Estimating Emissions from Sources of Air Pollution
6.2 Estimating Emissions from On-Road Mobile Sources
6.2.6 Determination of the Distribution of Vehicle Technologies in a Fleet
6.2.6.1 Introduction
The distribution of vehicle technologies in a fleet is one of the most important parameters that must be established in order to estimate the emissions from the on-road vehicle fleet. The collection of this data can be a challenge in many locations, but a number of approaches have been used to allow the accumulation of reasonably accurate information for input into vehicle emission models. It should be understood that when the “distribution of vehicle technologies” is referred to that the total amount or fraction of total driving by each technology is what is needed. This is different from the distribution of vehicles in the static fleet. For example, if the fleet of interest contains 50,000 carbureted vehicles and 50,000 fuel injected vehicles it might be concluded that 50% of the fleet driving was carbureted vehicles and 50% are fuel injected vehicles. While this is a true statistic for the static fleet, the carbureted vehicles will likely be older than the fuel injected vehicles. Older vehicles are normally driven less than newer vehicles. Thus, the fuel injected vehicles might actually contribute 65% of the on-road driving while the carbureted vehicles contribute 35% of the driving due to the difference in driving per vehicle. It is the amount or fraction of on-road driving that would be of interest for estimating vehicle emissions. Figure 6.2.6-1 illustrates vehicle use by age for several cities around the world.
If older vehicles were driven the same as newer vehicles, the curves in figure 6.2.6-1 would be straight lines. It is clear that this is not the case and that the change in mileage between the two years 0 and 2 is significantly different than for the two years 10 to 12.
6.2.6.2 Use of Model Year Distributions for Estimation of Vehicle Technology Distributions
The model year of a vehicle is a typical parameter that might be associated with vehicle technology. Thus, the emissions modeler might conclude that if the distribution of model years is known, then the distribution of technologies would be known. This is wrong for two reasons. First, it is seldom correct that a single model year contains only one type of technology. Depending upon the model year and the location of interest, a given model year will likely contain a variety of technologies. Thus, in order to use the model year to characterize the fleet technology distribution, the distribution of technologies within the model year must be established. Second, as illustrated in figure 6.2.6-1, even if the distribution of technologies by model year is known, different technologies might be driven differently. This fact must be accounted for to derive the fraction of driving that each technology contributes to the on-road totals within a model year.
If use of model year is the only option for establishing the on-road distribution of driving, then some attempt must be made to establish the technology distribution for each model year by fraction of driving. In the United States, carbureted on-road gasoline vehicles ceased to be produced in the late 1990s, however, in Kazakhstan, new carbureted vehicles are still being sold in some cases. In most countries, carbureted vehicles are not being produced as of 2005.
Equation 6.2.6-2 can be used to approximate vehicle use by vehicle age if the average annual vehicle use is known. This equation is based on studies carried out by ISSRC in 11 countries.
Where DA represents the actual annual driving for vehicles of age A, Davg represents the overall fleet average annual driving per vehicle, and A = the age in years of the vehicle group of interest.
Figure 6.2.6-3 indicates the overall fleet average annual driving measured in the 2001-2006 timeframe in several cities around the world that might be considered a pointer for estimating Davg for a location.
6.2.6.3 Use of Vehicle Registration Data or Inspection Maintenance Data
A common way to develop information on the technology distribution of vehicles is to refer to local vehicle registration data. Vehicle data is normally kept at the national or provincial (state) level of government in most countries. This approach, while better than no data at all, can be fraught with errors. The problems are two-fold. First, in most developing countries, vehicles are registered once, at the time of purchase. If the vehicle is damaged and removed from the fleet or sold to a different location, there may be no update to the registration database. Second, vehicles are not operated the same amount each year. This was illustrated earlier in 6.2.6-1. The reduced driving for older vehicles means that the driving distribution derived from the registration database can be very inaccurate. Figure 6.2.6-4 shows the fraction of driving in Los Angeles that would be developed based on numbers derived directly from motor vehicle registration data (Division of Motor Vehicle data) compared to data collected by video taping roadways then using license plate data to look up technologies (On-Road Vehicle Number), and data from surveying parking lots (Parking Lot Vehicle Number). This graphic is used with permission from CECERT at the University of California at Riverside.
As can be seen in figure 6.2.6-4, the on-road video measurements (On-Road Vehicle Number), which should be the most accurate, agree well with the parking lot surveys (Parking Lot Vehicle Number) but vary considerably from direct use of the registration data (DMV). It should be noted, however, that if equation 6.2.6-2 is used in conjunction with the registration data in figure 6.2.6-4 that a distribution much closer to the actual measured on-road distribution is derived. For example, considering vehicles that are 10 years old, the registration database shown in figure 6.2.6-4 (DMV) indicates that by number, these vehicles are 58% of the Los Angeles fleet. However, equation 6.2.6-2 indicates that a vehicle of zero years is operated 1.341 times the fleet-wide average driving while a vehicle aged 10 years will operate at 0.790 of the fleet-wide average driving. Thus, the 1998 DMV data should be increased from 8% to 1.341*8% or 10.7%. The 1998 data should be reduced from 4.7% to 0.79*4.7% or 3.7%. The results are closer to the actual on-road measured values in figure 6.2.6-4, although not perfect. This process could be applied to all years of a registration database to derive an approximate on-road driving distribution to use in emissions estimation.
Inspection and Maintenance records might be used to estimate the distribution of technologies in the fleet for locations that have such a program. Since vehicles are often required to be inspected annually or semi-annually, this data may be more up to date than the general statewide or countrywide registration data. One has to assume that people are getting their vehicles tested as they are supposed to in order to use this source of data. Also, vehicles will show up in the fractions that they exist in the static fleet. Equation 6.2.6-5 (seen in the previous section) on something like this will have to be used to adjust the data to the correct on-road driving statistics that are needed.
6.2.6.4 Roadway Observations
A more accurate way to determine the on-road distribution of technologies is by actually observing roadways in the area of interest and to record the distribution of vehicles that operate on the roads in this area. In this case, direct measurements of the driving distributions can be made; although on-road observations provide only general technology information because it is impossible to observe the engine configurations in vehicles as they pass by. In Los Angeles, where the license plate data is up to date with annual registration requirements and available in an online database, license plate numbers can be used and very specific information gathered about the vehicles. In most cases, only the general distribution of passenger cars compared to motorcycles, trucks, taxis, and buses can be determined using roadway observations.
Since it is typically not possible to observe all roadways in the area of interest, a selection of roadways must be made to carry out the needed observations. The selected roadways must be representative of the overall area of interest. Thus, some of the roadways should be highways while some should be arterials and residential streets. It is also important to sample in the different regions of a city such as a lower income area, a middle-income area, a high-income area, and a commercial area. The results for the data collection for these different road types can then be averaged or emission factors can be developed for each part of a city. In studies carried out by ISSRC, it is typical to select an upper income section of a city, a lower income section of a city, and a central (normally commercial) area of a city. A highway, arterial, and residential type street are selected in each city section. Data is collected hour by hour from early morning until late evening. Such a study can be carried out over a nine-day study period if only one sample of each hour is taken. This data is then analyzed to establish the general fleet on-road technology distribution. When studies were carried out in Mexico City, due to the size of the city, six sections of the city were studied and studies were carried out on more than eighteen days for all of the hours to improve the results of the study. This approach required several weeks of effort.
The described studies can be done by posting several persons at a location and have them each record the number of certain types of vehicles that pass by. For example, one person could be assigned to passenger vehicles, another person to trucks, and so on. The numbers can then be combined to produce the needed statistics. Modern digital video cameras have been found to be optimum for this type of data collection. The advantage of a video tape is that it can be stopped and the image frozen in order to capture the numbers of different types of vehicles especially in very busy traffic situations where it is hard for an observer to keep up with all of the vehicles passing by. Figure 6.2.6-6 shows a video camera set up for observing a roadway in Sao Paulo, Brazil and students reading tapes collected in Beijing, China.
It is important to reiterate that on-road observations can only provide information on the general distribution of vehicles on the roadways such as the fraction of trucks compared to buses, compared to passenger vehicles and so on. The types of data gleaned from video surveys are shown in table 6.2.6-7 for ten studies around the world carried out by ISSRC.
*There are a large number of unofficial taxis in many of these cities. These numbers reflect official taxis that are easily identified.
In addition to the fractions of vehicles shown in table 6.2.6-7, the video surveys also yield the number of vehicles on a roadway by hour of the day when the videos are taken. This data can be very helpful in estimating overall vehicle miles traveled as will be discussed in another section.
As noted earlier, it is difficult to get more differentiation of the vehicle fleet from videotape information other than that shown in table 6.2.6-7. A test was made to see how well videotape reviewers could differentiate between different sizes of passenger cars. The results were not very good. In order to differentiate between different size trucks and buses, the study directors provide pictures of a variety of small, medium, and large buses and trucks to educate the videotape readers as to how to distribute the observed trucks and buses into the different size classes. The results were better in this case since there is a little clearer differentiation between different sized buses and trucks.
Another technique must be used to establish the exact technologies of the passenger vehicle fleet, the truck fleet, the bus fleet, and the other components of the on-road fleet. In order to establish the more detailed information about the fleet of interest, parking lot surveys and business surveys must be carried out in association with the on-road vehicle observations. These types of surveys, when combined with the video data, can provide excellent data on the specific on-road distribution of vehicle technologies.
6.2.6.5 Parking Lot and Business Surveys to Establish Specific Vehicle Technology Distributions
The most straightforward way of determining the exact technology distribution of passenger vehicles is to carry out a vehicle-by-vehicle survey. This can be done in one of two ways. Vehicle operators can be surveyed at service stations or parking lot entrances or surveys can be carried out by studying parked vehicles on the streets and in parking lots. Either approach correlates well with the on-road fleet distribution as illustrated in 6.2.6-4. Each approach has advantages and disadvantages. The survey of vehicle drivers allows an accurate determination of vehicle model year since most owners know this about their vehicle and the hood of the vehicle can be opened to check out engine parameters. The odometer reading of the vehicle can be established and sometimes drivers will carry paperwork about their vehicles. The negative side to this type of survey is the time that it takes and sometimes vehicle owners are not willing to be interviewed or have their vehicle inspected. The alternate approach is to simply go into parking lots or along streets and record data about the unoccupied vehicles. This approach is much faster than the interview process but requires an expert vehicle mechanic who is able to determine model year and engine configuration by looking at the vehicle from the outside and under the vehicle. Odometer readings must be determined by looking through the vehicle windows to read the odometer, which does not work with modern digital odometers that turn off when the vehicle is not running. Owners returning to their vehicles and finding them being inspected can be concerned about the purpose of the inspection, but proper identification and documentation usually alleviates this concern. Figure 6.2.6-8 shows parked vehicle inspections in Nairobi, Kenya and Beijing, China.
To establish the specific technology distributions for trucks and buses, businesses that own truck or bus fleets must be surveyed. It has also been found possible in the case of trucks to carryout a survey at government roadside truck stops for weight, safety, or tax purposes.
As the case for any survey, the places selected for survey must, overall, be representative of the fleet of interest. In the case of parking lot surveys, locations such as grocery stores, which service a broad range of clients from different economic strata, should be sought out. Another option would be to carry out surveys in parking lots in different parts of an area to study all economic strata of the region. It has also been found important for passenger car studies to survey at least 1,000 vehicles. Most ISSRC studies are designed to collect data on 1,600 vehicles. They involve a two-person team that includes an expert in vehicle identification, usually an experienced mechanic, and a person, often a student, to record data. Two teams can typically survey 220 vehicles per day in parking lots and along the street. If it is decided to interview drivers as they fill their gasoline tanks or as they enter parking lots, the process will take about two to three times as long depending upon the rate of vehicles that show up at the point of interview.
Diesel vehicles do not require as large of a survey sample. This is because the variety of possible engine technologies is much smaller than for passenger vehicles. A random sampling of 100-200 buses and 100-200 trucks can be adequate to get a good fix on the distribution of technologies in the fleet with the present range of engine technologies. The range of bus and truck technologies will likely increase in the future as particulate filters and NOx controls are used to control trucks and buses. This change will require larger surveys to insure that accurate results will be obtained.
The distribution of vehicle technologies in a fleet is one of the most important parameters that must be established in order to estimate the emissions from the on-road vehicle fleet. The collection of this data can be a challenge in many locations, but a number of approaches have been used to allow the accumulation of reasonably accurate information for input into vehicle emission models. It should be understood that when the “distribution of vehicle technologies” is referred to that the total amount or fraction of total driving by each technology is what is needed. This is different from the distribution of vehicles in the static fleet. For example, if the fleet of interest contains 50,000 carbureted vehicles and 50,000 fuel injected vehicles it might be concluded that 50% of the fleet driving was carbureted vehicles and 50% are fuel injected vehicles. While this is a true statistic for the static fleet, the carbureted vehicles will likely be older than the fuel injected vehicles. Older vehicles are normally driven less than newer vehicles. Thus, the fuel injected vehicles might actually contribute 65% of the on-road driving while the carbureted vehicles contribute 35% of the driving due to the difference in driving per vehicle. It is the amount or fraction of on-road driving that would be of interest for estimating vehicle emissions. Figure 6.2.6-1 illustrates vehicle use by age for several cities around the world.
If older vehicles were driven the same as newer vehicles, the curves in figure 6.2.6-1 would be straight lines. It is clear that this is not the case and that the change in mileage between the two years 0 and 2 is significantly different than for the two years 10 to 12.
6.2.6.2 Use of Model Year Distributions for Estimation of Vehicle Technology Distributions
The model year of a vehicle is a typical parameter that might be associated with vehicle technology. Thus, the emissions modeler might conclude that if the distribution of model years is known, then the distribution of technologies would be known. This is wrong for two reasons. First, it is seldom correct that a single model year contains only one type of technology. Depending upon the model year and the location of interest, a given model year will likely contain a variety of technologies. Thus, in order to use the model year to characterize the fleet technology distribution, the distribution of technologies within the model year must be established. Second, as illustrated in figure 6.2.6-1, even if the distribution of technologies by model year is known, different technologies might be driven differently. This fact must be accounted for to derive the fraction of driving that each technology contributes to the on-road totals within a model year.
If use of model year is the only option for establishing the on-road distribution of driving, then some attempt must be made to establish the technology distribution for each model year by fraction of driving. In the United States, carbureted on-road gasoline vehicles ceased to be produced in the late 1990s, however, in Kazakhstan, new carbureted vehicles are still being sold in some cases. In most countries, carbureted vehicles are not being produced as of 2005.
Equation 6.2.6-2 can be used to approximate vehicle use by vehicle age if the average annual vehicle use is known. This equation is based on studies carried out by ISSRC in 11 countries.
Where DA represents the actual annual driving for vehicles of age A, Davg represents the overall fleet average annual driving per vehicle, and A = the age in years of the vehicle group of interest.
Figure 6.2.6-3 indicates the overall fleet average annual driving measured in the 2001-2006 timeframe in several cities around the world that might be considered a pointer for estimating Davg for a location.
6.2.6.3 Use of Vehicle Registration Data or Inspection Maintenance Data
A common way to develop information on the technology distribution of vehicles is to refer to local vehicle registration data. Vehicle data is normally kept at the national or provincial (state) level of government in most countries. This approach, while better than no data at all, can be fraught with errors. The problems are two-fold. First, in most developing countries, vehicles are registered once, at the time of purchase. If the vehicle is damaged and removed from the fleet or sold to a different location, there may be no update to the registration database. Second, vehicles are not operated the same amount each year. This was illustrated earlier in 6.2.6-1. The reduced driving for older vehicles means that the driving distribution derived from the registration database can be very inaccurate. Figure 6.2.6-4 shows the fraction of driving in Los Angeles that would be developed based on numbers derived directly from motor vehicle registration data (Division of Motor Vehicle data) compared to data collected by video taping roadways then using license plate data to look up technologies (On-Road Vehicle Number), and data from surveying parking lots (Parking Lot Vehicle Number). This graphic is used with permission from CECERT at the University of California at Riverside.
6.2.6-4 Distribution of Vehicle Technologies Estimated Using Different Approaches
As can be seen in figure 6.2.6-4, the on-road video measurements (On-Road Vehicle Number), which should be the most accurate, agree well with the parking lot surveys (Parking Lot Vehicle Number) but vary considerably from direct use of the registration data (DMV). It should be noted, however, that if equation 6.2.6-2 is used in conjunction with the registration data in figure 6.2.6-4 that a distribution much closer to the actual measured on-road distribution is derived. For example, considering vehicles that are 10 years old, the registration database shown in figure 6.2.6-4 (DMV) indicates that by number, these vehicles are 58% of the Los Angeles fleet. However, equation 6.2.6-2 indicates that a vehicle of zero years is operated 1.341 times the fleet-wide average driving while a vehicle aged 10 years will operate at 0.790 of the fleet-wide average driving. Thus, the 1998 DMV data should be increased from 8% to 1.341*8% or 10.7%. The 1998 data should be reduced from 4.7% to 0.79*4.7% or 3.7%. The results are closer to the actual on-road measured values in figure 6.2.6-4, although not perfect. This process could be applied to all years of a registration database to derive an approximate on-road driving distribution to use in emissions estimation.
Inspection and Maintenance records might be used to estimate the distribution of technologies in the fleet for locations that have such a program. Since vehicles are often required to be inspected annually or semi-annually, this data may be more up to date than the general statewide or countrywide registration data. One has to assume that people are getting their vehicles tested as they are supposed to in order to use this source of data. Also, vehicles will show up in the fractions that they exist in the static fleet. Equation 6.2.6-5 (seen in the previous section) on something like this will have to be used to adjust the data to the correct on-road driving statistics that are needed.
6.2.6.4 Roadway Observations
A more accurate way to determine the on-road distribution of technologies is by actually observing roadways in the area of interest and to record the distribution of vehicles that operate on the roads in this area. In this case, direct measurements of the driving distributions can be made; although on-road observations provide only general technology information because it is impossible to observe the engine configurations in vehicles as they pass by. In Los Angeles, where the license plate data is up to date with annual registration requirements and available in an online database, license plate numbers can be used and very specific information gathered about the vehicles. In most cases, only the general distribution of passenger cars compared to motorcycles, trucks, taxis, and buses can be determined using roadway observations.
Since it is typically not possible to observe all roadways in the area of interest, a selection of roadways must be made to carry out the needed observations. The selected roadways must be representative of the overall area of interest. Thus, some of the roadways should be highways while some should be arterials and residential streets. It is also important to sample in the different regions of a city such as a lower income area, a middle-income area, a high-income area, and a commercial area. The results for the data collection for these different road types can then be averaged or emission factors can be developed for each part of a city. In studies carried out by ISSRC, it is typical to select an upper income section of a city, a lower income section of a city, and a central (normally commercial) area of a city. A highway, arterial, and residential type street are selected in each city section. Data is collected hour by hour from early morning until late evening. Such a study can be carried out over a nine-day study period if only one sample of each hour is taken. This data is then analyzed to establish the general fleet on-road technology distribution. When studies were carried out in Mexico City, due to the size of the city, six sections of the city were studied and studies were carried out on more than eighteen days for all of the hours to improve the results of the study. This approach required several weeks of effort.
The described studies can be done by posting several persons at a location and have them each record the number of certain types of vehicles that pass by. For example, one person could be assigned to passenger vehicles, another person to trucks, and so on. The numbers can then be combined to produce the needed statistics. Modern digital video cameras have been found to be optimum for this type of data collection. The advantage of a video tape is that it can be stopped and the image frozen in order to capture the numbers of different types of vehicles especially in very busy traffic situations where it is hard for an observer to keep up with all of the vehicles passing by. Figure 6.2.6-6 shows a video camera set up for observing a roadway in Sao Paulo, Brazil and students reading tapes collected in Beijing, China.
It is important to reiterate that on-road observations can only provide information on the general distribution of vehicles on the roadways such as the fraction of trucks compared to buses, compared to passenger vehicles and so on. The types of data gleaned from video surveys are shown in table 6.2.6-7 for ten studies around the world carried out by ISSRC.
*There are a large number of unofficial taxis in many of these cities. These numbers reflect official taxis that are easily identified.
In addition to the fractions of vehicles shown in table 6.2.6-7, the video surveys also yield the number of vehicles on a roadway by hour of the day when the videos are taken. This data can be very helpful in estimating overall vehicle miles traveled as will be discussed in another section.
As noted earlier, it is difficult to get more differentiation of the vehicle fleet from videotape information other than that shown in table 6.2.6-7. A test was made to see how well videotape reviewers could differentiate between different sizes of passenger cars. The results were not very good. In order to differentiate between different size trucks and buses, the study directors provide pictures of a variety of small, medium, and large buses and trucks to educate the videotape readers as to how to distribute the observed trucks and buses into the different size classes. The results were better in this case since there is a little clearer differentiation between different sized buses and trucks.
Another technique must be used to establish the exact technologies of the passenger vehicle fleet, the truck fleet, the bus fleet, and the other components of the on-road fleet. In order to establish the more detailed information about the fleet of interest, parking lot surveys and business surveys must be carried out in association with the on-road vehicle observations. These types of surveys, when combined with the video data, can provide excellent data on the specific on-road distribution of vehicle technologies.
6.2.6.5 Parking Lot and Business Surveys to Establish Specific Vehicle Technology Distributions
The most straightforward way of determining the exact technology distribution of passenger vehicles is to carry out a vehicle-by-vehicle survey. This can be done in one of two ways. Vehicle operators can be surveyed at service stations or parking lot entrances or surveys can be carried out by studying parked vehicles on the streets and in parking lots. Either approach correlates well with the on-road fleet distribution as illustrated in 6.2.6-4. Each approach has advantages and disadvantages. The survey of vehicle drivers allows an accurate determination of vehicle model year since most owners know this about their vehicle and the hood of the vehicle can be opened to check out engine parameters. The odometer reading of the vehicle can be established and sometimes drivers will carry paperwork about their vehicles. The negative side to this type of survey is the time that it takes and sometimes vehicle owners are not willing to be interviewed or have their vehicle inspected. The alternate approach is to simply go into parking lots or along streets and record data about the unoccupied vehicles. This approach is much faster than the interview process but requires an expert vehicle mechanic who is able to determine model year and engine configuration by looking at the vehicle from the outside and under the vehicle. Odometer readings must be determined by looking through the vehicle windows to read the odometer, which does not work with modern digital odometers that turn off when the vehicle is not running. Owners returning to their vehicles and finding them being inspected can be concerned about the purpose of the inspection, but proper identification and documentation usually alleviates this concern. Figure 6.2.6-8 shows parked vehicle inspections in Nairobi, Kenya and Beijing, China.
To establish the specific technology distributions for trucks and buses, businesses that own truck or bus fleets must be surveyed. It has also been found possible in the case of trucks to carryout a survey at government roadside truck stops for weight, safety, or tax purposes.
As the case for any survey, the places selected for survey must, overall, be representative of the fleet of interest. In the case of parking lot surveys, locations such as grocery stores, which service a broad range of clients from different economic strata, should be sought out. Another option would be to carry out surveys in parking lots in different parts of an area to study all economic strata of the region. It has also been found important for passenger car studies to survey at least 1,000 vehicles. Most ISSRC studies are designed to collect data on 1,600 vehicles. They involve a two-person team that includes an expert in vehicle identification, usually an experienced mechanic, and a person, often a student, to record data. Two teams can typically survey 220 vehicles per day in parking lots and along the street. If it is decided to interview drivers as they fill their gasoline tanks or as they enter parking lots, the process will take about two to three times as long depending upon the rate of vehicles that show up at the point of interview.
Diesel vehicles do not require as large of a survey sample. This is because the variety of possible engine technologies is much smaller than for passenger vehicles. A random sampling of 100-200 buses and 100-200 trucks can be adequate to get a good fix on the distribution of technologies in the fleet with the present range of engine technologies. The range of bus and truck technologies will likely increase in the future as particulate filters and NOx controls are used to control trucks and buses. This change will require larger surveys to insure that accurate results will be obtained.
