- Email: [email protected]
Local aspects of vehicular pollution
Pergamon
Atmospheric
PII: S1352-2310(96)00205-l
LOCAL ASPECTS OF VEHICULAR M. J. CLIFFORD,
R. CLARKE
Environmew Vol. 31, No. 2, pp. 211-276, 1997 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 1352-2310/97 $15.00 + 0.00
POLLUTION
and S. B. RIFFAT
Institute of Building Technology, Department of Architecture and Building Technology, University of Nottingham, University Park, Nottingham, NG7 2RD, U.K. (First received 2 April 1995 and injnalform
13 March 1996)
Abstract-We consider the local aspects of vehicular pollution by the use of 1: 10 scale models. The model automobiles ‘were placed in a low-velocity wind tunnel to simulate a queue of slow-moving city traffic and tracer gasses were then injected into the air flow though the tail pipes fitted to two of the models. Measurements show that the exhaust gases are entrained in the wake of the vehicle from which they were emitted, and are dispersed mainly by the passage of the following car. The spatial distribution of tracer gas along and across the models suggests that the level of pollution received by a commuter in a slow-moving heavy traffic may be strongly influenced by the pollution produced from the car immediately in front. Copyright c 1996 Elsevier Science Ltd Key word index: Traffic pollution, wind tunnel testing, air quality modelling.
INTRODUCTION
Air pollution from vehicles is of increasing concern, and the issue is of prime importance in many cities in the U.K. and the rest of the developed world. The health aspects of pollution are not yet fully understood, but the increase in asthma, other respiratory diseases and certain types of cancer are being linked to pollutants emitted from vehicles such as carbon monoxide, oxides of nitrogen, and volatile organic compounds (Chan et al., 1991a, b). Consequently, it is of great importance to measure the individuals’ exposure to these pollutants in the course of their daily activities, both in the workplace, and whilst commuting to and from work (Peterson and Sabersky, 1975a; Ott et al., 1988; Ha:yes, 1989; Ott, 1990). Studies have shown that levels of pollution experienced whilst commuting by road to and from the workplace are much higher than those experienced at work or at home, and an individual receives most of his daily dose of pollution whilst commuting (Flachsbart and Yo, 1987). Hence these levels and the resulting dose of pollution should be closely monitored. Several methods have been developed which detect the concentration of various pollutants at the roadside (Brice and Roesler, 1966; Chan et al., 1994), but the levels of pollutants inside a vehicle are generally greater than these roadside measurements (Chan et al., 1991a, b). There may be several reasons for this. Firstly, the level of pollution decays away from the source, and consequently the levels around the vehicles in traffic will be higher than the levels at the roadside. This makes careful location of monitoring equipment essential. Secondly, the method of ventilation used by indi-
vidual vehicles may serve to concentrate the levels of pollutants inside the vehicle if the air conditioning and the state of windows, sun roof, etc. are not effective, or conversely an efficient ventilation system may reduce the levels (Stricker and Dick, 1988). Conflicting reports exist as to whether it is better to drive with the windows open or closed in heavy traffic (Peterson and Sabersky, 197513;Colwill and Hickman, 1980). The air space inside a vehicle can act as a buffer against sudden high levels of pollution, but this would also slow down the beneficial effect of driving from an area with a high level of pollution into an area with a lower level (Flachsbart and Yo, 1987). Simplified models assuming complete mixing inside the vehicle have been proposed to measure this buffer effect (Ott and Willits, 1981; Ott et al., 1994). An alternative approach is to measure the pollutant concentration inside a vehicle whilst driving in traffic. Several techniques are available, either using a mobile gas sensor, or collecting samples of vehicular air for subsequent analysis. The mobile gas detector method has the advantage of not only giving the total pollution dose, but also the pollutant concentrations at various landmarks which can be compared with climate, traffic density, queuing time at junctions, and so on. However, it may be necessary to collect gas samples or to use some other total exposure method to detect volatile organic compounds. Several studies have been carried out along these lines (Peterson and Allen, 1982; Flachsbart et al., 1987; Chan et al., 1991a, b; Ott et al., 1994), and the results may be useful in estimating the average exposure of a typical commuter. However, a more local aspect of the problem was shown by Colwill and Hickman (1980) in an 271
212
M. J. CLIFFORD et al.
shapes. Wind tunnel testing has been used to examine the theory that the pollution to which a driver is exposed may be a local phenomenon, and in the future this will be followed up with experimental observations in traffic and computational fluid dynamics, which is the subject of current research.
early study which considered the levels of pollution in several types of vehicles driven along the same route. The levels varied considerably from vehicle to vehicle. Clearly a more fundamental approach is needed if general statements are to be made about the exposure of commuters to pollutants. An important aspect ignored by the studies already mentioned is that most of the pollution emitted from a vehicle comes from a specific location: the exhaust. This may lead to significant local effects in the concentration of pollutants around the traffic. Given that the levels of pollution are highest in slowmoving traffic, on calm days, especially when approaching or accelerating away from road junctions, it may be beneficial to consider the dispersion of the pollution produced at the exhaust by the wind, and the air flow around the adjacent vehicles in a more controlled setting. The hypothesis here is that the levels of pollution in a vehicle may be strongly influenced by the levels of pollution produced from the exhaust of the vehicle immediately in front. Whilst this theory falls down at high speed with strong cross winds, at low speed, and on calm days, i.e. the worst case, it seems more plausible. The experimental observations of Chaney (1978) bear this out. He drove at a constant speed of 52 m.p.h. along a highway as various vehicles overtook his vehicle and pulled into his lane about 50 ft ahead. At the differential speed from 3 to 10 m.p.h., he noticed significant peaks in the level of carbon monoxide that were directly attributable to the vehicle in front. Clearly, driving behind such a vehicle for a significant time in slow-moving traffic would have an effect on the level of pollution inside a vehicle. Such observations have prompted the current work, which is concerned with modelling the dispersion of pollutants around three simplified vehicle
-
+
+
+
+
+
WIND TUNNEL TESTINGLOW-VELOCITY ATMOSPHERIC WIND TUNNEL
The tunnel used for this series of tests was based on a small open-jet wind tunnel developed for teaching purposes. The tunnel has a maximum flow rate of 4.5 ms-’ and a working section of width l.Om, height 0.75 m and length 2.25 m. Two layers of honeycomb are fitted in the bellmouth to straighten the incoming flow and are followed by a 0.5 mm mesh screen. Beyond the entry section there is a 0.75 m settling chamber to allow disturbances caused by the honeycomb and screen to decay. A transformer is positioned between the working section and the fan to convert the l.Om x 0.75 m cross section to a suitable shape for the fan mounting. The entry to the transformer is larger than the settling chamber exit to allow for some expansion of the flow through the working section. The fan speed is controlled by a variable autotransformer which varies the voltage supplied to the drive motor. The tunnel wind speed was measured to determine any velocity variations across the tunnel section. Two sets of data were recorded, the first measured 0.5m downstream of the settling chamber, and the second a further 1 m downstream. Only negligible velocity variations were found ( + 3%), although the turbulence intensity was higher at the second location. The flow through the tunnel was found to be stable up to 0.2m from the edge of the table.
+
+
--
+ I \
/
;
Fig. 1. MIRA vehicle with pressure tappings.
Local aspects of vehicular pollution SCALE MODEL CONSTRUCTION
Three hollow model cars based on the Motor Industry Research Association (MIRA) model were constructed from 1Omm medium density fibreboard (MDF) at a scale of 1: 10 (Fig. 1). The cars were approximately 40 cm long. The models were deliberately simple, and no attempt was made to model fine detail. Cars 1 and 2 were fitted with a single injection tube to simulate the engine exhaust which was positioned on the right-hand side at the rear of the two vehicles. This had 1 mm internal diameter (ID) to allow simulation of the measured flow of exhaust from a car whilst the engine was idling (15-20 m s-l). SF, tracer gas was injected at the exhaust of car 1, and N,O tracer gas was injected at the exhaust of car 2 at a rate of 1 emin-’ and 250,000 parts per million (ppm) to simulate the carbon monoxide concentration in a typical exhaust. Car 3 was fitted with several static pressure tappings (holes) to allow measurement of gas concentration and pressure around the car body surface. These tappings were machined from 0.8 ID brass tubing, and fitted on all of the upper body surfaces. The modals were painted and no attempt was made to measure surface roughness. Plastic tubing was attached to the tappings through a hole in the wind tunnel table to allow the tappings to be connected to the instrumentation.
INSTRUMENTATION
Twenty of the model pressure tappings were connected to a selection box. Gas analysis was carried out using a Binos 1000 gas analyzer. The instrumentation
273
was connected to a Datron 1055 data logging system controlled by a portable PC (Fig. 2).
MODEL ARRANGEMENT
The three model cars were placed in line on the tunnel floor 20 cm apart to simulate a full scale gap of 2 m (approximately half a car length). The car with the pressure tappings (car 3) was placed at the end of the line. Both the exhausts were switched on, and readings of SF6 (car 1) and NzO (car 2) concentration were taken from all of the pressure tappings on car 3 with the wind speed set at 1.5 m s-l. Since the flow around the vehicles was turbulent, it was necessary to average the results over 20s on each tapping. When the test had been completed, the wind speed was altered. The test procedure was then repeated over a range of wind tunnel speeds. Hence data on the proportion of pollution received from cars 1 and 2 by car 3 was obtained for a range of operating conditions.
RESULTS
The data from each of the tests at a constant wind tunnel speed shows that there is considerable variation in the tracer gas concentration over the surface of car 3 from car 2. This can be seen in Fig. 3 where the concentration of NzO (that is, the pollution from car 2) along the centre-line of car 3 is plotted for three wind tunnel speeds. A sketch of car 3 is added to aid visualization. The three curves show the same general trend in that the pollution concentration falls sharply
Tunnel Inlet L
““M:
i
(0.75lhin)
Fig. 2. Wind tunnel setup: NzO supply to car 2 is not shown.
M. J. CLIFFORD et al.
274
Concentration (ppm) 1.5mls
N,O (car 3mi.s N,O (car 4.5mIs N,O (car f
Fig. 3. Concentration of NzO (car 2) along the centre-line of car 3.
Concentration (ppm) 120
80 60
\/
J
Fig. 4. Concentration of SF6 (car 1) along the centre-line of car 3.
across the front hood (bonnet) of car 3, and continues to fall more gradually along the roof, but recovers slightly towards the rear of the vehicle. This trend is more pronounced at low wind speed. The reason for this pattern can be deduced by considering the flow pattern around the vehicle. As expected, the concentration falls along the length of the vehicle as the distance from the source increases. The rise at the rear of the vehicle is due to the tracer gas travelling around the sides of the vehicle, and being mixed in the turbulent wake. Figure 4 shows the concentration of SF6 (that is, the pollution from car 1) along the centre-line of car 3 at three tunnel speeds. Since the exhaust from car 1 has to travel further before it reaches car 3, the readings are lower due to the dissipative effects of the turbulent flow around car 2. This can be seen in Fig. 4 where the proportion of tracer gas is plotted for the pressure tappings along the centre-line of car 3. The three curves again show a similar pattern, with a gradual fall in concentration along the length of the vehicle
until the rear screen where the concentrations rise markedly before falling again. Again, the trend is more pronounced at low wind speed. However, the overall range in concentration over the body surface at any one wind tunnel speed is considerably less than in the previous case. Thus it appears that the flow is relatively well-mixed due to the turbulence encountered. Next, we consider the proportion of pollution received from car 1 and car 2 by car 3 at a fixed wind tunnel speed. Figures 5a-c show the concentration of each tracer gas at three wind tunnel speeds. The graphs are very similar, and show that the proportion of tracer gas received by car 3 from car 2 to that of car 1 decreases from about 4: 1 to 3: 1 along the front hood (bonnet). The concentrations converge further along the roof, and actually meet at the lowest speed, but diverge at the rear of the vehicle. Hence, at low speed, and at locations along the bonnet, a large proportion of the pollution received by car 3 comes from car 2.
Local aspects of vehicular pollution
215
Concentration (ppm)
Concentration (ppm)
250 ,
(b)
Concentration (ppm)
1
Fig. 5. Concentration of N20 {car 2) and SF6 (car 1) along the centre-line of car 3 at (a) 4.5 ms-‘, (b) 3 ms-’ and (c) 1.5ms-‘.
The effect of wind tunnel speed is demonstrated in Fig. 6 where the concentrations of the two tracer gases at the pressure tappings corresponding to the air inlet vents of car 3 are plotted over a range of wind tunnel speeds. The results show that the concentrations of
both pollutants fall with wind speed. The proportion of tracer gas received by car 3 from car 2 to that of car 1 is around 2.5: 1 at each of the tappings, and this value is relatively independent of wind tunnel speed.
216
M. J. CLIFFORD
Concentration
et al
(ppm)
300 Left Vent 250
SF, (car 1) -mLeft Vent N,O (car 2) -+Right Vent SF, (car I) -2Right Vent N,O (car 2) +
1
Wind Speed (m/s)
Fig. 6. Concentration of N20 (car 2) and SF, (car 1) at the air vents of car 3 for a range of wind tunnel speeds.
CONCLUSIONS
These results represent a modest but valuable start in understanding the mechanisms involved in the transportation of pollutants around vehicles in slowmoving traffic. It has been shown that around 75% of the total amount of pollution received at the surface of a car comes from the car in front in the wind tunnel. No allowance has been made for crosswinds, but these could be considered in future research. However, a realistic highway study is highly desirable to validate these results in the real world. Acknowledgements-The authors would like to acknowledge the assistance of the Motor Industry Research Association (MIRA) in supplying the geometry for the simplified vehicle shape, the referee for his comments, and the financial support of the Engineering and Physical Sciences Research Council, U.K.
REFERENCES
Brice R. M. and Roesler J. F. (1966) The exposure to carbon monoxide of occupants of vehicles moving in heavy traffic. J. Air Pollut. Control Ass. 6, 591. Chan C. C., ozkaynak H., Spengler J. D. and Sheldon L. (1991a) Driver exposure to volatile organic compounds, CO, ozone, and NO* under different driving conditions. Enoir. Sci. Technol. 25, 964912.
Chan C. C., Spengler J. D., ozkaynak H. and Lefkopoulou M. (1991b) Commuter exposure to VOCs in Boston, Massachusetts. J. Air Waste Man. Ass. 41, 15941600. Chan L. Y., Hung W. T. and Qin Y. (1994) Vehicular emission exposure of bicycle commuters in the urban area of Guangzhou, South China (PRC). Envir. Int. 20, 169-177. Chaney L. W. (1978) Carbon monoxide emissions measured from the interior of a travelling automobile. Science 199, 1203-1204.
Colwill D. M. and Hickman A. J. (1980) Exposure of drivers to carbon monoxide. J. Air Polk. Control Ass. 30. 1316-1319. Flachsbart P. G. and Yo C. Ah. (1987) Microenvironmental models of commuter exposure to carbon monoxide from vehicle exhaust. Final report under contract no. 68-024084, U.S. EPA, Environmental Monitoring Systems Laboratory, Research Triangle Park, North Carolina. Flachsbart P. G., Mack G. A., Howes J. E. and Rodes C. E. (1987) Carbon monoxide exposure to Washington commuters. J. Air Pollut. Control Ass. 37, 135. Hayes S. R. (1989) Estimating the effect of being indoors on total personal exposure to outdoor air pollution. J. Air Pollut. Control Ass. 39, 1453-1461. Ott W. R. (1990) Total human exposure: basic concepts, EPA field studies, and future research needs. J. Air Waste Man. Ass. 40, 966. Ott W. R. and Willits N. W. (1981) CO exposures of occupants of motor vehicles: modelling the dynamic response of the vehicle. SIMS Technical Report No. 48, Department of Statistics, Stanford University, Stanford, California. Ott W. R., Thomas J., Mage D. and Wallace L. (1988) Validation of the simulation of human activity and pollution exposure (SHAPE) model using paired days from the Denver, Colorado, carbon monoxide field study. Atmospheric Environment 22, 2101&2113. Ott W. R., Switzer P. and Willits N. W. (1994) Carbon monoxide exposures inside an automobile travelling on an urban arterial highway. J. Air Waste Man. Ass. 44, 1010-1018. Peterson G. A. and Sabersky R. J. (1975a) Urban exposure to carbon monoxide. Arch .&uir. Hlth. 25, 1028-1032. Peterson G. A. and Saberskv R. J. (1975b) Measurement of pollutants inside an auto&obile. j. Air dollut. Control Ass. 25, 1028-1033. Peterson W. B. and Allen R. (1982) Carbon monoxide exposure to Los Angeles commuters. J. Air Pollut. Control Ass. 32, 826. Stricker R. and Dick A. (1988) Numerical simulation of car aerodynamics and interior climate. In Proc. 1st Con& Supercomputer Applications in the Automotive Industry (edited by Marino C.), Ziirich, Switzerland, October 1986, pp. 99-123. Cray Research.
Atmospheric
PII: S1352-2310(96)00205-l
LOCAL ASPECTS OF VEHICULAR M. J. CLIFFORD,
R. CLARKE
Environmew Vol. 31, No. 2, pp. 211-276, 1997 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 1352-2310/97 $15.00 + 0.00
POLLUTION
and S. B. RIFFAT
Institute of Building Technology, Department of Architecture and Building Technology, University of Nottingham, University Park, Nottingham, NG7 2RD, U.K. (First received 2 April 1995 and injnalform
13 March 1996)
Abstract-We consider the local aspects of vehicular pollution by the use of 1: 10 scale models. The model automobiles ‘were placed in a low-velocity wind tunnel to simulate a queue of slow-moving city traffic and tracer gasses were then injected into the air flow though the tail pipes fitted to two of the models. Measurements show that the exhaust gases are entrained in the wake of the vehicle from which they were emitted, and are dispersed mainly by the passage of the following car. The spatial distribution of tracer gas along and across the models suggests that the level of pollution received by a commuter in a slow-moving heavy traffic may be strongly influenced by the pollution produced from the car immediately in front. Copyright c 1996 Elsevier Science Ltd Key word index: Traffic pollution, wind tunnel testing, air quality modelling.
INTRODUCTION
Air pollution from vehicles is of increasing concern, and the issue is of prime importance in many cities in the U.K. and the rest of the developed world. The health aspects of pollution are not yet fully understood, but the increase in asthma, other respiratory diseases and certain types of cancer are being linked to pollutants emitted from vehicles such as carbon monoxide, oxides of nitrogen, and volatile organic compounds (Chan et al., 1991a, b). Consequently, it is of great importance to measure the individuals’ exposure to these pollutants in the course of their daily activities, both in the workplace, and whilst commuting to and from work (Peterson and Sabersky, 1975a; Ott et al., 1988; Ha:yes, 1989; Ott, 1990). Studies have shown that levels of pollution experienced whilst commuting by road to and from the workplace are much higher than those experienced at work or at home, and an individual receives most of his daily dose of pollution whilst commuting (Flachsbart and Yo, 1987). Hence these levels and the resulting dose of pollution should be closely monitored. Several methods have been developed which detect the concentration of various pollutants at the roadside (Brice and Roesler, 1966; Chan et al., 1994), but the levels of pollutants inside a vehicle are generally greater than these roadside measurements (Chan et al., 1991a, b). There may be several reasons for this. Firstly, the level of pollution decays away from the source, and consequently the levels around the vehicles in traffic will be higher than the levels at the roadside. This makes careful location of monitoring equipment essential. Secondly, the method of ventilation used by indi-
vidual vehicles may serve to concentrate the levels of pollutants inside the vehicle if the air conditioning and the state of windows, sun roof, etc. are not effective, or conversely an efficient ventilation system may reduce the levels (Stricker and Dick, 1988). Conflicting reports exist as to whether it is better to drive with the windows open or closed in heavy traffic (Peterson and Sabersky, 197513;Colwill and Hickman, 1980). The air space inside a vehicle can act as a buffer against sudden high levels of pollution, but this would also slow down the beneficial effect of driving from an area with a high level of pollution into an area with a lower level (Flachsbart and Yo, 1987). Simplified models assuming complete mixing inside the vehicle have been proposed to measure this buffer effect (Ott and Willits, 1981; Ott et al., 1994). An alternative approach is to measure the pollutant concentration inside a vehicle whilst driving in traffic. Several techniques are available, either using a mobile gas sensor, or collecting samples of vehicular air for subsequent analysis. The mobile gas detector method has the advantage of not only giving the total pollution dose, but also the pollutant concentrations at various landmarks which can be compared with climate, traffic density, queuing time at junctions, and so on. However, it may be necessary to collect gas samples or to use some other total exposure method to detect volatile organic compounds. Several studies have been carried out along these lines (Peterson and Allen, 1982; Flachsbart et al., 1987; Chan et al., 1991a, b; Ott et al., 1994), and the results may be useful in estimating the average exposure of a typical commuter. However, a more local aspect of the problem was shown by Colwill and Hickman (1980) in an 271
212
M. J. CLIFFORD et al.
shapes. Wind tunnel testing has been used to examine the theory that the pollution to which a driver is exposed may be a local phenomenon, and in the future this will be followed up with experimental observations in traffic and computational fluid dynamics, which is the subject of current research.
early study which considered the levels of pollution in several types of vehicles driven along the same route. The levels varied considerably from vehicle to vehicle. Clearly a more fundamental approach is needed if general statements are to be made about the exposure of commuters to pollutants. An important aspect ignored by the studies already mentioned is that most of the pollution emitted from a vehicle comes from a specific location: the exhaust. This may lead to significant local effects in the concentration of pollutants around the traffic. Given that the levels of pollution are highest in slowmoving traffic, on calm days, especially when approaching or accelerating away from road junctions, it may be beneficial to consider the dispersion of the pollution produced at the exhaust by the wind, and the air flow around the adjacent vehicles in a more controlled setting. The hypothesis here is that the levels of pollution in a vehicle may be strongly influenced by the levels of pollution produced from the exhaust of the vehicle immediately in front. Whilst this theory falls down at high speed with strong cross winds, at low speed, and on calm days, i.e. the worst case, it seems more plausible. The experimental observations of Chaney (1978) bear this out. He drove at a constant speed of 52 m.p.h. along a highway as various vehicles overtook his vehicle and pulled into his lane about 50 ft ahead. At the differential speed from 3 to 10 m.p.h., he noticed significant peaks in the level of carbon monoxide that were directly attributable to the vehicle in front. Clearly, driving behind such a vehicle for a significant time in slow-moving traffic would have an effect on the level of pollution inside a vehicle. Such observations have prompted the current work, which is concerned with modelling the dispersion of pollutants around three simplified vehicle
-
+
+
+
+
+
WIND TUNNEL TESTINGLOW-VELOCITY ATMOSPHERIC WIND TUNNEL
The tunnel used for this series of tests was based on a small open-jet wind tunnel developed for teaching purposes. The tunnel has a maximum flow rate of 4.5 ms-’ and a working section of width l.Om, height 0.75 m and length 2.25 m. Two layers of honeycomb are fitted in the bellmouth to straighten the incoming flow and are followed by a 0.5 mm mesh screen. Beyond the entry section there is a 0.75 m settling chamber to allow disturbances caused by the honeycomb and screen to decay. A transformer is positioned between the working section and the fan to convert the l.Om x 0.75 m cross section to a suitable shape for the fan mounting. The entry to the transformer is larger than the settling chamber exit to allow for some expansion of the flow through the working section. The fan speed is controlled by a variable autotransformer which varies the voltage supplied to the drive motor. The tunnel wind speed was measured to determine any velocity variations across the tunnel section. Two sets of data were recorded, the first measured 0.5m downstream of the settling chamber, and the second a further 1 m downstream. Only negligible velocity variations were found ( + 3%), although the turbulence intensity was higher at the second location. The flow through the tunnel was found to be stable up to 0.2m from the edge of the table.
+
+
--
+ I \
/
;
Fig. 1. MIRA vehicle with pressure tappings.
Local aspects of vehicular pollution SCALE MODEL CONSTRUCTION
Three hollow model cars based on the Motor Industry Research Association (MIRA) model were constructed from 1Omm medium density fibreboard (MDF) at a scale of 1: 10 (Fig. 1). The cars were approximately 40 cm long. The models were deliberately simple, and no attempt was made to model fine detail. Cars 1 and 2 were fitted with a single injection tube to simulate the engine exhaust which was positioned on the right-hand side at the rear of the two vehicles. This had 1 mm internal diameter (ID) to allow simulation of the measured flow of exhaust from a car whilst the engine was idling (15-20 m s-l). SF, tracer gas was injected at the exhaust of car 1, and N,O tracer gas was injected at the exhaust of car 2 at a rate of 1 emin-’ and 250,000 parts per million (ppm) to simulate the carbon monoxide concentration in a typical exhaust. Car 3 was fitted with several static pressure tappings (holes) to allow measurement of gas concentration and pressure around the car body surface. These tappings were machined from 0.8 ID brass tubing, and fitted on all of the upper body surfaces. The modals were painted and no attempt was made to measure surface roughness. Plastic tubing was attached to the tappings through a hole in the wind tunnel table to allow the tappings to be connected to the instrumentation.
INSTRUMENTATION
Twenty of the model pressure tappings were connected to a selection box. Gas analysis was carried out using a Binos 1000 gas analyzer. The instrumentation
273
was connected to a Datron 1055 data logging system controlled by a portable PC (Fig. 2).
MODEL ARRANGEMENT
The three model cars were placed in line on the tunnel floor 20 cm apart to simulate a full scale gap of 2 m (approximately half a car length). The car with the pressure tappings (car 3) was placed at the end of the line. Both the exhausts were switched on, and readings of SF6 (car 1) and NzO (car 2) concentration were taken from all of the pressure tappings on car 3 with the wind speed set at 1.5 m s-l. Since the flow around the vehicles was turbulent, it was necessary to average the results over 20s on each tapping. When the test had been completed, the wind speed was altered. The test procedure was then repeated over a range of wind tunnel speeds. Hence data on the proportion of pollution received from cars 1 and 2 by car 3 was obtained for a range of operating conditions.
RESULTS
The data from each of the tests at a constant wind tunnel speed shows that there is considerable variation in the tracer gas concentration over the surface of car 3 from car 2. This can be seen in Fig. 3 where the concentration of NzO (that is, the pollution from car 2) along the centre-line of car 3 is plotted for three wind tunnel speeds. A sketch of car 3 is added to aid visualization. The three curves show the same general trend in that the pollution concentration falls sharply
Tunnel Inlet L
““M:
i
(0.75lhin)
Fig. 2. Wind tunnel setup: NzO supply to car 2 is not shown.
M. J. CLIFFORD et al.
274
Concentration (ppm) 1.5mls
N,O (car 3mi.s N,O (car 4.5mIs N,O (car f
Fig. 3. Concentration of NzO (car 2) along the centre-line of car 3.
Concentration (ppm) 120
80 60
\/
J
Fig. 4. Concentration of SF6 (car 1) along the centre-line of car 3.
across the front hood (bonnet) of car 3, and continues to fall more gradually along the roof, but recovers slightly towards the rear of the vehicle. This trend is more pronounced at low wind speed. The reason for this pattern can be deduced by considering the flow pattern around the vehicle. As expected, the concentration falls along the length of the vehicle as the distance from the source increases. The rise at the rear of the vehicle is due to the tracer gas travelling around the sides of the vehicle, and being mixed in the turbulent wake. Figure 4 shows the concentration of SF6 (that is, the pollution from car 1) along the centre-line of car 3 at three tunnel speeds. Since the exhaust from car 1 has to travel further before it reaches car 3, the readings are lower due to the dissipative effects of the turbulent flow around car 2. This can be seen in Fig. 4 where the proportion of tracer gas is plotted for the pressure tappings along the centre-line of car 3. The three curves again show a similar pattern, with a gradual fall in concentration along the length of the vehicle
until the rear screen where the concentrations rise markedly before falling again. Again, the trend is more pronounced at low wind speed. However, the overall range in concentration over the body surface at any one wind tunnel speed is considerably less than in the previous case. Thus it appears that the flow is relatively well-mixed due to the turbulence encountered. Next, we consider the proportion of pollution received from car 1 and car 2 by car 3 at a fixed wind tunnel speed. Figures 5a-c show the concentration of each tracer gas at three wind tunnel speeds. The graphs are very similar, and show that the proportion of tracer gas received by car 3 from car 2 to that of car 1 decreases from about 4: 1 to 3: 1 along the front hood (bonnet). The concentrations converge further along the roof, and actually meet at the lowest speed, but diverge at the rear of the vehicle. Hence, at low speed, and at locations along the bonnet, a large proportion of the pollution received by car 3 comes from car 2.
Local aspects of vehicular pollution
215
Concentration (ppm)
Concentration (ppm)
250 ,
(b)
Concentration (ppm)
1
Fig. 5. Concentration of N20 {car 2) and SF6 (car 1) along the centre-line of car 3 at (a) 4.5 ms-‘, (b) 3 ms-’ and (c) 1.5ms-‘.
The effect of wind tunnel speed is demonstrated in Fig. 6 where the concentrations of the two tracer gases at the pressure tappings corresponding to the air inlet vents of car 3 are plotted over a range of wind tunnel speeds. The results show that the concentrations of
both pollutants fall with wind speed. The proportion of tracer gas received by car 3 from car 2 to that of car 1 is around 2.5: 1 at each of the tappings, and this value is relatively independent of wind tunnel speed.
216
M. J. CLIFFORD
Concentration
et al
(ppm)
300 Left Vent 250
SF, (car 1) -mLeft Vent N,O (car 2) -+Right Vent SF, (car I) -2Right Vent N,O (car 2) +
1
Wind Speed (m/s)
Fig. 6. Concentration of N20 (car 2) and SF, (car 1) at the air vents of car 3 for a range of wind tunnel speeds.
CONCLUSIONS
These results represent a modest but valuable start in understanding the mechanisms involved in the transportation of pollutants around vehicles in slowmoving traffic. It has been shown that around 75% of the total amount of pollution received at the surface of a car comes from the car in front in the wind tunnel. No allowance has been made for crosswinds, but these could be considered in future research. However, a realistic highway study is highly desirable to validate these results in the real world. Acknowledgements-The authors would like to acknowledge the assistance of the Motor Industry Research Association (MIRA) in supplying the geometry for the simplified vehicle shape, the referee for his comments, and the financial support of the Engineering and Physical Sciences Research Council, U.K.
REFERENCES
Brice R. M. and Roesler J. F. (1966) The exposure to carbon monoxide of occupants of vehicles moving in heavy traffic. J. Air Pollut. Control Ass. 6, 591. Chan C. C., ozkaynak H., Spengler J. D. and Sheldon L. (1991a) Driver exposure to volatile organic compounds, CO, ozone, and NO* under different driving conditions. Enoir. Sci. Technol. 25, 964912.
Chan C. C., Spengler J. D., ozkaynak H. and Lefkopoulou M. (1991b) Commuter exposure to VOCs in Boston, Massachusetts. J. Air Waste Man. Ass. 41, 15941600. Chan L. Y., Hung W. T. and Qin Y. (1994) Vehicular emission exposure of bicycle commuters in the urban area of Guangzhou, South China (PRC). Envir. Int. 20, 169-177. Chaney L. W. (1978) Carbon monoxide emissions measured from the interior of a travelling automobile. Science 199, 1203-1204.
Colwill D. M. and Hickman A. J. (1980) Exposure of drivers to carbon monoxide. J. Air Polk. Control Ass. 30. 1316-1319. Flachsbart P. G. and Yo C. Ah. (1987) Microenvironmental models of commuter exposure to carbon monoxide from vehicle exhaust. Final report under contract no. 68-024084, U.S. EPA, Environmental Monitoring Systems Laboratory, Research Triangle Park, North Carolina. Flachsbart P. G., Mack G. A., Howes J. E. and Rodes C. E. (1987) Carbon monoxide exposure to Washington commuters. J. Air Pollut. Control Ass. 37, 135. Hayes S. R. (1989) Estimating the effect of being indoors on total personal exposure to outdoor air pollution. J. Air Pollut. Control Ass. 39, 1453-1461. Ott W. R. (1990) Total human exposure: basic concepts, EPA field studies, and future research needs. J. Air Waste Man. Ass. 40, 966. Ott W. R. and Willits N. W. (1981) CO exposures of occupants of motor vehicles: modelling the dynamic response of the vehicle. SIMS Technical Report No. 48, Department of Statistics, Stanford University, Stanford, California. Ott W. R., Thomas J., Mage D. and Wallace L. (1988) Validation of the simulation of human activity and pollution exposure (SHAPE) model using paired days from the Denver, Colorado, carbon monoxide field study. Atmospheric Environment 22, 2101&2113. Ott W. R., Switzer P. and Willits N. W. (1994) Carbon monoxide exposures inside an automobile travelling on an urban arterial highway. J. Air Waste Man. Ass. 44, 1010-1018. Peterson G. A. and Sabersky R. J. (1975a) Urban exposure to carbon monoxide. Arch .&uir. Hlth. 25, 1028-1032. Peterson G. A. and Saberskv R. J. (1975b) Measurement of pollutants inside an auto&obile. j. Air dollut. Control Ass. 25, 1028-1033. Peterson W. B. and Allen R. (1982) Carbon monoxide exposure to Los Angeles commuters. J. Air Pollut. Control Ass. 32, 826. Stricker R. and Dick A. (1988) Numerical simulation of car aerodynamics and interior climate. In Proc. 1st Con& Supercomputer Applications in the Automotive Industry (edited by Marino C.), Ziirich, Switzerland, October 1986, pp. 99-123. Cray Research.