EVS28 KINTEX, Korea, May 3-6, 2015 Effect of driving pattern parameters on fuel-economy for conventional and hybrid electric city buses Ming CHI 1, Hewu WANG 1, Minggao OUYANG 1 1 Author 1 State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing,China, chiming0208@163.com (Tsinghua University, State Key Laboratory of Automotive Safety and Energy, Beijing 100084, China) Abstract Driving pattern parameters is of great importance not only to the construction of the driving cycle but also to the energy control strategy for HEV.This study is aimed at investigating which driving pattern parameter has a main effect on the fuel consumption of two kinds of vehicles, conventional vehicle and HEV. The driving data are from Nanjing for the conventional vehicle and Hengyang for the HEV.The data included speed and fuel consumption collected in real traffic. In this paper, 15 traditional driving pattern parameters were calculated. These included average speed, average driving speed, average acceleration, average deceleration, mean length of a driving period, proportion of speed interval and so on. Regression analysis on the relation between driving pattern parameters and the fuel-use showed that the proportion of speed interval and the percentage of idling time have the most impact on the fuel consumption for the traditional car. While, for the HEV the analysis showed that the average speed and the average acceleration have considerable effects on the fuel consumption. At the same time, the result shows the equation of linear regression can be used to predict the fuel-use if the 15 driving pattern parameters are given. Keywords: hybrid electric city buses, fuel-economy, regression analysis, driving pattern parameters 1 Introduction Driving pattern affects the fuel consumption and emissions of vehicles together with other influences.[1, 2] Driving pattern parameters can explain much of the information in driving patterns.[2] Thus the effects of driving pattern parameters on fuel consumption draw considerable attention. The VSP (vehicle specific power) -based approach was used by H. Christopher Frey et al [3]. The influence of key factors, such as: speed, acceleration, and road grade, on the fuel consumption was showed clearly. Ericsson Eva [2] explored 64 driving pattern parameters influence on fuel-use. Nine driving pattern parameters of great impact on the fuel-use was found and the regression analysis Empirically based methods were commonly used in modelling fuel consumption, using driving pattern parameters. Haikun Wang et al [4] developed a VSP-based fuel consumption model. By using the method, he found the sensitivity of the fuel-use of acceleration for a traditional vehicle. While this model cannot analyse several factors at the same time. Michel Andre, Mario Rapone [5] showed the speed and acceleration, constitutes a good basis for emission modelling. This paper told the driving pattern parameters can be utilized to develop an emission estimation approach. EVS28 International Electric Vehicle Symposium and Exhibition 1
In this paper, 15 traditional driving pattern parameters were calculated. These included average speed, average driving speed, average acceleration, average deceleration, mean length of a driving period, proportion of speed interval and so on. Regression analysis on the relation between driving pattern parameters and the fueluse showed that the proportion of speed interval and the percentage of idling time have the most impact on the fuel consumption for the traditional car. While, for the HEV the analysis showed that the average speed and the average acceleration have considerable effects on the fuel consumption. At the same time, the result shows the equation of linear regression can be used to predict the fuel-use if the 15 driving pattern parameters are given. 2 Methodology 2.1 Driving pattern parameters The driving data we used were collected from the real traffic. For the traditional vehicle from Nanjing, 5 days of driving were collected using 5 buses and 1 days of driving for HEV from Hengyang. Not only speed and fuel-consumption were included in the data, but also longitude and latitude, which were of great use. Table 1 driving pattern parameters Driving pattern Denotation parameters average speed v m Average driving speed v mr maximum speed v max average acceleration a m(over 0.1m/s 2 ) average deceleration d m(under-0.1m/s 2 ) average time of c microtrips average lengths of l microtrips acceleration root RMS mean square % of idling time p i % of accelerating time p a % of decelerating time p d % of constant time p c % of time in speed p 0-20 interval 0_20 % of time in speed p 20-40 interval 20_40 % of time in speed p 40-60 interval 40_60 The data were all from driving of buses, which had fixed routes every day. According to the specificity of buses, the data were divided into many microtrips, depending on the longitude and latitude. In every microtrip, the buses drove from a fixed stop to another and stayed for over 5 minutes. In this paper, 15 traditional driving pattern parameters were calculated for each microtrip. All the parameters calculated are given in Table 1. Each of the driving pattern parameters contains some information about traffic or vehicles. And each can reflect different aspects of traffic. 15 driving pattern parameters are used, so most of the different characteristics of driving patterns are covered. 2.2 The regression analysis Regression analysis is widely used for prediction and forecasting. To investigate which driving pattern parameters had important influences on fuel consumption, a regression analysis was performed with fuel consumption as a dependent variable and 13 driving pattern parameters as independent variables. 2 parameters were excluded to avoid high degree of linear dependence within the independent variables. As is known, p i is of importance for conventional vehicles. Therefore, percentage of constant time was removed in the analysis. For percentage of time in different speed intervals, the percentage of time in speed interval 40_60 accounts for the least percentage and compared with other intervals, the percentage is relatively less. Thus, the percentage of time in speed interval 40_60 was removed. The following is the model of the regression analysis: y 0 1x1... nxn (1) As the dependent variable, fuel consumption is y in the above equation. The n above is 13. x... 1 xn are representing the 13 driving pattern parameters as independent variables. 0, 1... n are unknown parameters. is random error, following Gaussian distributions. And also should be independent characters. Those are the hypothesis of the model. During the regression, the number of measurements, N, should be larger than the number of unknown parameters, n. That is also one of the reasons why we divided the data into many microtrips. The larger N is, the better accuracy we get. EVS28 International Electric Vehicle Symposium and Exhibition 2
The regression analysis was performed using the software SPSS. In the results of regression analysis, some indexes can be obtained, which can tell the accuracy of the model and also can tell if the hypothesis of the model is true. One is R 2, representing the accuracy, and the other one is sig, for the truth of the hypothesis. 3 Driving pattern parameters in two cities Different driving pattern parameters can tell the different traffic traditions between the two cities and also the different performance between the two vehicles. Because of the different traffic conditions and technologies of the vehicles, fuel consumption in these cities is different. The fuel consumption of the conventional vehicle is 42.42L per 100 kilometres and the HEV is 32.36L per 100 kilometres. We can see the traditional one use almost 32.36% more than HEV. As we can see in Figure 1, average speed and average driving speed are almost the same. However, maximum speed in Nanjing is much higher than Hengyang. This difference may not be caused by a number of reasons. First the traffic condition may differ greatly between different time. When the condition is good, the speed can be up to a high level. However, for Hengyang the condition may be almost the same. The average speed for Hengyang is 11.3% higher than Nanjing which tells the better traffic condition. Second the conventional vehicle used in Nanjing may have better performance than the HEV used in Hengyang. The next driving pattern parameters are acceleration and declaration, which represent the variation of the speed. Figure 2 acceleration and declaration in Nanjing and Hengyang As showed in Figure 2, in both cities, average deceleration is larger than average acceleration. The reason is about the specificity of buses. During the driving, buses must stop and start frequently. While starting vehicles, drivers should be careful of people around, so the acceleration is smaller. While parking, drivers should get to the locations, so the deceleration is larger. Average acceleration and average deceleration in Hengyang are both higher than Nanjing. This reflects during the driving of the buses in Hengyang, quick acceleration and deceleration are needed, which tells the distance between the stops in Hengyang is shorter. Figure 3 percentage of time in different states Figure 1 speed of Nanjing and Hengyang Figure 3 tells the percentage of time in different states for buses. Idling time in Nanjing, which is of great importance, makes up a high proportion. This accounts for the high fuel consumption of the traditional vehicle. In Hengyang, idling time is much lower, which means better traffic condition. As for the percentage of accelerating time and decelerating time, in both cities, accelerating time makes up a higher proportion. Combined with the EVS28 International Electric Vehicle Symposium and Exhibition 3
result showed in Figure 2,these show a good driving behaviour of bus drivers. Figure 4 percentage of time in different speed intervals Speed intervals are also of great importance for fuel consumption, especially for traditional vehicles. That is because of the influence of working conditions for the engine is significant. Working in a low speed interval means a poor efficiency, and causes higher fuel consumption. While working in a low speed interval, HEV can use the motor instead of the engine. Its sensitivity of efficiency to speed intervals is much lower. Motors can maintain a high efficiency no matter which speed interval they are in. At the same time, because of the importance of the low speed intervals and the proportion, during the regression analysis, the percentage if time in speed interval 40_60 was removed. For buses in cities, because of the limitation of speed and the specialty of frequent starting and stopping, the percentage of time in speed interval 0_20 makes up a high proportion. As is shown in Figure 4, the percentage of time in speed interval 0_20 in both cities is much higher, especially in Nanjing. The much higher percentage means the worse traffic and worse fuel consumption, particularly for conventional vehicles. As is shown above, the traffic condition for Nanjing is worse than Hengyang, which is influenced by the population of the cities, the population of vehicles or some other influences. At the same time, traffic condition is an important influence on fuel consumption for vehicles. Therefore the necessary for Nanjing to use HEV or EV instead of traditional vehicles is significant. In this way, fuel consumption and emissions will be better, which is of great importance for the future. 4 The effect of driving pattern parameters on CV and HEV 4.1 Results of the regression analysis A model for the prediction of fuel consumption was derived from regression analysis. As described in Table 2, the result of CV is better than HEV. All of the results are acceptable. For the values of sig., both are less than 0.05. This means the model for HEV and CV is effective. The method of regression analysis can be used here. With regard to R 2, both of the two values are larger than 0.6, which means the prediction model is acceptable. The method of regression analysis for fuel-use prediction with driving pattern parameters is useful. Table 2 results of regression analysis CV HEV Sig. 0.000 0.005 R 2 0.746 0.634 4.2 Driving pattern parameters effects on conventional vehicle As described in Table 3, driving pattern parameters are ordered by their effects on conventional vehicle. The effects are graded according to the absolute values of standardised B in the regression analysis. There are four grades here. Standardised B gives the size of the effect (in S.D.s), if the driving pattern parameter changes by one S.D. Table 3 Driving pattern parameters effects on CV Driving pattern parameters Effects % of idling time ++++ average speed ++++ % of time in speed interval 0_20 ++++ average acceleration +++ acceleration root mean square +++ average driving speed ++ % of accelerating time ++ average lengths of microtrips ++ average deceleration + average time of microtrips + maximum speed + % of decelerating time + % of time in speed interval 20_40 + EVS28 International Electric Vehicle Symposium and Exhibition 4
In the Table 3, it is found that, the driving pattern parameters: the percentage of idling time, average speed and the percentage of time in speed interval 0_20 are of great importance for CV. The result shows that idling is one of the most effective influences for CV. While idling, the engine is still working, meaning the waste of fuel-use. While working in a low speed interval, engine has poor efficiency, causing higher fuel consumption. Average speed means the energy a vehicle needs. So v m is also influential. The driving pattern factor of acceleration is also of great importance for CV. The result is supported by an earlier study. 26 parameters were investigated, and the factor of stop and percentage of time at speed under 15 was found influential. [6] Parameters about microtrips are less important. Percentage of time in speed interval 20_40, percentage of accelerating time is more influential than decelerating time. The values of accelerating and decelerating are more important than the percentages. 4.3 Driving pattern parameters effects on HEV For HEV, because of the difference of the controlling and working, the regression result was different. As is shown in Table 4, the factor of speed is of great influence for HEV. Average speed, maximum speed and % of accelerating time are influential elements. Table 4: Driving pattern parameters effects on HEV Driving pattern parameters Effects average speed ++++ maximum speed ++++ % of accelerating time ++++ average acceleration +++ acceleration root mean square +++ average driving time +++ % of decelerating time +++ average time of microtrips ++ average lengths of microtrips ++ % of idling time ++ % of time in speed interval 20_40 + % of time in speed interval 0_20 + average deceleration + It is showed that the energy HEV needs is most important. This also means the efficacy for HEV is much higher. The driving pattern factor of acceleration is also of great importance for HEV which is almost the same as CV. Percentages of time in different speed intervals are less important. While the percentages of accelerating and decelerating are much more influential. The values of accelerating and decelerating are less influential than the percentages of accelerating and decelerating. Contrary to CV, it is shown that the percentage of idling time is at the end, which means the least influential. 4.4 Comparisons of Driving pattern parameters effects on CV and HEV As is described above, effects of driving pattern parameters on CV and HEV are different. First of all, the importance of percentage of idling time is different. For CV, while idling, engine is still working, causing the waste of fuel use. But for HEV, because of the controlling, while idling engine can be off so that the influence is much less. The reason for percentage of time in speed interval 0_20 is the same. As for the driving pattern factor of acceleration, we can see the influence is almost the same. While for deceleration, effects for CV are more than HEV. This is because batteries and motor in HEV can recover the energy during the time of decelerating, which also means less fuel consumption for HEV. 5 Conclusions and Discussions (1) Traffic in Hengyang is better than Nanjing, which cause less fuel consumption for vehicles in Hengyang. A higher percentage of idling time, lower average speed and higher percentage of time in speed interval 0_20 reflect the worse traffic in Nanjing. The necessary for Nanjing to use HEV or EV instead of traditional vehicles is significant. In this way, fuel consumption and emissions will be better, which is of great importance for the future. (2) Rough estimation can be done by the regression analysis. As is shown above, both values of R 2 for HEV and traditional vehicles are enough. The estimation was proved to be available. (3) Because of the different operating principle, effects of driving pattern parameters on conventional vehicles and HEV are different. For traditional vehicles, the percentage of idling time, average speed and the percentage of time in speed interval 0_20 are of great importance. While EVS28 International Electric Vehicle Symposium and Exhibition 5
for HEV, average speed, maximum speed and % of accelerating time take account. With more data, the regression analysis of HEV can be much more accurate. And the regression can be used to analyse the influence of other kind of vehicles, such as electric vehicles. With the most effective parameters, constructing driving cycles, designing of the vehicles and so on can be done better. For much more accurate estimation, another model can be used. Acknowledgments The authors gratefully acknowledge the support of this research by the State Key Laboratory of Automotive Energy and Safety Center of Chinese Automotive Energy in Tsinghua University. They thank Mr.Tang from CSR, who provided the driving data of HEV buses. References: [1]E. Ericsson, "Driving pattern in urban areasdescriptive analysis and initial prediction model," Bulletin 185/3000, 2000-01-01 2000. [2]E. Ericsson, "Independent driving pattern factors and their influence on fuel-use and exhaust emission factors," Transportation Research Part D: Transport and Environment, vol. 6, pp. 325-345, 2001-01-01 2001. [3]H. C. Frey, N. M. Rouphail, H. Zhai, T. L. Farias, and G. A. Gonçalves, "Comparing real-world fuel consumption for diesel-and hydrogen-fueled transit buses and implication for emissions," Transportation Research Part D: Transport and Environment, vol. 12, pp. 281-291, 2007-01-01 2007. [4]H. Wang, L. Fu, Y. Zhou, and H. Li, "Modelling of the fuel consumption for passenger cars regarding driving characteristics," Transportation Research Part D: Transport and Environment, vol. 13, pp. 479-482, 2008-01-01 2008. [5]M. André and M. Rapone, "Analysis and modelling of the pollutant emissions from European cars regarding the driving characteristics and test cycles," Atmospheric Environment, vol. 43, pp. 986-995, 2009-01-01 2009. [6]E. Ericsson, "The relation between vehicular fuel consumption and exhaust emission and the characteristics of driving patterns,", 2000, pp. 137-147. Authors Ming CHI received her B.E. Degree in automotive engineering from Tsinghua University, Beijing, China. She is now pursuing M.E. Degree in automotive engineering in Tsinghua University, China. Her research interests include driving cycle and driving pattern. Hewu WANG is the Associate Professor and Deputy Director of the China-US Clean Vehicle Consortium from Tsinghua University, China. His research interests include the alternative fuels, industrialization and demonstrations of electric vehicle. 2 nd Author s biography Minggao OUYANG, is the Professor from Tsinghua University, China. He is Directing the State Key Laboratory of the Automotive Safety and Energy, Tsinghua University, China. EVS28 International Electric Vehicle Symposium and Exhibition 6