Measurement and Modelling of Motor Vehicle Related Air Toxics Along Urban Streets Health Canada Toxic Substances Research Initiative Project #55, Final Report May 2001 (Revised November 2002) Principal Investigator Deniz Karman Department of Civil and Environmental Engineering Carleton University Collaborating Partner Lisa Graham Emissions Research and Measurement Division Environment Canada EXECUTIVE SUMMARY This is the revised final report for Project # 55 of the Toxic Substances Research Initiative (TSRI) of Health Canada. The project is a collaborative effort between Carleton University and Environment Canada’s Emissions Research and Measurement Division (ERMD). The original final report was submitted in May 2001. The revised version has been prepared in response to comments received from the TSRI secretariat in March 2002. The objectives of the project, in summary, are: i) ii) iii) iv) To establish a database of motor vehicle related toxic substance concentrations and PM2.5 mass concentrations at nose-level along a busy downtown street by measurements in the two extremes of weather (Summer and Winter) in a typical Canadian city. To compare and correlate the short term (2 hour periods of peak traffic volume) concentrations of toxic substances and fine particulate matter measured at nose-level with the regional air quality monitoring data of longer duration (24 hours) measured at other urban sites. To compare and correlate the short-term concentrations of toxic substances measured at nose-level with the in-vehicle concentrations on typical commuting trips. To determine the contribution of motor vehicle traffic to the measured toxic substance concentrations and fine particulate matter by comparisons with motor vehicle emission data. An ambient sampling program was carried out on Slater Street in Ottawa, and in vehicles on long commuting trips, using equipment designed and built ERMD. Carleton University graduate students and research assistants were employed in sampling and data processing while most of the sample analyses were carried out at ERMD's laboratories. Two periods of fieldwork were completed: 17 January – 25 February 2000, and, 17 July – 11 August 2000. Volatile organic compound (VOC), semi-volatile organic compound (SVOC), carbonyl compound, and particulate matter (PM2.5) samples were collected. A source emission study was also undertaken at ERMD laboratories using two transit buses and two light duty vehicles to characterize emissions from these vehicles under simulated driving conditions. The main findings for the gas-phase measurements among the micro-environments studied, can be summarized as: - The temporal variation from day to day of pollutant concentrations observed in micro-environments is much higher than the spatial variation observed among diverse micro-environments such as the roadside on a busy downtown street, the rooftop on a parking structure, the inside of a commuter car, and the inside of a Measurement and Modelling of Motor Vehicle Related Air Toxics 2 - - - transit bus. The in-vehicle concentrations are generally the highest among these micro-environments Despite the large temporal variations, the median values of 24 hr average concentrations recorded at the National Air Pollution Surveillance (NAPS) ambient monitoring station on Slater street at a height of 4 m are in general agreement with the median values of 2 hour average concentrations recorded at nose-level along the same street. There are noticeable differences between 2-hour average concentrations recorded at nose level and 24-hour average concentrations recorded at a height of 10 m, on the roof of an adjacent parking structure. Nearly all of the difference is due to the sample averaging time since the comparison of 24 hr average concentrations at these two locations shows reasonably close agreement. The in-vehicle concentrations observed during this study in Ottawa are of the same magnitude as those reported by studies in other cities but in the lower half of the range of values reported. PM2.5 measurements were attempted only at the roadside station with the cyclone-filter combination used in this study. Therefore, particulate phase measurements do not offer the same opportunity for comparison as the gas-phase measurements. However, the chemical speciation data for particulate matter can be used in comparisons with NAPS stations with PM2.5 sampling capability. The speciation is also being used in source apportionment studies for fine particulate matter in urban air. The database of 2-hour VOC and PM2.5 concentrations in microenvironments is an important asset to complement the database from ambient monitoring stations that generally comprise daily concentrations for the same compounds. While ambient monitoring data capture the long term trends and provide a basis for quantifying human exposure, the short-term, microenvironment data collected during this study can give an indication of the range of exposure values and hence the uncertainties associated with health risk assessment. The database generated during this study, as well as comparable data from a similar study at this location in 1994 are being made available electronically through a web site: http://www.carleton.ca/~dkarman/OMDB. The measured concentrations of hydrocarbons on Slater Street in 2000 (in the range ~70200 micrograms per cubic meter) are explained dominantly by light duty gasoline vehicle exhaust in the cold start and hot start modes of operation. These source contributions are determined by using locally measured light duty gasoline vehicle exhaust profiles. Locally derived evaporative emission profiles and the extension of ambient data to heavier hydrocarbons may enable finer resolution between exhaust vs. evaporative emissions for gasoline vehicles, and gasoline vs. diesel exhaust emissions. Measurement and Modelling of Motor Vehicle Related Air Toxics 3 CONTENTS 1. INTRODUCTION, 5 2. METHODOLOGY, 6 3. SUMMARY OF FINDINGS, 7 3.1 Gas phase samples at the roadside and in-vehicle microenvironments, 7 3.2 Particulate matter samples at the roadside microenvironment, 18 3.3 Vehicle Emissions sampling, 24 3.4 Chemical Mass Balance Receptor Modelling, 28 4. DISSEMINATION OF RESULTS, 36 OMDB, The Ottawa Microenvironment Database 5. REFERENCES, 37 FIGURES 1. Distribution of gas phase concentrations at the roadside, winter, 8 2. Distribution of gas phase concentrations at the roadside, summer, 9 3. Comparison of 24 hour and 2 hour VOC concentrations, 10 4. Distribution of summer VOC concentrations in 1994 and 2000, 11 5. Comparison of 1994 and 2000 summer VOC median concentrations, 12 6. Comparison of 24 hr roadside and rooftop VOC concentrations, 13 7. Comparison of 2 hr roadside and 24 hr rooftop VOC concentrations, 15 8. Distribution of gas phase concentrations in commuter vehicles, winter, 16 9. Distribution of gas phase concentrations in commuter vehicles, summer, 17 10. Comparison of benzene concentrations with other studies, 21 11. Distribution of PM2.5 concentrations, winter and summer, 22 12. Comparison of PM measurements from 1994 and 2000, 23 13. Elemental and organic carbon in PM2.5 samples, 23 14. Metals in PM2.5 samples, 24 15. Mass emission rates of C2-C26 compounds from buses, 26 16. Normalized C2-C12 emission profiles of light duty vehicles, 28 17. Source contribution estimates during sampling sessions, 32 18. Uncertainties in source contribution estimates, 35 19. Goodness of fit for individual compounds, 35 TABLES 1. Comparison of select compounds in different micro-environments, 19 2. In-vehicle concentrations of select compounds from other studies, 20 3. Mass emission rates of selected compounds from buses, 27 4. Source HC profiles used in CMB modelling, 30 5. Summary output from CMB8 Modelling, 31 Measurement and Modelling of Motor Vehicle Related Air Toxics 4 1. INTRODUCTION An accurate assessment of the effects of motor vehicle related toxic emissions on urban populations is a relatively complex task, involving the quantitative characterization of the traffic, the emissions from that traffic, the exposure of individuals, and the health risks associated with that exposure. Toxic compound concentrations can show high variability over time and location, particularly in specific microenvironments such as sidewalks along busy streets, within vehicles, parking garages, near major point sources etc. The different health effects of various emissions is particularly important in view of the tradeoffs that may be involved among different types of emissions associated with alternative or reformulated fuels or transportation technologies. For particulate matter, respirable particles (those less than 2.5 m diameter), and the changes in chemical composition of the different size ranges are becoming the focus rather than the mass concentration of particles. The Emissions Research and Measurement Division (ERMD) of Environment Canada and the Department of Civil and Environmental Engineering at Carleton University have collaborated in field studies aimed at detailed characterization and quantification of motor vehicle related air pollutants in urban micro-environments, starting with a study partially funded by Health Canada in 1994. The present study, carried out in the winter and summer of 2000 complemented and expanded the 1994 study, and was funded by the Toxic Substances Research Initiative (TSRI) as Project #55. The objectives of these studies are: i) To establish a database of motor vehicle related toxic substance concentrations and PM2.5 mass concentrations at nose-level along a busy downtown street by measurements in the two extremes of weather (summer and winter) in a typical Canadian city. Such a database will be an important element for analyzing the total exposure of urban populations to toxic substances and fine particulate matter. ii) To compare and correlate the short term (2 hour periods of peak traffic volume) concentrations of toxic substances and fine particulate matter measured at noselevel with the regional air quality monitoring data of longer duration (24 hours) measured at other urban sites. Direct comparisons of nose-level concentrations with data available from the National Air Pollution Surveillance (NAPS) network will enhance the utility of NAPS data for exposure analysis. iii) To compare and correlate the short-term concentrations of toxic substances measured at nose-level with the in-vehicle concentrations on typical commuting trips. Such measurements and comparisons will provide useful data for estimating the components of total exposure for individuals who commute to the downtown area for work-related trips. iv) To determine the contribution of motor vehicle traffic to the measured toxic substance concentrations and fine particulate matter by comparisons with motor vehicle emission data. Motor vehicle emission regulations have accomplished very significant reductions in CO, NOx, and HC emissions but the effect of these reductions on the individual species of interest from a human health perspective is Measurement and Modelling of Motor Vehicle Related Air Toxics 5 relatively complex. Source apportionment for the substances of most interest would enable the assessment of effectiveness for emission control measures. Ottawa is an appropriate choice of location for the measurement of motor vehicle related toxic and reactive compounds, as it has virtually no heavy industry that could be contributing to the ambient loadings. The extremes of Canadian climate also provide some special challenges in the assessment of traffic related emission impacts under different conditions. 2. METHODOLOGY An ambient sampling program was carried out on Slater Street in Ottawa, and in vehicles on long commuting trips, using equipment designed and built by Emissions Research and Measurement Division (ERMD), Environment Canada. Carleton University graduate students and research assistants were employed in sampling and data processing while most of the sample analyses were carried out at ERMD's laboratories. A source emission study was also undertaken at ERMD laboratories using samples of heavy-duty buses and light duty vehicles to characterize emissions from these vehicles under simulated driving conditions. The ambient sampling stations were located beside a major public transit bus stop on a one-way artery. The primary roadside sampling station was set-up at pedestrian noselevel, approximately 1.5 m above the sidewalk, 60 cm from the curbside, on the south sidewalk of Slater Street. A second sampling station was set-up on the rooftop of a threestorey parking structure adjacent to the nose-level sampler. A permanent National Air Pollution Surveillance (NAPS) network station sampler was located at 88 Slater Street, within two blocks of the nose-level sampling station. The rooftop sampler was installed to enable correlation with the data acquired at the NAPS sampling station and nose-level station. The NAPS and rooftop stations provided observations at different heights (4 m and 10 m) and sampling periods (24 h) relative to the nose-level. During periods of high traffic and pedestrian volumes, two-hour, nose-level ambient samples were obtained for VOCs, SVOCs, carbonyl compounds, and fine particulate matter (PM2.5). In-vehicle VOC and carbonyl samples were collected for typical long commuting trips in the region during the same 2-hour periods. On selected days, 24-hour samples were simultaneously obtained with the 2-hour samples during the day to enable cross correlation with the NAPS ambient monitoring station at 88 Slater Street. Some cumulative, 6-hour samples were also collected to obtain sufficient mass on PM2.5 filters. For the source emission characterization study, two light duty vehicles and two buses were operated on chassis dynamometers at ERMD laboratories, using commercial fuel and standard driving cycles and test procedures at two different temperatures (-10 C and 24 C), with 2-3 repeats for each set of conditions. Mass emission rates of regulated and non-regulated gaseous emissions and PM emissions were determined. The particulate matter emissions measurements were conducted as part of the PERD funded project “Determination of the concentration, composition and sources of atmospheric Measurement and Modelling of Motor Vehicle Related Air Toxics 6 carbonaceous particles in Canada”. The detailed gaseous emissions measurements were conducted as part of the TSRI project. 3. SUMMARY OF FINDINGS Over 100 compounds were quantified in the gas phase samples obtained at the roadside and in-vehicle micro-environments and 40 elemental species quantified in the PM2.5 samples obtained at the roadside micro-environment. These compounds were also quantified in the source emission samples from vehicles operated at the ERMD laboratories. The discussion here will focus on the most abundant subset of these species. The full data set is available on a dedicated web page as detailed in Section 4 below. 3.1 Gas phase samples at the roadside and in-vehicle microenvironments Gas phase concentrations show significant variation from day to day and between sessions as demonstrated in Figures 1 and 2. The closed boxes in these figures show the 25th and 75th percentile of the data observed during the sampling session of 20 days, with the horizontal line in the box showing the median value. The caps on the vertical bars show the 10th and 90th percentile of the data while free standing horizontal bars represent outliers. In the winter, the afternoon sampling session shows noticeably higher concentrations. This can be attributed to the large number of vehicle which enter the traffic in the coldstart mode after having been parked in downtown locations during the day, whereas for the morning and noon sessions it may be expected that most vehicles arrive at the downtown location in a hot stabilized mode of operation. In the summer, the concentrations are generally lower compared to winter and the relative rise in the afternoon is not particularly noticeable. It might be expected that the effect of the cold-start mode is less pronounced under summer conditions. The 24 hr samples in Figures 1 and 2 were obtained at roadside, nose-level stations and serve to distinguish the effects of sampling time. Ambient VOC concentrations are also monitored regularly at the NAPS station on Slater Street, two blocks away from the roadside station, on the basis of one 24 hr sample every six days, collected at a height of 4 m from the street. Comparison with this ambient monitoring station data is informative because it gives an indication of the representativeness of routine air quality monitoring data for the actual exposure of pedestrians during the commuting rush periods in this microenvironment. Figure 3 compares the median concentrations of the most abundant species averaged over 24 hours at the NAPS station and the roadside samples analyzed by ERMD. Measurements at the two stations show remarkable agreement, with minor scatter. This is somewhat surprising since the analysis of 1994 data at these same locations had shown marked differences both quantitatively and qualitatively: 2 hour samples at the roadside stations had shown higher total non-methane hydrocarbons by a factor of 2-5 and noticeably different hydrocarbon speciation fingerprints explained by different source contributions. Measurement and Modelling of Motor Vehicle Related Air Toxics 7 Figure 1 (concentrations in ng/L) Measurement and Modelling of Motor Vehicle Related Air Toxics 8 hy l et ene ac ha e n p r ty l e e op n y e p r le n is o o p a e b n i s o u ta e bu ne t x 2 n -b u e n e m ta -b n u e x 2 n -p e t a n m n e x 3 -p e t a n m n e -p t a e ne b e n ta n nz e m to e n e & F o p lu e rm -x n A c a ld y le e et eh ne al y de de A c hyd et e on e et hy l e en a c th a e e n p r ty l e e op n y e p r le n is o o p a e b n i s o u ta e b ne n ut x 2 -b u e n e m ta - n n bu e x 2 -p e t a n m n e x 3 -p e t a n m nt e -p a n e e b e n ta n nz e m to e n F o & p lu e e rm -x n A c a ld y le e et eh ne al y de de A c hyd et e on e et (SLATER STREET - WINTER - 24hr) Figure 2 (concentrations in ng/L) Box Plots Top 15 NMHC Species & 3 Most Abundant Carbonyls (SLATER STREET - SUMMER ) 40 30 20 10 0 40 Measurement and Modelling of Motor Vehicle Related Air Toxics 9 7:30 11:30 3:30 24Hr 30 20 10 0 Figure 3 NAPS winter median conc. (ng/L) NAPS summer median conc. (ng/L) Comparison of NAPS 24-Hr (4m) VOC Measurements with ERMD 24-Hr Nose-level (1.5 m) VOC Mesurements 6 4 2 0 0 2 4 6 ERMD winter median conc. (ng/L) 6 4 2 0 0 2 4 6 ERMD summer median conc. (ng/L) Measurement and Modelling of Motor Vehicle Related Air Toxics 10 (n g /L ) 80 80 0 e t h y le a c n e e t yl p r e n o p e yl e n p r e o p is o b a n e u t is o b a n u t e yl e n n e b u 2 m ta n e -b u t n a n p e e 2 m n t a n 2 e b 2 m u t e n -p e e 3 m n t a n -p e e n m ta n e -c n -h yc e lo x a p e n e n t a n b e cy e n z cl o h e n e x e e n to e lu e e b e n e n z m & e n p e xy le o n e xy le n e C o n c e n tr a tio n s Figure 4 B o x P lo ts o f T o p 2 0 C o m p o u n d s fo r 1 9 9 4 C u r b s id e a n d 2 0 0 0 S la te r 1 5 :3 0 m e a s u r e m e n ts 2 0 0 0 S la t e r 1 5 : 3 0 60 40 20 0 1 9 9 4 C u r b s id e 1 5 : 3 0 60 40 20 Measurement and Modelling of Motor Vehicle Related Air Toxics 11 The distribution, shown in Figure 4, of concentrations observed for a group of compounds in the afternoon sampling sessions during the Summer 1994 and Summer 2000 studies demonstrates that the Summer 2000 concentrations are indeed noticeably lower than those in 1994. However, one should be careful not to interpret this as a general decline as the number of observation days in 2000 is only four, while it was over 15 in 1994. The fewer number of days in 2000 may therefore be contributing to the narrower spread. Comparison, in Figure 5, of the median concentrations observed for all three sampling sessions in 1994 and 2000 shows more difference for the morning sessions than the afternoon sessions shown in Figure 4. Figure 5. 2000 Slater Street Median Conc. (ng/L) Comparison of 1994 and 2000 Summer Median Concentrations 40 7:30 11:30 15:30 30 20 10 0 0 10 20 30 40 1994 Curbside Median Conc. (ng/L) The "rooftop" station used during the 2000 study also provides some opportunity for comparing the temporal and spatial variation of concentrations observed on Slater Street. This "station was initially located at the top level of the parking structure adjacent to the roadside station, about 10 m above the road, to obtain SVOC samples for comparison with roadside data as the NAPS station did not have SVOC data. During the course of sampling, VOC canisters were also deployed at this station. 24 hr average samples were obtained, enabling comparison with both the NAPS station and the roadside station. Figure 6 compares 24 hr average concentrations for the most abundant species between the rooftop and the nose level stations for two days, one in the winter, and one in the summer. There is some scatter in the winter comparison, while the summer comparison shows somewhat higher concentrations at the nose level station. When 2 hr average concentrations at nose level are compared with 24 hr average rooftop concentrations measured during the same day (Figure 7), differences do arise, most notably in the Measurement and Modelling of Motor Vehicle Related Air Toxics 12 afternoon session. There are a number of days for which this comparison can be made, but they do not all show the trend in Figure 7. While differences are more likely to arise for the afternoon or noon samples there are a few days when 2-hour average concentrations correlate well with the 24 hr average concentrations. Although dispersion modelling in this complex environment is not among the objectives of the study it is to be expected that atmospheric mixing conditions play the dominant role in the observed differences or agreements between these measurements. Figure 6 Comparison of ERMD 24-Hr Nose-level (1.5 m) VOC Measurements with ERMD 24-Hr Rooftop (10 m) VOC Mesurements Summer (29 Jul) 8 ERMD Nose-level conc. (ng/L) ERMD Nose-level conc. (ng/L) Winter (6 Feb) 8 6 4 2 0 0 8 6 4 2 ERMD Rooftop conc. (ng/L) 6 4 2 0 6 4 2 0 ERMD Rooftop conc. (ng/L) Measurement and Modelling of Motor Vehicle Related Air Toxics 13 8 Figure 7 8 8 6 6 roof.01Feb.24hr roof.01Feb.24hr 2-hr Nose-level Measurements vs 24-hr Rooftop Measurements for same Days (Concentrations in ng/L). 4 2 0 4 2 0 0 2 4 6 8 0 2 nose.01Feb7.30 4 6 .nose.01Feb.11.30 20 roof.01Feb.24hr 15 10 5 0 0 5 10 .nose.01Feb.3.30 Measurement and Modelling of Motor Vehicle Related Air Toxics 15 15 20 8 Measurement and Modelling of Motor Vehicle Related Air Toxics 16 ne ac et ne y le pr n op e y le pr ne op is o a n e bu t is o a n e bu x2 t m ene -b ut an ne p x2 ent a m n -p en e ta ne to lu m en x1 & 2 4 p -x e -t m y l e n -b en e ze ne ha hy et et le le ne h ac an et e p r y le n op e yl pr ene op is o a n bu e is o ta n x2 but e m en -b e n - u ta n x2 pen e m -p t a n e en ta ne t x 1 m & o lu e 24 p n -t m -x y e l e -b en ne ze n n - in e un da de n ca ne hy et et Figure 8. (concentrations in ng/L) Box Plots Top 15 NMHC Species & 3 Most Abundant Carbonyls (IN-CAR & IN-BUS - WINTER) ) 60 40 20 0 In-Car (PM) In-Bus (PM) In-Car (AM) In-Bus (AM) 60 40 20 0 hy le et ne ac ha e n p r ty le e op ne n y le x 2 -b u n e m ta -b n n - u ta e x2 pe ne m n x 3 -p e ta n e m nt -p a n en e t x3 ben ane m ze -h n ex e x 1 m to a n e 2 4 & p lu e -tm -x n e y x 3 -b e n l e n e F o ip - z e n rm to l e A c a ld u e n et eh e al yd de e h A c yd et e on e et hy a c le e ne p r ty le op ne y p r le n e op n a x 2 -b u n e m ta n -b e n ut x 2 -p e a n e m n ta x3 pen ne m ta -h n e ex a m to lu n e &p e x 1 x 3 -x y n e 2 4 e - le -tm to n e -b l u e n en e z x 3 n -d e e n e F o -i p -t c a n rm o l u e A c a ld e n e et eh al yd de e h A c yde et on e et Figure 9 (concentrations in ng/L) Box Plots Top 15 NMHC Species & 3 Most Abundant Carbonyls (IN-CAR & IN-BUS - SUMMER) 30 20 10 0 In-Car (PM) In-Bus (PM) In-Car (AM) In-Bus (AM) 30 20 10 0 Measurement and Modelling of Motor Vehicle Related Air Toxics 17 Concentrations measured in commuter vehicles also show significant variation from day to day (Figures 8 & 9) but no discernible pattern between sampling sessions comparable to that observed for the winter roadside station samples. It should be remembered that the commuter vehicles travel over a fairly long distance that encompasses a variety of traffic patterns, in contrast to the pattern of "arrival in the morning, departure in the afternoon" experienced on a downtown location like Slater Street. They are therefore less likely to be influenced by the cold-start effect mentioned above for the roadside station in the winter. For the middle 50% of the in-vehicle data, the variation between winter and summer is less pronounced relative to the variation observed at the roadside, but the in-vehicle winter concentrations are again generally higher. There are very large maxima observed during the winter session for the in-car data, which can be caused by the car 's relative position in traffic, e.g. following a high emitting vehicle for a considerable period of time. The ambient inside a small car is quite responsive to changes in the ambient ahead of the car due to the positioning of the fresh air intake on the car. In contrast, the ambient inside a bus is relatively isolated from such changes and the winter data for the bus show much less variation than the car. The summer session shows more variation when bus windows may be expected to be at least partially open. When in-vehicle concentrations are compared to those at the roadside, the median in-vehicle concentrations are noticeably higher than the roadside concentrations. Table 1 presents a more detailed comparison of in-vehicle and roadside concentrations for a select group of compounds important from a human health perspective. For these compounds the in-vehicle concentrations are quite comparable to roadside concentrations during winter. During the summer they are higher for the in-vehicle concentrations by a factor of around two. This spatial variation among micro-environments, while noticeable, is much less than the temporal variation observed (in Figures 1, 2, 8, and 9) for any one of these micro-environments over the period of the sampling study. The in-vehicle microenvironment, which generally shows the highest concentrations for gas-phase compounds, has been studied by others with whom comparisons are possible. Table 2 presents in-vehicle concentrations of selected compounds observed in seven other studies and Figure 10 presents a graphical comparison for Benzene. One of the studies (Jo and Park 1999) with very high concentrations and variability has been left out to avoid distorting the scale too much. The in-vehicle benzene concentrations observed during this study in Ottawa are of the same magnitude as those reported by studies in other cities but in the lower half of the range of values reported. To summarize the main findings for the gas-phase measurements that offer comparisons among all the microenvironments studied, the following can be said. - The temporal variation from day to day of pollutant concentrations observed in micro-environments is much higher than the spatial variation observed among diverse micro-environments such as the roadside on a busy downtown street, the rooftop on a parking structure, the inside of a commuter car, and the inside of a transit bus. Despite the large temporal variations, the median values of 24 hr Measurement and Modelling of Motor Vehicle Related Air Toxics 18 average concentrations recorded at the ambient monitoring station on Slater street at a height of 4 m are in general agreement with the median values of 2 hour average concentrations recorded at nose-level along the same street. - There are noticeable differences between 2-hour average concentrations recorded at nose level and 24-hour average concentrations recorded at a height of 10 m, on the roof of an adjacent parking structure. Nearly all of the difference is due to the sample averaging time since the comparison of 24 hr average concentrations at these two locations shows reasonably close agreement. Table 1. Comparison of select compound microenvironments for different seasons concentrations in different Winter mean concentrations, (micrograms/m3) Compound In-car In-bus Roadside Car/Road Bus/Road Benzene 5.06 ± 2.69 3.38 ± 1.38 4.24 ± 3.17 1.2 0.8 Toluene 18.05 ± 14.48 9.62 ± 5.15 14.41 ± 13.74 1.3 0.7 Ethylbenzene 3.09 ± 3.24 2.58 ± 1.23 2.49 ± 2.62 1.2 1.0 m,p-xylene 9.25 ± 9.97 6.93 ± 3.29 7.33 ± 7.41 1.3 0.9 o-xylene 3.93 ± 4.15 3.04 ± 1.29 2.87 ± 2.88 1.4 1.1 1,3-butadiene BDL BDL 0.38 ± 0.64 NA NA Formaldehyde 7.00 ± 2.60 3.50 ± 3.36 3.89 ± 1.59 1.8 0.9 Acetaldehyde 2.00 ± 2.04 2.50 ± 2.46 2.80 ± 1.17 0.7 0.9 Summer mean concentrations (micrograms/m3) Compound In-car In-bus Roadside Car/Road Bus/Road Benzene 5.33 ± 1.50 3.46 ± 1.00 2.16 ± 0.86 2.5 1.6 Toluene 15.66 ± 7.45 11.53 ± 3.58 7.75 ± 3.04 2.0 1.5 Ethylbenzene 2.90 ± 1.30 3.57 ± 1.77 1.36 ± 0.78 2.1 2.6 m,p-xylene 6.00 ± 2.32 7.06 ± 2.77 3.17 ± 1.84 1.9 2.2 o-xylene 2.23 ± 0.81 2.85 ± 1.01 1.20 ± 0.73 1.9 2.4 1,3-butadiene 1.13 ± 0.81 1.08 ± 0.58 0.81 ± 0.48 1.4 1.3 Formaldehyde 10.40 ± 2.39 10.50 ± 2.97 8.19 ± 11.55 1.3 1.3 Acetaldehyde 6.01 ± 1.97 5.93 ± 2.05 2.39 ± 0.61 2.5 2.5 3.2 Particulate matter samples at the roadside microenvironment Figure 11 presents the minimum, 25th percentile, 75th percentile, and maximum PM2.5 mass concentration for wintertime and summer-time sampling sessions. Median mass concentrations for the morning and evening sessions are similar and somewhat higher than the midday sessions. The summer median mass concentrations for morning and evening sessions were approximately 50% higher than in winter. Weekend 6-hr samples were collected by accumulating the sample over the three 2-hr sampling periods used during weekdays in an effort to obtain sufficient mass on the filter for metals analysis. The median concentrations for these samples are lower than the 2-hour average samples. Measurement and Modelling of Motor Vehicle Related Air Toxics 19 Table 2. Mean in-vehicle concentrations (g/m3) of VOCs and carbonyl compounds for comparable studies. m&pxylene 30.5 17.2 71.8 75.7 40.60 Ethyl benzene 8.8 4.9 24.8 18.9 12.5 11.3 14.10 11.4 6.2 1,3butadiene 3.3 2.4 Form aldehyde NA Acet aldehyde NA 53 41.9 NA NA NA NA NA NA 34.60 25.6 13.1 17.00 BDL 6 3.9 5.80 4.70 23.2 12.8 3.50 3.0 9.30 6.4 3.90 3.6 BDL 24.3 10.1 91.5 75.8 7.7 31.0 NA 2.7 2.5 NA NA NA 5.1 8.8 NA 4.8 1.7 NA NA NA 3.6 NA NA 5.5 2.1 NA NA NA 15.6 1.6 NA NA NA NA 5.2 0.5 NA NA 20.3 3.5 NA NA NA NA 6.3 0.8 NA NA Reference Study Benzene toluene Chan et al., 1991a Raleigh, NC, 1988 South Korea, in-car, winter 1996-1997 South Korea, in-bus, winter 1996-1997 Toronto, winter 1994 Toronto, summer 1994 Birmingham, UK, 1996 Boston, winter 1989 New Jersey, winter 1991 Sydney, Australia, 1996 (morning commute) Sydney, Australia, 1996 (evening commute) 11.6 6.9 46.5 27.3 60.2 50.9 181 158 21.3 19.2 84.1 41.6 20.10 Jo and Park, 1999 Novamann, 1994 Leung and Harrison, 1999 Chan et al., 1991b Weisel et al., 1992 Duffy and Nelson, 1997 Measurement and Modelling of Motor Vehicle Related Air Toxics 20 28.8 16.1 o-xylene Figure 10. Mean concentrations of benzene (g/m3) standard deviation for the present study and previous studies in the literature 80 70 Concentration (ug/m3) 60 50 40 30 20 10 98 9 96 r1 te ,w in to n Bo s in rm Bi 19 94 gh am ,U K, m m su ro nt o, To nt o, ro To er r1 te w in ,N ei gh 19 99 4 98 8 C ,1 20 -b in w a, tta O Measurement and Modelling of Motor Vehicle Related Air Toxics 21 R al us , -b us , su m w in te m er r2 00 00 0 00 20 O tta w a, in O tta w a, w a, tta O in -c in ar -c ,s um ar ,w in te m er r2 00 0 0 Figure 11. PM2.5 mass concentration for a) 2000 wintertime sampling sessions and b) 2000 summertime sampling sessions. (Minimum, 25th percentile, 75th percentile, and maximum values indicated) 50 Concentration, ng/L 45 40 35 30 25 20 15 10 5 0 7:30 to 9:30 a.m. 11:30 a.m. to 1:30 p.m. 3:30 to 5:30 p.m. weekend 6 hr 3:30 to 5:30 p.m. weekend 6 hr Weekday 24 hr 50 Concentration , ng/L 45 3) 40 35 30 25 20 15 10 5 0 Weekday 24 hr 7:30 to 9:30 a.m. 11:30 a.m. to 1:30 p.m. The 24-hr samples were collected to compare the effects of averaging times on observed concentrations and coincided with 24 hour VOC samples collected on NAPS sampling days. The 24 hour median concentration was lower in the summer at 10.3 g/m3 compared to 18.3 g/m3 observed in the winter but there were only two samples in the summer. The median PM2.5 mass concentrations for the winter and summer 2000 are compared with TSP mass concentration measurements from the summer of 1994 in Figure 12. While TSP concentration is an order of magnitude higher than the PM 2.5 concentration, there is also a different pattern between TSP sessions and PM 2.5 sessions. The TSP mass concentration declines as the day progresses, whereas PM2.5 mass concentrations are higher during the morning and afternoon sessions when there was a higher volume of traffic than the midday session. Measurement and Modelling of Motor Vehicle Related Air Toxics 22 Figure 12. Mass concentration for winter 2000 PM2.5, summer 2000 PM2.5 and summer 1994 TSP. 160 Winter 2000 140 Summer 2000 Summer 1994 TSP Concentration, ng/L 120 100 80 60 40 20 0 7:30 to 9:30 a.m. 11:30 a.m. to 1:30 p.m. 3:30 to 5:30 p.m. weekend 6 hr NAPS 24 hr The elemental carbon and organic carbon content of PM2.5 for winter and summer 2000 sessions are presented in Figure 13. The median OC concentrations for the morning and evening sessions were higher than the noon and 6 hr sessions for both winter and summer, similar to the trends observed for total PM2.5 mass. Figure 13: OC and EC concentrations for winter and summer 2000 roadside PM2.5 samples (minimum, median and maximum values indicated) 30 EC OC Median Concentration (ug/m 3) 25 20 15 10 5 0 7:30 -9:30 Winter 7:30 -9:30 Summer 11:30 - 1:30 Winter 11:30 - 1:30 Summer 3:30 -5:30 Winter 3:30 -5:30 Summer w eekend 6 hr w eekend 6 hr Winter Summer NAPS 24 hr Winter NAPS 24 hr Summer The OC content comprised 70 to 90 % of the total carbon content for the winter samples and 45 to 75% for the summer samples. Elemental carbon on the other hand showed different trends between summer and winter with noticeably higher absolute values and shares in the summer. Measurement and Modelling of Motor Vehicle Related Air Toxics 23 Figure 14 presents concentrations of the 13 most abundant metals in PM2.5 samples for each wintertime sampling session. The predominant elements observed were silicon, sulphur and chlorine. The chlorine is likely from road salt, which is in abundant use in Ottawa during the winter months. Figure 14. Minimum, median and maximum PM2.5 metals concentration (g/m3) during each wintertime sampling session. 4.5 7:30 to 9:30 a.m. 11:30 a.m. to 1:30 p.m. 4.0 3:30 to 5:30 p.m. weekend 6 hr 24 hr 3 Min, Median and Max Concentration (ug/m ) 3.5 3.0 2.5 2.0 1.5 1.0 0.5 Metal Measurement and Modelling of Motor Vehicle Related Air Toxics 24 Zn Fe Mn Sc Ca K Cl S P Si Al Mg Na 0.0 3.3 Vehicle emissions sampling The emissions from two diesel city transit buses and two gasoline light duty vehicles were measured and characterized, primarily with a view to developing source emission profiles that were later used in source apportionment by Chemical Mass Balance receptor modelling. Diesel Buses The two in-use urban buses of different vintages were tested at 20 C and -10 C on a chassis dynamometer over the Central Business District (CBD) test cycle. The first bus had an older technology engine and was part of an exhaust after-treatment demonstration study conducted by ERMD. This bus was tested in two configurations, with the OEM muffler installed and with the oxidation catalyst installed. This particular bus sees in-use service with the local transit authority with the oxidation catalyst installed, but other buses of the same type have no exhaust after-treatment device installed. The second bus was a new, state-of-the-art urban bus just introduced into service by the local transit authority. Approximately 165 volatile organic compounds (non-methane hydrocarbons) in the range of C1 to C26 were determined in the samples collected from buses. Figure 15 presents the mass emission rate profiles over the entire range of compounds while Table 3 gives details of emission rates for the sub-set of compounds selected for the receptor modelling as outlined in Section 3.4. Complete data are available at the OMDB web site (7). The oxidation catalyst used with the older bus achieves noticeable reductions for some compounds while there seem to be increases in others. The new technology achieves much lower emission rates than the older bus in either configuration. For example, Benzene emissions are reduced by approximately 10% with the oxidation catalyst at 20 C while the new technology bus reduces benzene emissions by more than 90%. Reductions are less at the cold temperature but the new technology vehicle still achieves reductions of around 90%. Passenger Cars Two light duty vehicles were also tested (a 1994 model year Neon and a 2000 model year Intrepid) at 24 C and -10 C, using Cold Start and Hot Start LA-4 test cycles with local summer grade commercial fuel. These tests enabled the determination of nonmethane hydrocarbon profiles in the C2 to C12 range that characterize the exhaust emissions when the vehicles are operated in the cold transient, hot stabilized and hot transient modes of the test procedure used to regulate emissions from light duty vehicles. Figure 16 presents the normalized profiles for the sum of compounds selected for receptor modelling. Complete data are available at the OMDB web site (7). Measurement and Modelling of Motor Vehicle Related Air Toxics 25 180 Emission Rate (mg/mile) 160 20 oC 140 Muffler 120 OxyCat 100 New 80 60 40 20 0 1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 Compound ID Number 180 160 o Emission Rate (mg/mile) -10 C 140 Muffler 120 OxyCat New 100 80 60 40 20 0 1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 Compound ID Number Figure 15. Mass emission rates of C2-C26 compounds from buses operated at two different temperatures. Measurement and Modelling of Motor Vehicle Related Air Toxics 26 CMB Species # 2 5 8 9 6 13 7 27 26 4 42 30 12 40 31 28 29 37 24 11 15 17 21 34 14 38 23 33 18 22 25 39 32 3 36 41 35 19 10 20 16 ERMD Analysis Diesel # 1 2 3 4 5 6 9 11 12 15 18 21 22 23 24 25 27 28 31 32 34 35 38 46 47 50 52 53 54 57 61 63 67 71 72 75 77 83 94 97 102 122 methane ethylene acetylene ethane propylene propane isobutene/1-butene n-butane t2-butene c2-butene 2m-butane 1-pentene 2m1-butene n-pentane 2m-13-butadiene t2-pentene 2m2-butene 22-dm-butane cyclopentane 23-dm-butane 2m-pentane 3m-pentane n-hexane m-cyclopentane 24-dm-pentane benzene cyclohexane 2m-hexane 23-dm-pentane 3m-hexane 224-tm-pentane n-heptane m-cyclohexane/22-dm-hexane 234-tm-pentane toluene/233-tm-pentane 2m-heptane 3m-heptane/… n-octane/t12-dm-cyH e-benzene m&p-xylene/34-dm-heptane o-xylene & 112-tm-cyH 124-tm-benz/tb-benz/1-decene Emission data in mg/mile Muffler Muffler OxyCat 20C -10C 20C 3.23 5.21 163.82 145.90 125.21 30.52 28.34 0.86 2.25 72.55 64.33 51.38 0.59 0.15 1.87 OxyCat -10C 4.62 129.83 5.36 2.27 53.41 0.27 New 20C New -10C 54.56 57.83 3.11 11.78 13.76 21.10 2.72 0.16 0.41 0.31 0.48 1.10 0.86 0.61 0.47 0.19 0.48 1.31 0.65 1.37 0.94 0.63 4.52 0.12 1.82 1.14 1.34 3.52 0.26 1.10 0.89 0.79 0.19 0.71 1.57 1.01 0.80 3.30 0.24 0.41 0.20 0.45 1.27 0.64 0.10 0.08 0.20 0.31 3.08 6.64 0.04 27.53 0.34 0.21 0.13 0.31 0.59 0.08 0.69 1.08 3.68 3.72 0.35 27.42 1.33 0.82 0.49 1.18 0.35 0.16 1.24 0.60 2.70 1.24 0.19 21.90 1.39 1.12 0.64 4.90 26.01 1.26 0.90 0.52 14.02 1.88 0.15 0.24 0.29 2.54 0.42 0.11 0.10 0.39 10.08 1.28 0.06 30.09 6.53 1.27 51.22 13.72 41.13 22.96 61.22 10.00 5.19 0.09 26.81 2.17 5.06 50.83 16.61 42.08 23.05 63.19 7.80 5.60 0.19 20.42 2.73 5.88 47.07 9.07 27.81 12.75 33.78 9.99 4.98 0.09 24.09 6.02 5.16 50.54 12.29 34.10 16.78 46.04 0.33 0.34 0.27 0.44 0.48 0.63 0.72 0.35 0.45 0.48 1.89 1.23 0.25 0.73 0.71 1.04 1.97 1.20 2.66 0.56 0.74 0.85 0.18 0.76 1.39 2.82 0.68 1.37 0.15 0.24 0.10 0.05 0.46 0.20 0.07 0.12 0.03 0.21 0.37 0.07 0.12 0.69 0.40 Table 3. Mass emission rates of selected compounds from buses operated at two different temperatures. Measurement and Modelling of Motor Vehicle Related Air Toxics 27 0.12 0.66 0.16 op a n- ne bu ta ne c2 -b ut en e 1pe nt en e npe nt an e t2 -p en te 22 ne -d m -b ut 23 an -d e m -b ut an 3m e -p en m ta -c ne yc lo pe nt an e be nz en e 2m -h ex an e 3m -h ex an e nhe 23 pt an 4tm e -p en ta ne 2m n-h oc m ep ta & ne ta 12 pne xy /t1 4le tm 2ne dm -b / en 34 -c ze -d yH m ne -h /tb e -b pt an en e z/ 1de ce ne pr e ne le n et ha et hy Fraction of named compounds hy le ne et ha ne pr op an e nbu ta ne c2 -b ut en 1pe e nt en ne pe nt a ne t2 -p 22 ent e -d ne m 23 but a -d ne m -b ut an 3m e -p m en -c ta yc ne lo pe nt an e be nz e 2m ne -h ex 3m ane -h ex an nhe e 23 p 4tm tan e -p en ta n2 ne m oc 12 m & -h ta p ep 4ne tm -xyl t /t1 an en -b 2e en e/ dm 34 ze -d cy ne m H /tb -h -b ep en ta z/ ne 1de ce ne et Fraction of named compounds LDGV (Intrepid) Exhaust, FTP at 20 C 0.18 0.16 0.14 0.12 0.1 0.08 Bag 1 Bag 2 Bag 3 0.06 0.04 0.02 0 LDGV Exhaust (Neon), FTP at - 10 C 0.16 0.14 0.12 0.1 0.08 Bag 1 Bag 2 0.06 Bag 3 0.04 0.02 0 Figure 16. Normalized C2-C12 HC emission profiles of light duty vehicles under different operating conditions Measurement and Modelling of Motor Vehicle Related Air Toxics 28 3.4 Chemical Mass Balance Receptor Modelling Source apportionment for the VOC concentrations measured on Slater Street was carried out using the Chemical Mass Balance receptor model, CMB8. (8) It can be expected that the total VOC concentrations measured in ambient samples on a major downtown street will be dominated by the emission sources related to mobile sources. Other emission sources such as architectural coatings, emissions related to graphic arts, etc. can be considered, although at significantly lower strengths. Major source types of interest for mobile source emissions are tailpipe emissions under different modes of operation, resting evaporative emissions associated with "hot soak" and diurnal heat buildup situations, "running" emissions that combine tailpipe and evaporative emissions for moving vehicles, and the emissions associated with the fuelling operations. A number of source profiles have been published characterizing these source types and the SPECIATE database provides access to a large compilation (9). These profiles can contain as many as 75 hydrocarbon species in the C2-C10 range, comprising saturated, unsaturated, cyclic and aromatic species. The profiles available in the SPECIATE database were augmented by the diesel and gasoline exhaust profiles developed through vehicle emission testing completed as part of the current study, as explained in Section 3.3 above. The complete list of profiles included in the CMB8 modelling exercise is presented in Table 4. The list of compounds to be used in receptor modelling was governed by the availability of component data in the SPECIATE profiles and the relative importance of the compound in terms of the total mass measured in the ambient samples. The CMB8 Application and Validation document (10) focuses on the target list of compounds for the PAMS (11) network in the U.S. and identifies the compounds used as fitting species in source apportionment studies with this data set. The CMB8 Application and Validation document is summarized in a review article (12). The list of chemical compounds used in the modelling exercise was arrived at by reconciling the list of compounds from the above sources with the list of compounds that were responsible for ~90% of the total measured mass in our ambient samples. The complete list is presented in Table 3, previously mentioned in connection with diesel bus source profiling. CMB8 modelling results are presented below for the median profiles for each of the three daily sampling sessions in the two seasons. Screening runs were first completed with various source profiles from Table 4 to arrive at a set that included the types of sources expected to contribute to ambient measurements and explain the observations. Table 5 presents CMB8 summary results with a profile set that includes some profiles from the SPECIATE database and some profiles that were obtained from the vehicle emission testing completed in this study. Figure 17 presents these data in graphical form. Measurement and Modelling of Motor Vehicle Related Air Toxics 29 CODE DIES1 DIES2 DIES3 DIES4 DIES5 DIES6 DIES7 DIES8 LPG01 GASL1 GASL2 GASV1 GASV2 GASV3 GASV4 GASV5 GASE1 GASE2 GASE3 GASE4 GASE5 GASE6 GASE7 GASE8 GASE9 HOSO1 HOSO2 HOSO3 HOSO4 HOSO5 HOST1 REFU1 COST1 COST2 COST3 COST4 COST5 COMB1 COMB2 DRYC1 PRNT1 PRNT2 ROAD WGAS1 WGAS2 WGAS3 WGAS4 HESP1 HESP2 DM20 DM-10 DOC20 DOC-10 DN20 DN-10 BAG120C BAG220C BAG320C BAG110C BAG210C BAG310C SPECIATE profile # and description #2530/Vehicle Exhaust - Van Nuys Tunnel, Diesel and Minimum Running Loss Subtracted - June 8-12, 1995 #2524/Vehicle Exhaust - Sepulveda Tunnel Diesel and Minimum Running Loss Subtracted - Oct. 3-4, 1995 #2525/Vehicle Exhaust - Sepulveda Tunnel Diesel and Maximum Running Loss Subtracted - Oct. 3-4, 1995 #2518/Vehicle Exhaust - Lincoln Tunnel Diesel and Minimum Running Loss Subtracted - Aug. 16-18, 1995 #2519/Vehicle Exhaust - Lincoln Tunnel Diesel and Maximum Running Loss Subtracted - Aug. 16-18, 1995 #2513/Vehicle Exhaust - Callahan Tunnel Diesel Exhaust Subtracted - Sept. 18-19, 1995 #2514/Vehicle Exhaust - Callahan Tunnel Diesel and Minimum Running Loss Subtracted - Sept. 18-19, 1995 #2015/Vehicle Exhaust - Callahan Tunnel Diesel and Maximum Running Loss Subtracted - Sept. 18-19, 1995 #2445/LPG from Servigas & Commercial de Juarez - 1996 #2448/Composite Gasoline Liquid, El Paso - 1996 #2447/Composite Gasoline Liquid from Los Angeles, Summer 1995 Fed Phase 1 RFG #2452/Gasoline Vapor, Hot-Soak, Downwind-Upwind Sample from the Astrodome - 1996 #2451/Gasoline Vapor, Hot-Soak, Downwind Sample from the Astrodome - 1996 #2453/Composite of 14 Gasoline Headspace Vapor Samples - 1996 #2454/Composite Gasoline Vapor from Los Angeles, Summer 1995 #2450/Composite Gasoline Vapor from Boston, Summer 1995, Fed Phase 1 RFG #2521/Vehicle Exhaust - Tuscarora Tunnel Light Duty Gasoline - 1995 #2527/Vehicle Exhaust - Fort McHenry Tunnel Light-Duty Gasoline - 1995 #1203/Light-Duty Gasoline Vehicles - Exhaust #1204/Light-Duty Gasoline Vehicles - Evaporative #1101/Light Duty Gasoline Vehicles - 46 Car Study #1313/Industry Average (circa 1990) Gasoline Exhaust #1186/Heavy Duty Gasoline Trucks #1315/11% MTBE Exhaust #1314/10% Ethanol Exhaust #2502/Vehicle Exhaust - Older Fleet (1983-1985) Hot Soak Evaporative #2495/Vehicle Exhaust - Current Fleet (1989) Hot Soak Evaporative #2452/Gasoline Vapor, Hot-Soak, Downwind-Upwind Sample from the Astrodome - 1996 #2451/Gasoline Vapor, Hot-Soak, Downwind Sample from the Astrodome - 1996 #1311/11% MTBE Hot Soak #2501/Vehicle Exhaust - Older Fleet (1983-1985) Hot Start #1100/Refueling #2505/Vehicle Exhaust - Tip O'Neill Garage (Boston) Cold Start - Sept.12-13, 1995 #2499/Vehicle Exhaust - Older Fleet (1983-1985) Cold Start #2492/Vehicle Exhaust - Current Fleet (1989) Cold Start #2506/Vehicle Exhaust - Cold-Start, Downwind Sample from the Astrodome - 1993 #2507/Vehicle Exhaust - Astrodome, Cold Start, Downwind-upwind.- 1993 #1001/Internal Combustion Engine - Natural Gas #9002/Internal Combustion - Average #9017/Drycleaning/Degreasing - Average #9026/Printing/Publishing - Average #2553/Offset Printing - Plant C, Room Composition #2564 Atlanta roadway #7003 Atlanta whole gas weighed average of all octanes #7004 Atlanta whole gas 87 octane #7005 Atlanta whole gas 89 octane Atlanta whole gas 92/93 octane Conner, T.L., Lonneman, W.A., Seila, R.L., 1995, AWMA Journal, Vol.45, pp. 383-394. Atlanta headspace gas 24C Conner, T.L., Lonneman, W.A., Seila, R.L., 1995, AWMA Journal, Vol.45, pp. 383-394. Atlanta headspace gas 32 C Conner, T.L., Lonneman, W.A., Seila, R.L., 1995, AWMA Journal, Vol.45, pp. 383-394. ERMD Diesel bus (original muffler) at 20 C ERMD Diesel bus (original muffler) at - 10 C ERMD Diesel bus with oxidation catalyst at 20 C ERMD Diesel bus with oxidation catalyst at - 10 C ERMD New Technology diesel bus at 20 C ERMD New Technology diesel bus at - 10 C ERMD FTP Phase 1 (Cold transient) at 20 C from Intrepid ERMD FTP Phase 2 (Hot stabilized) at 20 C from Intrepid ERMD FTP Phase 3 (Hot transient) at 20 C from Intrepid ERMD FTP Phase 1 (Cold transient) at - 10 C from Neon ERMD FTP Phase 2 (Hot stabilized) at - 10 C from Neon ERMD FTP Phase 3 (Hot transient) at -10 C from neon Table 4. Source HC profiles used in CMB modelling Measurement and Modelling of Motor Vehicle Related Air Toxics 30 SITEID ROSUM7 ROSUM11 ROSUM15 DATE Median Median Median ST 7:30 11:30 15:30 DR 2 2 2 SIZE VOC VOC VOC CONC 75.98 68.52 89.08 UCONC 15.45 15.96 18.96 RSQUAR 0.88 0.87 0.88 CHISQUAR 1.72 2.11 2.03 PCMASS 94.79 97.49 99.13 9 6.82 5.15 4.64 LPG01 0.82 0.63 0.60 20 0.68 1.14 1.23 GASE4 1.03 0.88 0.88 SITEID ROWIN7 ROWIN11 ROWIN15 DATE Median Median Median ST 7:30 11:30 15:30 DR 2 2 2 SIZE VOC VOC VOC CONC 72.86 92.36 195.33 UCONC 15.15 17.60 28.50 RSQUAR 0.98 0.97 0.97 CHISQUAR 0.52 0.57 0.71 PCMASS 103.12 102.32 100.92 9 7.01 7.68 4.65 LPG01 1.10 1.27 1.63 23 GASE7 41 PRNT1 54 DOC20 58 BAG120C 60 BAG320C 0.98 1.96 2.45 2.88 39.28 6.42 21.82 3.66 1.25 1.70 0.00 0.00 42.29 4.79 16.98 3.02 0.52 2.25 0.00 0.00 67.68 5.84 14.24 3.54 38 COMB1 41 PRNT1 55 DOC-10 61 BAG110C 63 BAG310C 2.46 2.37 4.74 1.44 1.93 1.97 4.37 2.58 38.50 5.05 16.13 6.04 3.59 2.96 4.58 1.67 1.12 2.19 7.15 3.77 48.82 6.12 21.56 7.28 1.20 5.66 0.54 2.72 2.46 3.04 2.57 4.02 142.93 12.26 42.77 13.98 Table 5. Summary output from CMB8 Modelling CONC: Measured concentrations, micrograms per cubic meter UCONC: Uncertainties in measured concentrations, micrograms per cubic meter RSQUAR, CHISQUAR: CMB8 goodness of fit criteria PCMASS: Modelled total concentration as % of measured total concentration 9, 20, 23, 38, 41, 54, 55, 58, 61, 63: Source profile numbers, the values in the rows containing these source numbers are the contribution of the respective source to the total modelled concentration. LPG01, GASE4, GASE7, COMB1, PRNT1, DOC20, DOC-10, BAG120C, BAG110C, BAG320C, BAG310C: Source codes, the values in the rows containing these source codes are the uncertainties in the contribution of the respective source, one line above. Measurement and Modelling of Motor Vehicle Related Air Toxics 31 Winter HC Source Apportionment 250.00 micrograms per cubic meter 200.00 BAG310C BAG110C DOC-10 PRNT1 COMB1 GASE7 LPG01 150.00 100.00 50.00 0.00 7:30 11:30 15:30 Sessions Summer HC Source Apportionment 100.00 90.00 micrograms per cubic meter 80.00 70.00 BAG320C BAG120C DOC20 PRNT1 GASE4 LPG01 60.00 50.00 40.00 30.00 20.00 10.00 0.00 7:30 11:30 15:30 Sessions Figure 17. Source contribution estimates during sampling sessions Measurement and Modelling of Motor Vehicle Related Air Toxics 32 Discussion of CMB8 Results For the summer sessions, the sources with estimable contributions were: LPG01 Liquefied petroleum gases GASE4 Light Duty Gasoline Vehicle evaporative emissions PRNT1 Printing/publishing emissions DOC20 Heavy Duty Diesel Vehicle exhaust BAG120C Light Duty Gasoline Vehicle Exhaust in the cold transient mode BAG320C Light Duty Gasoline Vehicle Exhaust in the hot transient mode For the winter sessions, the sources with estimable contributions were: LPG01 Liquefied petroleum gases GASE7 Heavy Duty Gasoline Vehicle Exhaust COMB1 Natural Gas vehicle exhaust PRNT1 Printing/publishing emissions DOC-10 Heavy Duty Diesel Vehicle exhaust BAG110C Light Duty Gasoline Vehicle Exhaust in the cold transient mode BAG310C Light Duty Gasoline Vehicle Exhaust in the hot transient mode In addition to the source profiles with estimable contributions listed above, the final set that was tried included the following two profiles that were not found to have estimable contributions: for the summer sessions: BAG220C Light Duty Gasoline Vehicle Exhaust in the hot stabilized mode (20 C) for the winter sessions: BAG210C Light Duty Gasoline Vehicle Exhaust in the hot stabilized mode (-10 C) The immediate result from the above modelling is that the measured concentrations are explained mostly by light duty gasoline vehicle exhaust. Contributions from other sources such as heavy-duty diesel vehicle exhaust or light duty gasoline vehicle evaporative emissions that may have been expected to be significant are at relatively low levels and associated with high uncertainties. The dominance of light duty gasoline vehicle exhaust is in contrast to source apportionment carried out earlier (13) on data collected in the Summer of 1994 when evaporative sources and heavy duty diesel exhaust were found to be more significant contributors. The 1994 data are all in the summer season while the 2000 data are split approximately 3:1 in favour of winter samples. The evaporative GASE4 profile appears in the summer sessions, albeit somewhat weaker than might be anticipated. Its absence from winter sessions is understandable. Although locally derived exhaust profiles (from ERMD) were Measurement and Modelling of Motor Vehicle Related Air Toxics 33 available for the modelling exercise, the evaporative profile used comes from a study in Atlanta and may not be particularly representative of evaporative profiles in Ottawa. In the absence of better resolution of evaporative type emissions from exhaust emissions it may be reasonable to suspect that some of the LDGV exhaust contribution may in fact be a surrogate for evaporative emissions. It was notable that among the HDDV exhaust profiles derived in this study, the profile with the oxidation catalyst gave the better fit. For the LDGV exhaust profiles, the cold start and hot start modes were favoured, relative to the hot stabilized mode of operation. The more distinctive features of HDDV exhaust relative to LDGV exhaust are manifested in the higher hydrocarbons C12-C26. When the list of compounds used in the modelling does not extend into this range due to lack of ambient data, it becomes more difficult to resolve the two types of exhaust. The absence of the GASE7 profile from the summer sessions is not easily explainable since the presence of heavy-duty gasoline vehicles in the fleet should not be season related. On the other hand, the contribution is a small one, associated with large relative uncertainty. The difference should therefore not be given undue weight. The presence of the COMB1 profile only in the winter is equally surprising from the vehicle fleet perspective but may be explained by possible contributions from stationary natural combustion sources in the winter. Its contribution is also small and associated with a large relative uncertainty. The overall statistical indicators of RSQUAR and CHISQUAR in Table 5 should be considered good (8), although the winter sessions have somewhat better indicators. The uncertainties in the source contribution estimates (t-statistics) should be considered good for the major sources but not adequate for the minor sources (8). Figure 18 demonstrates the relative importance of the uncertainties for the particular case of the winter 7:30 session. The goodness of fit in terms of individual components is demonstrated in Figure 19 for the same session. While most compounds show good agreement between measured and modelled concentrations (thus resulting in the good overall statistics mentioned above), some compounds have significant difference between measured and modelled concentrations justifying further investigation into source and ambient profiles. Conclusions of Source Apportionment from CMB8 Modelling The measured concentrations of hydrocarbons on Slater Street in 2000 (in the range ~70200 micrograms per cubic meter) are explained dominantly by light duty gasoline vehicle exhaust in the cold start and hot start modes of operation. These source contributions are determined by using locally measured light duty gasoline vehicle exhaust profiles. Locally derived evaporative emission profiles and the extension of ambient data to heavier hydrocarbons may enable finer resolution between exhaust vs. evaporative emissions for gasoline vehicles, and gasoline vs. diesel exhaust emissions. Measurement and Modelling of Motor Vehicle Related Air Toxics 34 Uncertainties in source contribution estimates (Winter 7:30) 50.00 45.00 40.00 micrograms per cubic meter 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 LPG01 GASE7 COMB1 PRNT1 DOC-10 BAG110C BAG310C -5.00 Figure 18. Uncertainties in source contribution estimates Measured and modelled profiles (Win 7:30) 9 Data 8 Model 7 micrograms per cubic meter 6 5 4 3 2 1 Measurement and Modelling of Motor Vehicle Related Air Toxics 35 TM BE MP XY HE NO CT 2M TM PN HE Figure 19. Goodness of fit for individual compounds NH EP 3M HE 2M BE NZ MC PE PE 3M DM BU E DM BU T2 P NP EN 1P EN C2 BU PR O NB UT -1 ET H EH Y 0 4. DISSEMINATION OF RESULTS The microenvironment database that has been generated in this study, including the earlier study in 1994, is being made available to interested researchers through a web page: http://www.carleton.ca/~dkarman/OMDB The web site includes detailed descriptions of the experimental methodology and pictorial orientation to the Slater street site, as well as complete data in electronic spreadsheet format. Additional figures, too numerous to include in this report are also available in PowerPoint format. It is hoped that in this way the data will be available to wider scrutiny, analysis, and discussion. The following conference presentations have been made: 1. Karman, D., Noseworthy, L., Graham, L. A., Oguz, O., Motor Vehicle Related Air Toxic Measurements Along an Urban Street, International Symposium on the Measurement of Toxic and Related Air Pollutants, Air & Waste Management Association & U.S. EPA, Research Triangle Park, September 2000. 2. Karman, D., Oguz, O., Akay, G., and Graham, L. A., Measurement of Air Toxics in the Cabins of Commuter Vehicles Under Summer and Winter Conditions in Ottawa, Canada, 11TH CRC ON-ROAD VEHICLE EMISSIONS WORKSHOP, San Diego, California, March 26-28, 2001. 3. Karman, D., Welburn, C., and Graham, L.A. Mass Emissions Rates and Chemical Characterization of PM2.5 Emissions from Two motorcycles, 11TH CRC ONROAD VEHICLE EMISSIONS WORKSHOP, San Diego, California, March 26-28, 2001. 4. Oguz, O., Tuncel, G., Karman, D., Trafikten Kaynaklanan Uçucu ve Yarı Uçucu Organik Bileşiklerin Belirlenmesi ve Mevsimsel Değişimlerinin İncelenmesi, IV. National Conference of Turkish Chamber of Environmental Engineers, İçel, Turkey, September 7-10, 2001. 5. Oguz, O., Tuncel, G., Karman, D., Measurement of Traffic Related Toxic Air Pollutants in an Urban Atmosphere, 2nd International Symposium on Air Quality Management at Urban, Regional and Global Scales, Istanbul, Turkey, September 2528, 2001. 6. Karman, D., Graham, L., Measurement and Modelling of Motor Vehicle Related Air Toxics Along Urban Streets, TSRI Regional Conference – Urban Air, Vancouver, November 2001. 7. Karman, D., Oguz, O., Akay, G., Graham, L., The Ottawa Microenvironment Database for Motor Vehicle Related Air Pollutants, TSRI National Conference, Ottawa, March 2002. Measurement and Modelling of Motor Vehicle Related Air Toxics 36 8. Karman, D., Noseworthy, L., Graham, L., Measurement and Characterization of PM2.5 in Urban Air, TSRI National Conference, Ottawa, March 2002. 9. Karman, D., Tuncel, G., Graham, L., Oguz, O., Spatial Variations in the Concentrations of Motor Vehicle Related Organic Air Pollutants in Ottawa, TSRI National Conference, Ottawa, March 2002. 10. Oguz, O., Tuncel, G., and Karman, Volatile Organics in an Urban Atmosphere, Symposium 2002 EUROTRAC2, Transport and Chemical Transformation in the Troposphere, Garmisch-Partenkirchen, Germany, 11-15 March, 2002. The following four papers are planned for submission to Atmospheric Environment: COMPARISON OF 2-HOUR AND 24-HOUR VOC MEASUREMENTS ALONG AN URBAN STREET, Karman, D., Noseworthy, L., Graham, L., Oguz, O. SEASONAL VARIATION OF PM2.5 CONCENTRATIONS AND CHARACTERISTICS ALONG AN URBAN STREET, Karman, D., Noseworthy, L., Graham, L. SOURCE APPORTIONMENT FOR MOTOR VEHICLE RELATED VOCs ALONG AN URBAN STREET, Karman, D., Graham, L., Oguz, O. SOURCE APPORTIONMENT FOR PM2.5 MEASUREMENTS ALONG AN URBAN STREET, Karman, D., Noseworthy, L., Graham, L. 5. REFERENCES 1. Lawryk, N.J., 1996. “Automobile commuter exposures to volatile organic compounds: Emissions, malfunctions, and policy”, Transportation Research Part A: Policy and Practice, Vol. 30, Iss. 1, pp. 55-56. 2. Wallace, L.A., 1993. In “Chemistry and Analysis of Volatile Organic Compounds in the Environment; Bloemen, H.J.TH.; Burn, J., Eds.; Blackie Academis and Professional (Chapman&Hall): Galsgow, 1993; pp. 1-24. 3. U.S. Environmental Protection Agency, 1993. “Motor Vehicle Related Air Toxics Study”, Office of mobile sources, Office of Air and Radiation: Ann Arbor, MI, USA. 4. U.S. Environmental Protection Agency, 1990. “Cancer Risk from Outdoor Exposure to Air Toxics”, Vol. 1, Final Report, EPA, 450/1-90-004a. Office of Air Quality Planning Standards, Research Triangle Park, NC, USA. 5. Chan, C.C., Ozkaynak, H., Spengler, J.D., and Sheldon, L., 1991. “Driver exposure to volatile organic compounds, CO, ozone, and NO2 under different driving conditions”, Environmental Science and Technology, Vol. 25, pp.964-972. Measurement and Modelling of Motor Vehicle Related Air Toxics 37 6. Wan-Kuen, J., and Kun-Ho, P., 1999. “Commuter exposure to volatile organic compounds under different driving conditions”, Atmospheric Environment, Vol. 33, pp. 409-417. 7. The Ottawa Micro-environment Database (OMDB) www.carleton.ca/~dkarman/OMDB 8. U.S. EPA 2001, CMB8 User's Manual, Office of Air Quality Planning and Standards, Research Triangle Park. 9. U.S. EPA 2000 & 2002, SPECIATE 3.1 and 3.2, EPA’s repository of Total Organic Compound (TOC) and Particulate Matter (PM) speciated profiles for a variety of sources for use in source apportionment studies. Available at: http://www.epa.gov/ttn/chief/software/speciate/index.html 10. CMB8 Applications and Validation Protocol for PM2.5 and VOCs, Desert Research Institute, Reno, Nevada, September 1998. 11. U.S. EPA, Enhanced Ozone Monitoring – PAMS, Photochemical Assessment Monitoring Stations, http://www.epa.gov/oar/oaqps/pams/general.html 12. Watson, J.G., Chow, J.C., Fujita, E.M., 2001, “Review of volatile organic compound source apportionment by chemical mass balance”, Atmospheric Environment, Vol.35, pp.1567-1584. 13. Karman, D., Wong, K., O'Leary, K., Graham, L., 1997, “Source Apportionment for VOCs in Micro-Environments using Chemical Mass Balance Receptor Modelling”, Environmental Research Forum Vols.7-8, pp.158-163, Proceedings of the 10th Regional IUAPPA Conference, Istanbul, 1997. Measurement and Modelling of Motor Vehicle Related Air Toxics 38