Sei sulla pagina 1di 8
Compendium of Papers Emissions and Fuel Consumption Estimation for Vehicles on the Road and Involved

Compendium of Papers

Emissions and Fuel Consumption Estimation for Vehicles on the Road and Involved in Injury Collisions

Rúben Silva, Guilhermina Torrão, Margarida C. Coelho *

Centre for Mechanical Technology and Automation / Department of Mechanical Engineering, University of Aveiro, Campus Universitário de Santiago, 3810-193 Aveiro, Portugal * Corresponding author:


The European Commission (EC) imposed mandatory CO 2 emissions of 130 g/km by 2015 for the new vehicle models. The European New Car Assessment Program (EuroNCAP) promotes safety developments, and credits car manufacturers for focusing on safety. Despite of EC regulations and EuroNCAP safety reinforcement, environmental and safety performances must be optimized by the automotive industry. The main objective of the current research was to analyze the effect of vehicles’ characteristics on crash severity, emissions and fuel use. For the safety and emissions analysis, the technical information of engine size, weight, wheelbase and age for those vehicles involved in injury collisions was extracted from a crash dataset. This dataset was developed for real crashes gathered for the time period of 2006-2008, in Portugal. A binary logistic model was developed to analyze the effect of vehicles characteristics on crash severity. Copert IV methodology was applied for to estimate the emissions of those vehicles. “Vehicle Specific Power” (VSP) methodology was applied to estimate emissions and consumption for two alternative routes choice: a motorway and a national rural road, that connect Aveiro and Porto cities, in Portugal. The preliminary safety research findings showed that engine size has an important effect on crash severity. The emissions and fuel consumption were lower for diesel vehicles, except for nitrogen oxides and particulate matter.

© 2012 The Authors.

Keywords: Copert IV; crashes; emissions; fuel consumption; vehicles; VSP

1. Introduction and Research Objectives

During the past decennia there has been a steady increase in traffic volume, which has resulted in continuously increasing traffic congestion-related problems, such as pollutants emissions, delays, crashes, injuries and casualties and economical losses. Recent status on road safety from World Health Organization (WHO) showed that more than 1.2 million people die on the world´s roads every year (WHO, 2009). The latest road safety indicators from the Portuguese National Authority for Road Safety (ANSR) show that during the year 2010, there have been a total of 35,426 crashes with injuries and fatalities on the Portuguese mainland roads. As consequence, there were 741 fatalities and 2637 serious injuries (ANSR, 2011). Crash testing is an information resource for consumers. However, EuroNCAP discourages consumers from comparing ratings of cars from different segments, and in real crashes, there is obviously no control on the vehicle categories involved


EWGT2012 – Compendium of Papers

(EuroNCAP, 2009a, 2009b). The main goal of this research was to analyze the effect of vehicle characteristics on safety, emissions and fuel consumption. The three specific objectives for this research were:

- To assess whether vehicle characteristics impact crash severity; - To estimate emissions for vehicles involved in real road crashes, with Copert IV methodology; - To compare the emissions estimates and fuel consumption for two alternative routes by applying the VSP methodology. The paper is organized as follows: chapter 2 provides a literature review, followed by methodology in chapter 3, results and discussion in chapter 4, and finally conclusions in chapter 5.

2. Literature Review

In the literature, most attention is paid to vehicle type and risk to the drivers, but not to its relation to crashworthiness. There is a lack of a methodology to estimate the effect of vehicles characteristics with the overall crash severity during vehicles collisions. Several studies have attempted to correlate safety and vehicle design features. Evans (2004) explored vehicle mass and size, and concluded that those variables are strongly correlated, which makes it difficult to determine the separate contribution of mass and size on crash risk. Wood and Simms (1997) showed that in collisions between cars of similar size, the fundamental parameters which determine the injury risk is associated to the size, i.e. the length of the vehicle. Furthermore, in collisions involving two cars of different size, the use of EuroNCAP ranking is not the most proper way to compare different safety performance, mainly due to the mass difference between the two vehicles involved in the crash (Coelho et al., 2010). Broughton showed that the driver casualty rate decreases with the size of his car, however the driver casualty increases with the size of the other car involved in the collision (Broughton, 2008). More recently, Wenzel (2010) suggested that the vehicle design would have a greater effect on occupant safety than fuel economy standards. Previous studies related to crash analyses have used a broad spectrum of statistical models. For example, statistical regression models are very popular to estimate injury severity (Li and Bai, 2008; Bedard et al., 2002; Boufous et al., 2008; Al-Ghamdi, 2002; and Chang and Wang, 2006). Kononen et al. (2011) have developed a logistic model for predicting serious injuries in vehicle crashes. However, regression models have assumptions and underlying relationships between the dependent and independent variables (Chang and Wang, 2006). Regarding fuel economy standards, Chen and Kockelman (2012) study suggested that moderate changes in vehicle weights and wheelbase are estimated to have small impacts on crash severity. In summary, much has been said about the high risk of low-mass cars in certain kinds of collisions, with focus only in the drivers’ injuries or fatalities and excluding vehicles’ passengers. There is no doubt that vehicle design can influence not only avoidance and crashworthiness, but also whether it endangers the occupants of the other vehicle(s) involved in the collision. In the preceding studies, the effect of differential vehicle weight and size and its impact on overall crash severity has not been studied. Hence, further research is needed to address the issue of the vehicle technical data effect on road crash risk, expressed by injuries or fatalities risk to all the occupants within the vehicles involved in the crash. In addition, the authors did not find any word on the trade-off of two different impacts (road safety and emissions / fuel use).

3. Methodology

The motivation for this research was to focus on the vehicle fleet characteristics and analyze those that might have a stronger impact on crash severity, which is expressed by the risk to drivers and passengers, based on real- world crash data. This analysis explores the contribution of vehicle specific characteristics on crash severity and

EWGT2012 – Compendium of Papers


on vehicles’ emissions and fuel independent. The independent variables in the dataset include: make and model, weight (mass), engine size (power), wheelbase, year of registration (age) and fuel type. Dependent variables were related to the occurrence or number of severe injuries and/or fatalities of the vehicles occupants. The research focused exclusively on post-crash consequences rather than on pre-crash contributing factors to the event, such as driver behavior or driver’s age. Figure 1 summarizes the steps that were developed, which includes the preliminary findings presented in this paper.

includes the preliminary findings presented in this paper. Figure 1 – Methodology overview towards an integrated

Figure 1 – Methodology overview towards an integrated analysis of environmental, fuel efficiency and safety vehicle performance.

3.1 Crash Selection and Dataset Development

Data for the crash severity predictive model were collected from the Road Traffic Departments of the Portuguese Road Safety Police National Republican Guard (GNR). Crash reports that involved injuries and/or fatalities outcomes were exclusively selected. A total of 1345 crashes reports were gathered for a time period from 2006 to 2008, for the urban areas of Aveiro and Porto, in Portugal. The vehicle technical features were requested to the Portuguese Institute for the Mobility and Inland Transportation (IMTT) and vehicle technical detailed information for 25% of the crashes (342/1354) was obtained. An integrated database was developed where each crash record and technical characteristics for each vehicle involved in the collision were combined into a unique crash observation. The developed database included a specification sheet for technical attributes of the vehicle’s registration plate. An example of the vehicle’s characteristics acquired from those databases is listed below: Brand (Toyota), Model (Corolla), Wheelbase (2465 mm), Size (4095 mm), Weight (1045 kg), Engine Size (1332 cm 3 ), Fuel (Gasoline).

3.2 Statistical Analysis

For the safety analysis and crash severity predictive modeling the Software, SAS® v9.2 and SAS®Enterprise MinnerTM6.2 was applied (SAS Institute Inc., 2009). The response variable (target) was driven from the crash outcomes, which were expressed by the number of injuries and/or fatalities, amongst the vehicles’ occupants. The


EWGT2012 – Compendium of Papers

Target was converted to a binary variable to define the probability of a Serious Injury or Fatality in a crash, and it was defined as FatalSIK. Hence if FatalSIK”1” the crash was severe. On the other hand, if FatalSIK”0” the crash was non severe. A binary logistic regression model was developed to predict the probability of a crash outcome resulting in FatalSIK”1”. A detailed description of a database as well as an overview of the safety models is described elsewhere (Torrão et al., 2011).

3.2. Fuel Consumption and Emissions

The estimation of road traffic emissions and fuel consumption was based on the application of two models:

Copert IV (CORINAIR, 2010) and VSP (US EPA, 2002; Frey et al., 2008). Copert IV is based on the European emission standards which are related to the acceptable limits for emissions of new vehicles sold in EU member states. Emission standards for passenger cars and light commercial vehicles are coded as Euro Stage: Euro1 to Euro 5 based on the year that the vehicle model starts to be sold in the market. EU regulations introduce different emission limits for diesel and gasoline vehicles. This model was used to estimate the emissions for the vehicles involved in injury collisions, at the crash dataset. Vehicles attributes, such as, engine size, model year (age) and fuel type were used as input, and road speed limit, as well. The VSP methodology was applied to estimate the emissions/fuel use for two alternative routes: a motorway and a national road. For light duty vehicles the following equation is used:

VSP = v[1.1×a + 9.81×sin(atan(grade)) + 0.132] + 0.000302×v 3


where VSP is the vehicle specific power (m 2 /s 3 ); v is the instantaneous vehicle speed (m/s); a is the instantaneous acceleration (m 2 /s); grade is the road grade (%). In this methodology, vehicle fuel use and emissions are stratified in 14 bins interval classification, from bin 1 through bin 14. Hence, bins 1 and 2, with negative values of VSP, are related to the deceleration or travel on the downgrade. Bin 3 addresses the emissions during idling periods. Bins 4 to 14 refers to increasing VSP values, related with higher speeds and accelerations. Each VSP Bin category is assumed to generate a fixed emission rate for each pollutant. The average emission rates by VSP bin for gasoline and diesel vehicles are indicated elsewhere (Coelho et al., 2009; Frey et al., 2008; US EPA, 2002). The totals for fuel use or emissions for a given scenario were calculated by multiplying the standardized average rates with the average route travel time. Since the crash population was gathered for the Porto metropolitan area, two routes were chosen to connect Aveiro and Porto cities. The two selected routes were the national road EN109 and the Motorway A29 and typical speed profiles previously measured in those roads were applied (Bandeira et al., 2011, 2012). The authorized speed limits are 90km/h and 120km/h for EN109 and A29, respectively. The extension of the EN109 route is approximately 74 km, while for A29 is 76 km.

4. Results and Discussion

This chapter presents the initial results for the safety analysis, followed by emissions estimates for the vehicles included in the crash dataset. Finally, a comparison of emissions and fuel use for two alternative routes is also discussed.

4.1. Safety Modeling – Logistic Regression for FatalSIK

During the data mining analysis, several types of models were attempted, and the target and the outcome

EWGT2012 – Compendium of Papers


percentage of each model were evaluated. The selected model to predict the target FatalSIK, based on the total profit for the training dataset (for those 342 crash observations) is illustrated in Figure 2 and described by the analytical equation indicated below.

2 and described by the analytical equation indicated below. (2) where P( Y=1 ) is the


where P(Y=1) is the probability of a serious and/or fatality occurrence (FatalSIK”1”) during a two vehicle collision, and ccV2 is the engine size for vehicle V2 involved in the crash.

is the engine size for vehicle V2 involved in the crash. Figure 2 – Effect of

Figure 2 – Effect of Engine Size of Vehicle 2 on FatalSIK for two-vehicle collisions

The likelihood ratio test for the global null hypothesis showed a Pr>ChiSq value of 0.0148. The analysis of the maximum likelihood estimates for the intercept and the ccV2 showed a Pr>ChiSq <0.0001 and 0.0106, respectively. The misclassification rate was 0.05. Despite the fact that these statistics were acceptable (low misclassification) and the finding that ccV2 was found to be of statistical importance, the predictive model did not have a satisfactory performance for the minority class being predicted, FatalSIK”1”. The percentage predictions matching the target “0” was 95.3% and those severe crashes which were misclassified as non-severe were at 4.7%. This analysis showed that in the training sample, the model predicts correctly all the non-severe crashes (FatalSIK”0”), however all the true positive (16 in total) were predicted as non-severe as well.

4.2. Road Traffic Emissions and Fuel Estimation– CopertIV and VSP Models

For the road traffic emissions analysis, Figure 3 shows the CO2 emissions based on Copert IV methodology for the vehicles involved in injury collisions. The illustrated CO2 emissions for passengers’ cars with gasoline engine are presented as function of vehicles’ age, in roads where the legal speed limit is set at 120 km/h. For the same vehicle engine size category, the newer models tend to emit less CO2, when compared with older vehicles models of the same engine size category. The results showed that engine size and speed have a strong effect in CO2 emissions. Figure 4 presents the results for CO2 emissions based on the VSP methodology. Gasoline vehicles emitted lower rates of CO2 emissions when compared with diesel vehicle engines. This finding is consistent with recent research (Wenzel, 2010; An et al., 2011).


EWGT2012 – Compendium of Papers

6 EWGT2012 – Compendium of Papers Figure 3 – CO 2 estimations for light duty vehicles

Figure 3 – CO 2 estimations for light duty vehicles with gasoline engine vs. vehicles’ age, in routes with 120 km/h speed limit, by Copert IV.

age, in routes with 120 km/h speed limit, by Copert IV. Figure 4 – CO 2

Figure 4 – CO 2 emissions (g/km) for gasoline vehicles (blue bars) and diesel vehicles (green bars) based on the average speed per kilometer (red line), by VSP methodology.

EWGT2012 – Compendium of Papers


EWGT2012 – Compendium of Papers 7 Figure 5 – Fuel estimation for two different routes: EN109

Figure 5 – Fuel estimation for two different routes: EN109 and A29.

Figure 5 presents the fuel estimation results based on the VSP methodology. The average fuel consumption was higher at the EN109 when compared to the A29. Even though A29 is a longer route (two kilometers more) and the expected average speed levels higher for motorways, the fuel demand was lower than in EN109. In main roads there are usually more traffic congestion and the speed profile tend to have more fluctuations, hence imposing higher fuel consumption. On the other hand, CO and NO x emissions rates were higher at A29 route.

5. Conclusions

The main conclusions of this study can be drawn in terms of the following findings:

- The binary logistic regression model for the FatalSIK prediction showed that the engine size of vehicle V2 increases the probability of a serious and/or fatal in a two vehicles collision.

- Regarding VSP emission results for the analysis of the effect of route choice (motorway and alternative

route), driving in national road EN109 required 33% more fuel consumption when compared with the driven in the motorway A29. In addition, CO 2 emissions were higher when travelling in the EN109.

- Regarding vehicle’s safety, emissions and fuel economy, engine size was proven a significant attribute and

this variable affects not only the crash severity outcome as it increases the CO 2 emissions, and consequently, the

fuel consumption. Future work will be focused on the development of a multi-objective analysis, in order to assess both dimensions (safety and emissions) and create an indicator that describes the existing trade-off. Further, a new strategy to plan road safety and environment policies must address not only the vehicle fleet, but also the route choice taken by the drivers.


This work is funded by FEDER through the Operation Program “Factores de Competitividade – COMPETE” and by National Funds through FCT – Fundação para a Ciência e a Tecnologia within the project PTDC/SEN- TRA/113499/2009. The authors also acknowledge the support of FCT for the Scholarship SFRH/BD/4135/2007.


EWGT2012 – Compendium of Papers


Al-Ghamdi, Al. (2002). Using logistic regression to estimate the influence of accident factors on accident severity. Journal of Accident Analysis & Prevention, 34(6), 729 - 741. An, F., Earley, R., & Green-Weiskel, L. (2011). United Nations Department of Economic and Social Affairs. Global Overview on Fuel Efficiency and Motor Vehicle Emission Standards: Policy Options and Perspectives for International Cooperation. May 2011.


ANSR (2012). Annual Report 2011. (in Portuguese). Portuguese National Authority for Road Safety. Bandeira, J., Almeida, T., Khattak, A., Rouphail, N., & Coelho, M. C. (2011). Characterization of urban environment to enhance the driver’s route choice – Application to urban and suburban routes. Proceedings of the 90th Transportation Research Board Annual Meeting, Washington D.C., January 2011. Bandeira, J., Carvalho, D., Khattak, A., Rouphail, N., & Coelho, M. C. (2012). Impacts on Emissions and Fuel Consumption of Route Choice Decision – Influence of Peak Hour in Different Realities. Proceedings of the 91st Transportation Research Board Annual Meeting, Washington D.C., January 2012. Bedard, M., Guyatt, G. H., Stones, M. J., & Hirdes, J. P. (2002). The independent contribution of driver, crash, and vehicle characteristics to driver fatalities. Journal of Accident Analysis & Prevention, 34(6), 717 - 727. Boufous, S., Finch, C., Hayen, A., & Williamson, A. (2008). The impact of environmental, vehicle and driver characteristics on injury severity in older drivers hospitalized as a result of a traffic crash. Journal of Safety Research, 39(1), 65 - 72. Broughton, J. (2008). Car driver casualty rates in Great Britain by type of car. Journal of Accident Analysis & Prevention, 40(4), 1543 - 1552. Chang, L., & Wang, H. (2006). Analysis of traffic severity: an application of non-parametric classification tree techniques. Journal of Accident Analysis & Prevention, 38(5), 1019 - 1027. Chen, T., & Kockelman, K. (2012). The Roles of Vehicle Footprint, Height, and Weight in Crash Outcomes: Application of a Heteroscedastic Ordered Proibit Model, Proceedings of the 91st Annual Meeting of the Transportation Research Board, Washington D.C., January 2012. Coelho, M. C., Frey, H. C., Rouphail, N. M., Zhai, H., & Pelkmans, L. (2009). Assessing methods for comparing emissions from gasoline and diesel light-duty vehicles based on microscale measurements. Transportation Research Part D, 14(2), 91 - 99. Coelho, M. C., Andrade, J., Soares, D., Frey, H.C., & Rouphail, N. (2010). A Vehicle Energy Use and Safety Information Support System, Proceedings of the 89th Annual Meeting of the Transportation Research Board, Washington, D.C., January 2010. CORINAIR (2010). EMEP/CORINAIR Emission Inventory Guidebook - 2009. Technical report No 9/2009, EEA. EuroNCAP (2009a). Euro NCAP - For safer cars | comparable Cars. European New Car Assessment Program., Accessed 23rd Nov., 2009. EuroNCAP (2009b). Frontal Impact. European New Car Assessment Program., Accessed 23rd Nov., 2009. Evans, L. (2004). How to make a car lighter and safer. Paper 2004-01-1172, SAE International. Frey, H. C., Zhang, K., & Rouphail, N. (2008). Fuel emissions comparisons for alternative routes, time day, road grade, and vehicles based on in-use measurements. Environmental Science and Technology, 42, 2483 - 2489. Kononen, D., Flannagan, C., & Wang, S. (2011). Identification and validation of a regression model for predicting serious injuries associated with motor vehicle crashes. Journal of Accident Analysis & Prevention, 43(1), 112 - 122. Li, Y., & Bai, Y. (2008). Development of crash-severity-index models for the measurement of work zone risk levels. Journal of Accident Analysis & Prevention. 40(5), 1724 - 1731. SAS Institute Inc. (SAS) (2009). Applied Analytics Using SAS®Enterprise MinerTM 5. Instructor-based training. Cary, NC, USA. ISBN


Torrão, G., Rouphail, N., & Coelho, M. C. (2011). Effect of Vehicle Characteristics on Crash Severity: Portuguese Experience, submitted to the Journal for Safety Research. US EPA (2002). Methodology for Developing Modal Emission Rates for EPA’s Multi-Scale Motor Vehicle & Equipment Emission System. Prepared by North Carolina State University for US Environmental Protection Agency, Ann Arbor. Wenzel, T. (2010). Analysis of the Relationship Between Vehicle Weight/Size and Safety, and Implications for the Federal Fuel Economy Regulation. Final Report prepared for the Office of Energy Efficiency and Renewable Energy, US Department of Energy. LBNL-3143E. March 2010. Wood, D., & Simms, C. (1997). Safety and the car size effect: a fundamental explanation. Journal of Accident Analysis & Prevention, 29(2), 139 - 151. World Health Organization (WHO) (2009a). Global Status Report on Road Safety. Time for Action. 2009., Accessed 22nd Apr., 2010.