emissions with mileage. The trend consisted of an increased frequency of high emitters with mileage, rather than an increase in emissions from all vehicles with increasing mileage.