Clean Techn Environ Policy (2005) 7: 279–284 DOI 10.1007/s10098-005-0279-x
O R I GI N A L P A P E R
Saad Abo-Qudais Æ Hani Abu Qdais
Performance evaluation of vehicles emissions prediction models
Received: 9 November 2004 / Accepted: 1 February 2005 / Published online: 20 July 2005 Ó Springer-Verlag 2005
Abstract Road traffic is a dominant source of urban air pollution. Therefore, it is necessary to quantify emission levels as accurately as possible to evaluate their impacts on the public health and the environment. Several models were developed to predict these emissions. These models can be grouped into three categories, namely, emission factors models, average speed models, and modal models. The prediction capability of most developed models is relatively poor. Therefore, there is a pressing need to improve the predictability of the existing models or to develop new ones with better accuracy. The main focus of this paper is to review different traffic emissions modeling efforts and to describe the effect of different factors on emission levels and modeling accuracy, so as to get reliable emission estimates. In addition, different models were evaluated for the prediction capability of certain emissions such as carbon monoxide (CO), nitrogen monoxide (NO), nitrogen dioxide (NO2) and hydrocarbons (HC). These models are mainly based on traffic volume, composition, and flow. The predicted values by one of the models were compared to measured values based on field surveys. The result of comparison indicated that there is a significant difference between the measured and predicted values. These differences ranged from 17% for NO2 to 72% in the case of CO, which suggests that the NO2 model has better predictability. This deviation in prediction may be attributed to the fact that prediction models ignored some of the parameters affecting vehicle emissions such as the type of fuel used, air–fuel ratio, engine compression ratio, spark timing, surrounding environment, wind effect, regional characteristics and high pollutants emitters effect.
S. Abo-Qudais Æ H. A. Qdais (&) Civil Engineering Department, Jordan University of Science and Technology, Irbid, Jordan, P.O.Box 3030 E-mail:
[email protected]
Introduction Vehicles’ emissions are considered as one of the major sources of air pollution in urban areas. Studies in the United States of America (USA) indicated that 45% of the pollutants released in the USA are a direct consequence of vehicle emissions (NRC 1995). During 1998, in the USA on-road vehicles were estimated by the US EPA to contribute 30% of hydrocarbon (HC) emissions, 32% of NOx, and 9% of particulate matter (PM) emissions (US EPA 1998). Motor vehicle emissions and fuel consumption depend on a large number of factors which may come under two broad categories (Cloke et al. 1998), namely technical factors related to the design and engineering of vehicle, and operational factors related to the way in which the vehicle is used. In order to quantify the emissions produced by traffic and the impacts of air pollution on the human health, accurate information on the composition and the concentration of the pollutants is needed. Mathematical models are one way that is used to predict the concentrations of various pollutants. Modeling approaches There have been many models developed worldwide to predict vehicular emissions for different road conditions, vehicles status, and driving modes. These models can be grouped under three main categories, namely, emission factors models, average speed models, and modal models. (Cloke et al. 1998). Emission factors models employ single emission factors for individual types of vehicles operating in a particular type of driving conditions. Examples of such models are California’s EMFAC, MOBILE (US EPA 1994), and CHINA- MOBILE emissions model (Hao et al. 2000). The existence of discrepancies between real world measurements and predictions of automotive emissions factor models has been reported by several
280
researches. One reason for this discrepancy is a difference between real-world vehicles, including the way they are maintained and the way they are driven, and estimates based on measurements of the vehicles used in the models (Robinson et al. 1996). The average speed models express average emission rates for each trip in terms of the average speed. The measurement takes place in a laboratory under dynamometer testing for a variety of simulated trips with different average speeds. Example of this type of model is COPERT (Computer program to calculate emissions from road traffic), which is mainly applied to mediumand large-scale emission estimates using average speeddependent emission factors. This type of model is not sensitive to variations in operational modes of vehicles, as it depends mainly on average speed without taking into account the influence of acceleration and deceleration (Barth et al. 1996). Thus, such models are not applicable in cases when changes of operational modes are encountered. The modal models were developed to estimate emissions at microscopic level based on extensive vehicle testing at high resolutions. Several researchers (Zachariadis and Samaras 1997; Kishi et al. 1996; Hansen et al. 1995) characterized the vehicle modal events by using a speed/acceleration matrix, which gives instantaneous emissions rates for different combinations of instantaneous speed and acceleration. Andre et al. (1995) used the product of speed and acceleration to estimate the vehicular emissions. The MODEM emission model was developed on this basis. Data on emissions and fuel consumption were classified into different classes of speed and speed acceleration products. Instantaneous emissions are then estimated by selecting values from the corresponding combinations of speed and speed acceleration product. Such modelling approaches are highly data intensive, which require time and manpower resources to collect the data. Ingalls and Garbe (1982) reported different models that predict the emission rates for noncreative pollutants in enclosed space, street canyons, and expressways. The reported model for street canyons predicts the concentration of pollutant (mass/volume) in terms of pollutant generation rate per meter, wind speed, and distance from source to receptor. In this model the building heights to street width ratio should be more than 0.3 in order for the model to be applicable. As part of environmental impact assessment of a road widening project, Abu Qdais and Abo-Qudais (2000) used empirical formulas to estimate the emission rates of different pollutants after the road widening. The vehicles, emissions in a given section of a roadway and length per area and time units is the sum of the product of the section length, the corresponding emission factor, and the number of vehicles of each category that drive along that section during the time span. The estimated rates of emissions as a result of road improvement were found to be within the acceptable limits of both European and American standards.
During the last two decades, Jordan has witnessed a rapid increase in population. In 1980 the country population was 2,218,000, which jumped to about 5 million in the year 2000. This increase in population was accompanied by increase in demand on the transportation sector. As a result, the number of the registered vehicles has increased by about 25-fold since 1970. The number of registered vehicles was 21,970 in 1970 and increased to 530,000 vehicles by the end of year 2000 (Jordan Traffic Institute 2000). Vehicles are a major source of air pollution in Amman and other major cities of Jordan. The problem is further complicated by the fact that Jordan is not an oil producing country and mainly depends on imported fuel. The fuel imported by Jordan is not as pure as required from an environmental point of view. The lead content of the Jordanian gasoline is 1%, while the diesel used in Jordan contains about 1.5% of sulfur, which is six times higher than the European Standards for DK3 diesel. Table 1, gives the sulfur content in the Jordanian diesel as compared to the diesel used in some countries of the world (Samara 1998). The maximum allowable limits of emissions from vehicles according to the Vehicles emissions standards issued by the Ministry of Interior in 1996 are 5% for carbon monoxide (CO), 6,000 ppm for hydrocarbons (HCS), and Smoke intensity (the ability of smoke to scatter the light due to the presence of PM) not greater than 65%. Although, these standards allow relatively high emission levels, as compared to other countries’ standards, the percentage of vehicles producing emission higher than those allowed in Jordanian standards is still high. In an attempt to evaluate the status of the vehicular emissions in Jordan, the Jordanian Society of Road Accidents Prevention has conducted a study in 1995. A total of 666 vehicles with different ages were tested for compliance with the emissions regulations. Table 2 shows the results of this study on vehicles driven by gasoline. From this table it can be observed that the vehicle age is a mojor factor affecting the emission level produced by the vehicles. The vehicle percentage that does not comply with the regulation is increasing with increase in the vehicle age. In order to phase out the old vehicles in Jordan, the Jordanian governemnt nowadays is allowing the public vehicles owners (taxies) to change their vehicles with new ones free of customs duty.
Evaluated models The Environmental Protection Agency (EPA) MOBILE model and the California Air Resources Board (CARB) Table 1 Sulfur content in diesel fuel (Samara 1999) Country
USA Algeria France Norway Jordan
Sulfur content (% wt.) 0.95
0.13
0.17
0.36
1.5
281 Table 2 Compliance of vehicle with emission standards in Jordan (Jordanian Society for Preventing Roads Accidents 1995) Vehicles manufacturing year
Percent of vehicles failed CO2test
Percent of vehicles with CO emission exceeded 5%
Percent of vehicles with HC emission exceeded 600 ppm
1980 and before 1981–1985 1986–1990 1995–1991
51.8% 28.5% 17.3% 23.3%
57% 51.4% 29% 30%
33.5% 20.4% 13.5% 16.6%
EMFAC model were the two main emission models commonly used in the USA. In these models, emission rates are predicted based on vehicle type and age, vehicle average speed, ambient temperature, and vehicle operating mode. Both models produce specific emission rates. These emission rates are multiplied by vehicle activities such as vehicle miles-traveled, number of trips, and vehicle-hours traveled in order to estimate total emission levels (NRC 1995). Current estimates of emission rates of both models are expressed as functions of average speeds and are based on vehicle testing on a limited number of driving cycles. The baseline emission rates for light duty vehicle are composed of three different phases: a cold-start phase, a stabilized phase and a hot-start phase. In the MOBILE model, the emissions from vehicles operating in all three phases are used to estimate baseline emissions. The baseline emission rates for a vehicle class is the average result from the three phases at an average speed of 31.6 km/h, which is the average tested speeds. In the EMFAC model, the baseline emission rate is derived from only the stabilized phase with an average operating speed of 25.6 km/h (Guensler et al. 1993). Emission rates at other average speeds are multiplied by the appropriate speed correction factor (SCF) associated with a vehicle class and the operating speed. The speed-corrected emission rates used in emission models are highly dependant on the average cycle speed (NRC 1995). Both models (MOBILE and EFMAC) were found to have many drawbacks (Ahn 1998). A limited set of driving cycles, which insufficiently represent specific traffic flow conditions, are used to estimate emission rates. Another problem with emission estimations is that the two models predict emission rates based of changes in average speed among traffic flow characteristics. Average trip speeds are not equivalent to link-specific speeds for portions of vehicle trips. This method of using average speeds cannot represent the distribution of speeds and accelerations of a trip, which vary by type of road and level of congestion (Frey et al. 2001). Ahn (1998) developed mathematical models to predict vehicle emissions. These models have the capability to predict single vehicle emission based on its speed and acceleration rate. The Georgia Tech’s MEASURE and UC riverside’s models were also developed to obtain data on vehicular emissions based on specific driving cycles. Some of the driving cycles used in these models have been developed specifically for driving purposes. (Frey et al. 2001).
These models did not consider other factors that might affect the vehicle emission rates; also it does not give any idea about the concentration of the pollutants in the vicinity of roads used by these vehicles. Al-Khateeb (1993) developed statistical models that have the capability to predict different pollutant concentrations in the air for traffic with frequent stopping. These models were a function of traffic volume and developed based on data collected from field studies. The models have the capability to predict the concentration of CO, nitrogen monoxide (NO), and nitrogen dioxide (NO2) pollutants in terms of traffic volume (vehicles per hour (vph)). The models have the following forms: CO ðppmÞ ¼ 0:004331 ðvphÞ0:993 NO ðppmÞ ¼ 0:000147 ðvphÞ0:949 NO2 ðppmÞ ¼ 0:000038 ðvphÞ1:008 For a traffic volume of 1,260 vph, the predicted pollutant emissions based on these models were 5.19 ppm, 0.13 ppm, and 0.05 ppm for CO, NO, and NO2, respectively. On the other hand, the measured values of the same pollutants at the same traffic volume were found to be 18.50 ppm, 0.22 ppm, and 0.06 ppm for CO, NO, and NO2, respectively, (Wanner et al. 1980). Figure 1 shows the predicted and measured values of the three pollutants based on the statistical models developed by Al-Khateeb (1993). It can be observed that the predicted values by the models for all the three pollutants are less than the measured ones. The difference ranges from 17% in the case of NO2. to 72% for CO.
Fig. 1 Predicted and measured values of different pollutants based on statistical models developed by Al-Khateeb (1993) CO concentrations in ppm, while NO and NO2 concentrations in ppb
282
This is in agreement with the results obtained by the other researchers (Washington and Guensler 1995; AlOmari 1998) who reported the poor prediction capability of such models. The poor performance of such models may be attributed to the fact that the developed models ignored some of the factors that affect the rate of pollutants emissions as discussed in the following section.
Physical parameters affecting prediction capability of the models The success of any developed model to predict pollutant emission depends on adequately defining the range of physical parameters affecting pollutants concentrations in the vicinity of the road under evaluation. Although some of the developed models considered part of these parameters, they ignored others that have significant effects on pollutant concentration. Defining accurate values of all physical parameters is a hard task. However, assuming reasonable values for these parameters, based on previous studies, will be better than ignoring them. The main physical parameters affecting pollutant concentration in the vicinity of roads are discussed below. Type and composition of used fuel Type and composition of fuel have a significant effect on pollutants emission. Diesel fueled engines emit different amounts and types of pollutants compared to those of gasoline engines. In Jordan, for example, about 73% of the vehicles use gasoline, while the rest are using diesel fuel (Traffic Department 2000). It was found that CO emission from gasoline engines is four times greater than those from diesel engines. Table 3 shows comparison of different emissions types emitted from both engines. The fuel composition varies according to crude source, used refining technique, and type and amount of additives. This variation in fuel composition will affect the emitted pollutants. For example, most of the gasoline used in Jordan contains 1% of lead, while many other countries are using unleaded gasoline. Another example, the diesel in Jordan, contains about 1.5% of sulfur. The amount of sulfur is considered high compared to that used in other countries as shown in Table 1.
Engine revolution per unit time Ingalls and Garbe (1982) reported that engine emmisions are greatly affected by vehicle speed. At low speed the emission of some polutant will be higher, while the emission of some other polutant will be lower. This statement might be more accurate if the vehicle speed is replaced by engine revolution per minute (rpm) or operating mode. The engine rpm is a function of gear level, road gradient, laod in the vehicle, acceleration, vehicle type, vehicle speed, and vehicle activity; acceleration, deceleration, or cruse speed. For example, the amount of HCs emission during acceleratin activity will be about 10% of that during deceleration activity, while the amount of nitrogenoxides emission during acceleration is about 50 times of that during deceleration activity. In general, emissions are higher under low speed and congested driving conditions. Also, the uncertainty of predicting the emitted pollutant concentration at cruse speed will be much less than that at acceleration and deceleration activities (NRC 1995). Table 4 shows the effect of the vehicle operation on the exhaust emissions. Some of the developed models, such as EMAC, developed by Californai Air Resources Board, and CALINE4, developed by Claifornia Department of Transportation, considered the effect of vehicle activity. However, these models are still in need to be improved for different stages in the same activity (Washington and Guensler 1995). Ambient temperature Ambient temperature affects both exhaust and evaporative emissions. The engine and emission control systems take longer to warm up at cold temperatures, which will lead to an increase in cold-start emissions. It was found that an increase in ambient temperature of 1° F will lead to a reduction of 1.3% and 2.8% in HC and CO emissions, respectively. On the other hand, an increase in temperature is usually associated with an increase in NOxemissions. (Samara 1996) Effect of high pollutant emitters Most of the developed models ignore the effect of high emitters on the pollutant concentration in the air.
Table 3 Typical exhaust emissions from gasoline and diesel engines Table 4 Effect of the vehicle velocity on the exhaust emissions Emissions
Gasoline engine (%)
Diesel engine (%)
CO HC NOX CO2 O2 SO2
0.8–5.2% 0.03–0.04% 0.2–0.06% 9.0–12.5% 3–5% 0.05–0.18%
0.1–1.6% 0.002–0.004% 0.15–0.04% 8.0–11.0% 7–10% 0.02–0.03%
Mode of operation
HC (ppm)
CO (% vol)
NOX (ppm)
CO2 (% vol)
H2 (% vol)
Idle Cruise Acceleration Deceleration
750 300 400 4000
5.2 0.8 5.2 4.2
30 1500 3000 60
9.5 12.5 10.2 9.5
13 13.1 13.2 13.0
283
Vehicles using large engines usually emit more pollutant than those using relatively small engines. Also, older vehicles usually emit higher amounts of pollutant. It was reported that CO emissions from 4 year old vehicles are about 1.5 times higher than that of 2 year old ones (Muttamara and Leong 2000). All the cited models were based on data collected randomly from vehicles with different ages and sizes. No attention has been paid to the effect of vehicle age and/or vehicle size on the emission rates. Roads geometric design Emission rates of vehicles depend on the geometric design of roads. Roads with facilities such as signalized intersections, toll booths, and weaving sections may increase the emission levels due to the engine enrichment from accelerations. Road grade is one of the main contributors affecting emission rates. On a steep grade, vehicles require more engine power, causing higher fuel consumption and consequently higher emissions. Road surface condition also has an affect on emission rates. Rough road surface require higher engine power to keep the same operating speed, so more pollutants will be emitted. Regional characteristics The developed models should take into account the impact of the conditions dominated within the region where the emissions will be predicted. The conditions such as land use, predominate wind direction, and humidity are of great importance for the process of emissions estimation. Consequently, the local data should be used to estimate various models’ coefficients so as to include the influence of those regional conditions. Air–fuel ratio The air–fuel ratio (A/F) by mass is one of the most important variables affecting the efficiency of catalytic converters and the level of exhaust emissions (Johnson 1988). Air–fuel ratio affects the amount of pollutant emitted from the engine. For example, NOx emission reaches its maximum value at A/F of 15, while CO and HC reach their minimum values at A/F value of 18. (Noel De Nerves 2000).
of pollutant per unit weight of consumed fuel might increase or decrease with increasing ratio compression ratio. At engine compression ratio of 8.5 the CO emissions reaches its minimum level, while HC reaches its maximum level. Since most of the developed models ignore many of the above mentioned factors, the prediction capability of these models is relatively poor. The evolution of new models that take into account some or all of these factors is crucial for many countries around the world. These models, if developed, will be useful tools for the decision makers and planners in the transportation and environmental sectors to base their decisions on a solid ground. Also such models will help in conducting the environmental impact assessment of the road construction projects.
Conclusions Traffic emissions are known to contribute significantly to urban air pollution. In an attempt to predict the traffic emission levels, and to assess their impacts on the public health and the environment, several models were developed. These models are capable of predicting emissions with various degrees of success. The difference between measured and predicted emissions by one of the evaluated models varies from 17% for NO2 to 72% for CO. This is may be attributed mainly to the fact that most of these models consider only few factors that are affecting the emissions rates and ignore many others. The emission levels predicted by these models are significantly lower than the measured values. This suggests that there is a need for more accurate models, which consider most of the factors that affect the emissions. These models will be useful tools in making decisions regarding traffic planning and management. The new models should be capable of measuring the change in emission levels resulting from strategies that affect the traffic flow. Furthermore, the models should be subjected to calibration in order to be used for prediction under conditions different from those under which they were developed. Efforts should be directed generally towards learning how to use such models, including setting up the appropriate inputs and parameters, and calibration. Model inputs should be evaluated for their significance on the model predictability. In case these parameters are found to be insignificant, they should not be considered in the model development.
References Engine compression ratio Another factor that has an effect on the amount of engines pollutants emissions is the compression ratio. Increase of compression ratio usually increases the efficiency, reducing the fuel consumption, and so decreases the amount of pollutants emitted. However, the amount
Abu Qdais H, Abo-Qudais S (2000) Environmental impact assessment of road construction projects. Environ Ecol 18(2):405–419 Al-Khateeb G (1993) Effect of traffic movement and volume on air pollution in irbid city. Master thesis, Civil Engineering Department, Jordan University of Science and Technology, Irbid, Jordan
284 Al-Omari A (1998) Interaction between urban planning, transportation system, and vehicle emissions in urban arterials. Master thesis, Civil Engineering Department, Jordan University of Science and Technology, Irbid, Jordan Barth M, An F, Norebck J, Ross M (1996) Modal emission modeling: a physical approach, transportation research record, No. 1520, pp 81–88 Cloke J, Boulter P, Davies GP, Hickman AJ, Layfield RE, McCare IS, Nelson PM (1998) Traffic management and air quality research program, Transportation Research Laboratory, Report 327 DOT, EPA (1993) Clean air through transportation: challenges in meeting national air quality standards Frey HC, Rouphail NM, Unal A, Colyar DJ (2001) Measurement of on-road tailpipe CO, NO, and hydrocarbon emissions using a portable instrument. In: Proceedings of the annual meeting of The Air and Waste Management Association, June 2001, Orlando, Florida Guensler F, Washington S, Sperling D (1993) A weighted disaggregate approach to modeling speed correction factors, Transportation Research Record, Washington, D.C., p 44 Hansen JQ, Winter M, Sorenson SC (1995) The influence of driving patterns on passenger car emissions. Sci Total Environ 169:129– 139 Hao J, He D, Wu Y, Fu L, He K (2000) A study of the emission and concentration distribution of vehicular pollutants in the urban areas of Beijing. Atmos Environ 34:453–465 Ingalls MN, Garbe RJ (1982) Ambient pollutant concentrations from mobile sources in microscale situations, Society of Automotive Engineers, Inc. Journal, paper No. 820787, pp 1–16 Johnson JH (1988) Automotive emissions, air pollution, the automobile, and public health, Health Effects Institute. National Academy Press, Washington D.C Jordan Traffic Institute (2000) Traffic accidents in Jordan, Directorate of Interior, Amman, Jordan, p 93 Jordanian Society for Preventing Roads Accidents (1996) Study on the emissions of vehicles in Jordan, Amaan Jordan, p 8 Joumard R, Jost P, Hickman J (1995) Influence of instantaneous speed and acceleration on hot passenger car emissions and fuel consumption. SAE 950928, Warrendale
Kishi Y, Katsuke S, Yoshikawa Y, Morita I (1996) A method for estimating traffic flow fuel consumption—using traffic simulations. JASE Rev 17:307–311 Ahn K (1998) Microscopic fuel consumption and emission modeling. Master thesis, Virginia Polytechnic Institute and State University, Blacksburg, VA, p 131 Muttamara S, Leong ST (2000) Monitoring and assessment of exhaust emission in bangkok street air. Environ Monit Assess 60:163–180 National Research Council (NRC) (1995) Expanding Metropolitan Highways: Implications for Air Quality and Energy Use. National Academy Press, Washington Nizich SV, McMullen TC, Misenheimer DC (1994) National air pollutant emissions trends, 1900–1993. EPA-454/R-94-027. Office of Air Quality Planning and Standards, Research Triangle Park, N.C., p 314 Noel De Nevers (2000) Air pollution control engineering, 2nd edn. McGraw Hill, New York, p 477 Robinson NF, Pierson WR, Gertler AW, Sagebiel JC (1996) Comparison of Mobile 4.1 and Mobile 5 predictions with measurements of vehicle emission factors in Fort McHenry and Tuscarora Mountin tunnels. Atmos Environ 30(12):2257–2267 Samara N (1996) Proceedings of the first national engineering conference. Amman, Jordan US EPA (1998) National air quality and emissions trends report, EPA 454/R – 003, Research Triangle Park, NC, 1998 Wanner HU, Deuber A, Satish J, Meier M, Sommer H (1980) Air polution in the vicinity of streets, Atmos Poll. Elsiever, Amsterdam, pp 99–107 Washington S, Guensler R (1995) Statistical assessment of vehicular carbon monoxide emission prediction algorithms, transportation research record, No. 1472, Washington D.C., pp 61– 68 Zachariadis Th, Samaras Z (1995) Comparative assessment of European tools to estimate TRAFFIC emissions. Int J Vehicle Design 18:312–325