Transportation Research Part D 41 (2015) 377–385
Contents lists available at ScienceDirect
Transportation Research Part D journal homepage: www.elsevier.com/locate/trd
Development of two driving cycles for utility vehicles P. Seers a,⇑, G. Nachin b, M. Glaus b a b
TFT lab., École de technologie supérieure, 1100 Notre-Dame W., Montréal H3C 1K3, Canada STEPPE, École de technologie supérieure, 1100 Notre-Dame W., Montréal H3C 1K3, Canada
a r t i c l e
i n f o
Article history: Available online 6 November 2015 Keywords: Driving cycle Driving characteristics Airport Suburban Utility vehicle
a b s t r a c t Driving cycles are used to assess vehicle fuel consumption and pollutant emissions. The premise in this article is that suburban road-work vehicles and airport vehicles operate under particular conditions that are not taken into account by conventional driving cycles. Thus, experimental data were acquired from two pickup trucks representing both vehicle fleets that were equipped with a data logger. Based on experimental data, the suburban road-work vehicle showed a mixed driving behavior of high and low speed with occasional long periods of idling. In the airport environment, however, the driving conditions were restricted to airport grounds but were characterized by many accelerations and few high speeds. Based on these measurements, microtrips were defined and two driving cycles proposed. Fuel consumption and pollutant emissions were then measured for both cycles and compared to the FTP-75 and HWFCT cycles, which revealed a major difference: at least a 31% increase in fuel consumption over FTP-75. This increased fuel consumption translates into higher pollutant emissions. When CO2 equivalent emissions are taken into account, the proposed cycles show an increase of at least 31% over FTP-75 and illustrate the importance of quantifying fleet speed patterns to assess CO2 equivalent emissions so that the fleet manager can determine potential gains in energy or increased pollutant emissions. Ó 2015 Elsevier Ltd. All rights reserved.
Introduction A wide variety of standard driving cycles are used to implement state emission regulations. The US FTP-75 and SFTP procedures, the soon-to-be-replaced European NEDC series, and the new Japanese JC08 cycle are among the best known. Standard driving cycles, however, are generic and may not be consistent with local network specifics, such as topography, topology, traffic, and driver behavior. When accurate small-scale emission rates have to be determined, specific driving cycles are developed to take into account local and particular conditions that may significantly influence emission levels. Many specific driving cycles have been proposed recently for Hong Kong (Hung et al., 2007), Pune (Kamble et al., 2009), Bangkok (Tamsanya et al., 2009), Mexico City (Schifter et al., 2005), Edinburgh (Saleh et al., 2009), just to name a few. Their singularities are proof of the diversity of urban road networks and driving behavior. Since vehicle fuel consumption and emissions are influenced by operating conditions (Wang et al., 2008; Booth et al., 2001), specific driving cycles are needed to sketch a portrait of fuel consumption and atmospheric emissions caused by fleets of vehicles. Furthermore, specific driving cycles may help in assessing a fleet’s impact on air quality and greenhouse-gas emissions, if modeled (Grieshop et al., 2012), or to assess well-to-tank energy consumption, as was recently proposed (Ma et al., 2011).
⇑ Corresponding author. http://dx.doi.org/10.1016/j.trd.2015.10.013 1361-9209/Ó 2015 Elsevier Ltd. All rights reserved.
378
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
This paper originated out of the need to assess the fuel consumption and CO2 equivalent emissions of two different fleets of pickup trucks. The first fleet—owned by the town of Salaberry-de-Valleyfield (SdV)—is used to transport workers and equipment in a suburban environment in Montréal, Canada. The second fleet consists of pickup trucks, but their use is confined exclusively to within the limits of Pierre Elliot Trudeau International Airport (PETIA), located in Montréal, Canada. It was determined that assessing the respective fleet fuel consumption and CO2 emissions meant acquiring a driving pattern representative of each fleet, which would then be compared to known driving cycles. The main objective of this paper is to use this comparison to propose two driving cycles for estimating the fuel consumption and emissions of both fleets. The paper is constructed as follows: First, the methodology used to acquire experimental data and its post-treatment are defined. The main driving characteristics of each fleet of vehicle are analyzed and compared to known driving cycles. Lastly, the proposed driving cycles are presented and discussed; the fuel consumption and pollutant emission results are also presented. Methodology A pickup truck (Ford F-150, 2010) belonging to Salaberry-de-Valleyfield (SdV) and one from the Pierre-Elliot Trudeau International airport (PETIA) were instrumented in order to determine driving characteristics and develop driving cycles. The PETIA vehicle is operated within the confines of the airport grounds not connected to public roads. The SdV vehicle is used to carry workers and equipment around Salaberry-de-Valleyfield, mostly in suburban areas and occasionally on highways. Both vehicles serve for specific applications that entail non-standard driving behaviors, which is the premise of this study. A complete methodology was designed to develop a driving cycle specific to each use. The three main steps of this methodology are presented briefly below. Data collection In the literature, car-chasing (see Hung et al., 2005, for example) and on-board (Saleh et al., 2009; Tamsanya et al., 2009 to name a few) measurements are used to collect vehicle speed over time. Herein on-board data acquisition was used for eight weeks on the PETIA vehicle and for six weeks on the SdV vehicle. An ISAAC DRU-800 data logger was connected to the vehicle’s OBD-II port and was configured to record instant wheel speed at a rate of 1 Hz whenever the engine was running. The data logger was installed and acquisition initiated without the driver’s knowledge. This element, combined with the relatively long acquisition period and the fact that the drivers were under no particular restrictions, made it possible to capture true driver behavior. At the end of the acquisition period, the data were downloaded to a computer for analysis. Data processing After collection, the data were transferred and posttreated in MATLAB R2009b with code developed in-house. The first step in the data posttreatment was characterizing and calculating the vehicles’ different indicators, such as average speed, maximum speed, and acceleration, as presented in Table 1. These indicators are similar to those used by Saleh et al. (2009), Hung et al. (2007), and Lin and Niemeier (2003). In Table 1, the terms v and a, respectively, refer to vehicle instant speed and acceleration rates. Herein, the vehicle is considered to be accelerating when its speed is greater than 5 km/h such as Hung et al. (2007) and Amirjamshidi and Roorda (2015) and its a is higher than 1 m/s2 as to obtain the same acceleration resolution than Hung et al. (2007). A set of five indicators focus on the distribution of five operating modes, namely, idling, accelerating, decelerating, cruising, and creeping. Positive kinetic energy (PKE) was used as a measure of aggressiveness (Lin and Niemeier, 2003). PKE expresses an average acceleration value; the average work performed by the engine per distance covered to provide
Table 1 Indicators for road-behavior characterization. Indicator
Definition
Average speed Maximum speed
Total covered distance/total recorded time Maximum recorded speed
Operating-mode distribution Idling Accelerating Decelerating Cruising Creeping Positive kinetic energy (PKE) SAFD matrix
v=0 a > 1 m/s2 v > 5 km/h a < 1 m/s2 v > 5 km/h a 2 [1; 1] m/s2 v > 5 km/h v 2 ]0; 5] km/h P 2 2 PKE ¼ D1 n1 8 v iþ1 > v i i¼1 v iþ1 v i Matrix of speed–acceleration frequency distribution with 5 km/h and 1 m/s2 increments, respectively
379
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
the vehicle’s kinetic energy (Watson et al., 1982). Lastly, the speed–acceleration frequency distribution matrix (SAFD) stores the speed–acceleration frequency distribution of the data (Nesamani and Subramanian, 2006). Cycle construction Data collected from the instrumented vehicles were recorded as trips, or speed traces that started when the engine was turned on and terminated when it was turned off. During data posttreatment, these trips were split into smaller microtrips for use in the assembly process. A microtrip is defined as any speed trace between any of the vehicle’s two consecutive stops (v = 0 km/h) and could be qualified as elementary blocks for cycle building. Idling events are not included in the microtrips, as in Tamsanya et al. (2009). In order to implement realistic idling times at the beginning and end of a driving cycle, average start idling and shutdown idling times were also calculated from the collected data. The cycle building process is iterative, since a large number of candidate cycles were built from randomly selected microtrips, similarly to the procedure used by Tamsanya et al. (2009) and Kamble et al. (2009). Thus, the procedure starts with a randomly selected initial microtrip. It is appended with a start idle time at its beginning, as well as another idle phase that extends the speed trace’s duration to the expected 1200 s. Both idle phases allow the driving cycles to carry the same average start and shutdown idling times, as in real-life operations. This combination constitutes an initial candidate cycle. Other microtrips were randomly selected and appended to the previous structures until a 1200-s cycle is defined including a shutdown idling event. The 1200-s duration of the driving cycles to be built in this study was dictated by the capacity of the sampling bags used during FTP emission measurements performed by a government agency. The cycle offering the best representative index (Dcycle) was kept as the normalized cycle. Therefore, for each candidate cycle, a Dcycle is calculated, making it possible to identify the one that best matches the parametric target defined by the vehicle’s actual road behavior. Dcycle is defined as:
Dcycle ¼
n X jhi ei j þ dSAFD minðh i ; ei Þ i¼1
ð1Þ
In Eq. (1), hi ei characterizes the difference between the proposed cycle and collected data with respect to the ith of the n indicators defined in Table 1. dSAFD quantifies the gap between the SAFD matrix of collected data and the SAFD matrix from the candidate cycle, calculated by a sum of squared differences between both matrices’ terms. The best possible match between a candidate cycle and real road behavior is achieved when Dcycle = 0. Therefore, the iterative process will return a cycle with the lowest Dcycle. In the proposed method, each of the n indicators possesses the same weight in the sum of the error. Unlike other techniques in the literature, however, Dcycle also takes into account the discrepancy between the SAFD matrix of the experimental data and that of the generated driving cycle. Results and analysis At the PETIA, six weeks of collection amounted to 28 h 25 min of data, which is comparable to the more than 26 h of data collected by Hung et al. (2005) for Zhuhai or the 34.6 h of data collected by Hung et al. (2007) for a Hong Kong cycle. In the case of SdV, 129 h27 min of recordings were collected over a period of eight weeks. Table 2 presents the main characteristics of both sites, along with a comparison to standard driving cycles in North America: UDDS is a cycle with city driving conditions, HWFET features highway driving at speed below 97 km/h (60 miles per hour), US06 is an aggressive driving cycle, SC03 is a supplemental test procedure with air conditioning and NYCC represents stop-and-go traffic conditions at low speed (U.S. Environmental Protection Agencya). The behavior characteristics of both sites differ significantly from that of North America standard cycles. In particular, the operating-mode distribution shows a much stronger tendency for idle. Moreover, there is a massive dissonance between high maximal speed and low average speeds. The behavior of the PETIA vehicle is particularly aggressive, as shown by
Table 2 Comparison between experimental data and driving cycles. Indicators
PETIA
SdV
UDDS urban
HWFET highway
US06 highway
SC03 urban
NYCC urban
Data collected Distance covered Avg. speed (km h1) Top speed (km h1)
28 h250 614 km 21.6 127.5
129 h270 2256 km 17.4 130.4
– – 31.5 91.2
– – 77.6 96.4
– – 77.2 129.2
– – 34.5 88.2
– – 11.4 44.6
Operating-mode distribution Idling Accelerating Decelerating Cruising Creeping
31.1% 20.9% 17.6% 28.1% 2.3%
43.6% 13.7% 12.1% 26.6% 4.0%
18.9% 21.0% 17.2% 42.1% 0.7%
0.8% 8.2% 8.4% 82.6% 0.0%
7.5% 22.0% 23.3% 46.1% 1.1%
19.5% 21.6% 18.6% 39.1% 1.2%
35.0% 19.9% 20.2% 17.9% 7.0%
PKE (m s2)
0.55
0.37
0.35
0.14
0.42
0.41
0.62
380
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
the high PKE index. Besides, both the PETIA and SdV vehicles are somewhat reminiscent of the NYCC in terms of operatingmode distribution and slow average speed. The PETIA vehicle is also comparable to NYCC as far as PKE index is concerned, while the SdV vehicle is more similar to the US06 highway and SC03 urban driving cycles for the same indicator. The high top speeds of both sites are, however, more similar to the US06 highway cycle, which is at the opposite end of the road-behavior spectrum. When comparing PETIA to SdV, it is noted that SdV has more idling time than PETIA due to the nature of the service provided by each fleet while the latter has more acceleration and deceleration. Figs. 1 and 2 illustrate the SAFD matrix for the PETIA and SdV data respectively, in which the probability, on a logarithm scale, is presented as a function of vehicle speed and acceleration to illustrate the kind of driving behavior. It can be observed when comparing Figs. 1 and 2 that SdV data show higher maximum speed and a greater area of acceleration above 10 km/h/s for speed below 50 km/h. On the other hand PETIA presents higher deceleration between 25 km/h and 90 km/h than SdV. Thus, the pickup trucks involved in this study show very particular trends that justify the need to develop specific driving cycles. In preparation for the driving-cycle assembly, all of the trips collected onboard the PETIA and SdV vehicles were split into microtrips, as defined previously. Table 3 shows the number of trips and microtrips available for each site, as well as some statistical properties about their lengths. Both sets of microtrips have similar breakdowns of duration up to the upper decile as well as close average durations. The longest SdV microtrips were twice as long as that of the PETIA vehicle. The durations of the short and medium-length microtrips followed a similar pattern in both cases, with longer microtrips (within the last decile) differing significantly. According to Table 2, the SdV’s pickup truck spent 43.6% of its time idling. This means that a 1200-s driving cycle should theoretically incorporate 523 s of idling and 677 s of motion. The two longest microtrips for the vehicle (see Table 3) are longer than the duration of the cycle to be generated. As such, they were rejected by the assembly algorithm. Finally, the average number of stops per kilometer of travel in the experimental data is 1.8 for PETIA and 1.9 for SdV which are similar to the 2.4 stops per kilometer recorded for Bangkok (Tamsanya et al., 2009). Pierre-Elliot Trudeau Airport (PETIA) Table 4 presents the PETIA vehicle’s driving-cycle characteristics and shows a Dcycle of 0.149. Of all the indicators used to measure the Dcycle of the driving cycles, only vehicle top speed differed by more than 5% as can be seen in Table 4. There is a close match between the distribution of the operating modes in the cycle and in real life: the highest absolute and relative gaps are, respectively, 1.3% and 4.6% (cruising). Moreover, the proposed driving cycle manages to reproduce the aggressive behaviors seen in real life, as shown by the PKE index that only differs by 0.4%. The proposed driving cycle matches the original data’s characteristics, from general tendencies—average speed and operating modes—to extreme events such as top speed. More particularly, the average time spent in acceleration, idle and deceleration are within less than 1.1% of the experimental data. One difference is observed for creeping for which the proposed cycle has a shorter proportion of time than the experimental data due to the cycle building process that identifies the cycle having the best representative index as defined by Eq. (1). Fig. 3 presents the PETIA’s vehicle’s driving cycle, which will be used for emission and fuel-consumption measurements.
110
1.49
100
1
90 0.5 80 −1
Speed (km.h )
0 70 −0.5
60 50
−1
40
−1.5
30
−2
20 −2.5 10 −3 −15
−10
−5
0
5
10
15
Acceleration (km.h−1.s−1) Fig. 1. Probability on a log scale of the vehicle’s acceleration as a function of vehicle speed at the PETIA site.
381
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
125
1.5 1
100
0.5
Speed (km.h−1)
0 75
−0.5 −1 −1.5
50
−2 −2.5
25
−3 −3.5 0 −15
−10
−5
0
5
10
15
−1 −1
Acceleration (km.h .s ) Fig. 2. Probablility on a log scale of the vehicle’s acceleration as a function of vehicle speed at the SdV site.
Table 3 Microtrip characterization.
Number of trips Number of microtrips Average duration Lower quartile Median Upper quartile Upper decile Length of the two Longest microtrips
PETIA
SdV
185 1087 66 s 8s 27 s 84 s 178 s 770 s 773 s
499 4333 62 s 9s 20 s 83 s 167 s 1261 s 1506 s
Table 4 Comparison between PETIA experimental data and driving cycle. Indicators
Vehicle
Cycle
Time Distance (km) Avg. speed (km/h) Top speed (km/h) Idle Acceleration Deceleration Creeping Cruising PKE (m/s2)
28 h 25 min 614 21.6 127.5 31.1% 20.9% 17.6% 2.3% 28.1% 0.554
20 min 7.4 22.2 120.6 31.2% 20.9% 17.8% 0.8% 29.4% 0.556
City of Salaberry-de-Valleyfield (SdV) While the SdV cycle presented in Fig. 4 has a higher Dcycle of 0.486, as can be seen in Table 5, there is a very good match with respect to operating-mode distribution, with differences of between 0.2% for idle and 2.2% for acceleration, which are better than the difference reported by Hung et al. (2007) or Manuel et al. (2006). The driving cycle also does a good job of reproducing the main indicators such as average vehicle speed, with a difference below 2% between the proposed driving
382
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
140 120
Speed (km/h)
100 80 60 40 20 0
0
200
400
600
800
1000
1200
1000
1200
Times (s) Fig. 3. PETIA driving cycle.
120 100
Speed (km/h)
80 60 40 20 0
0
200
400
600
800
Time (s) Fig. 4. Salaberry-de-Valleyfield driving cycle.
Table 5 Comparison between SdV experimental data and driving cycle. Indicators
Vehicle
Cycle
Time Distance Avg. speed (km/h) Top speed (km/h) Idle Acceleration Deceleration Creeping Cruising PKE (m/s2)
129 h 27 min 2256 17.4 130.4 43.6% 13.7% 12.1% 4.0% 26.6% 0.368
20 min 5.7 17.1 100.7 43.5% 13.4% 12.1% 4.2% 26.8% 0.417
cycle and the experimental data. A 10.8% difference in PKE is observed in Table 5 between the proposed cycle and the experimental data which is comparable to the difference reported for the same indicator by Hung et al. (2007) (8.4%) or the 20% reported by Han et al. (2012). The greatest difference between the proposed cycle and the experimental results is related to top speed: a difference of 22.8%, which is similar to the difference reported by Hung et al. (2007) for the suburban cycle. This difference between the experimental data and the driving cycle is attributed to a single event among the 4333 microtrips during which this top speed was attained. Further data analysis has shown that only 4 microtrips reached a top speed over
383
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
110 km/h and that these microtrips were too long to be selected by the assembly algorithm. Nevertheless, the top speed reached in the driving cycle is still within the 98th percentile of the fastest speeds among the experimental data and, as such, the cycle is deemed representative of driver behavior. Fig. 4 illustrates the cycle’s speed over time. Discussion As mentioned previously, experimental data from both sites show important idling times and aggressive driving that differ very significantly from the behaviors in standard cycles used in North America. These specificities could not be taken into account by existing standard cycles for emission measurements. Table 6 shows how the experimental data from the proposed driving cycle differs from standard cycles, as quantified by Dcycle (defined above) and showing the uniqueness of the driving patterns compared to standard cycles. The higher the representative index the higher is the difference between the proposed driving cycles and the driving cycle of comparison. Hence, the PETIA driving characteristics are far from the HWFET characteristics. Furthermore both proposed cycles are reminiscent to the UDDS but differ greatly with respect to driving cycle average speed and percentage of time in cruising and idling. Lastly, as the main objective of the paper is to evaluate fuel consumption and pollutant emissions associated to the driving pattern of each fleet the pickup trucks were tested over both proposed cycles and along the HWFET and FTP-75, as they are North American reference cycles for measuring fuel consumption and pollutant emissions following the U.S. Code of Federal Regulations (US CFR) Title 40 Part 86 (U.S. Environmental Protection Agencyb). Table 7 compares carbon monoxide (CO), nitrogen oxide (NOx), total hydrocarbons (THC), fuel consumption, and CO2 equivalent emissions. The aggressiveness of the proposed cycles translates into fuel consumption that is twice that of highway driving (HWFCT) as well as 31% and 35% higher for the PETIA and SdV cycles, respectively, compared to FTP-75. The proposed cycles show major differences when pollutant emissions are concerned. Indeed, the same pickup truck was used for the 4 cycles presented in Table 7 and the more aggressive proposed cycles, such as for the PETIA, translate into NOx emissions an order higher than for that obtained for the FTP-75 or HWFCT. CO and THC emissions are also significantly higher when comparing both proposed cycles together. CO and NOx emissions are nearly twice as high for the PETIA cycle compared to the SdV cycle, while the THC values are nearly twice higher for the SdV cycle, while both fuel consumptions are similar. Finally, the last comparison for Table 7 is with respect to CO2 equivalent emission over the respective cycles. The experimental data show that the fleet vehicles from the proposed cycle emit at least 31% more CO2 equivalent when PETIA is compared to the FTP-75. A slightly higher difference is observed if SdV is used, which illustrates the importance of driving-cycle definition in predicting a fleet’s pollutant emissions or in estimating fleet CO2 emissions. Fuel consumption and pollutant emissions have been strongly related to the microtrips PKE and average speed (Watson, 1995) and fuel consumption of vehicle fed with gasoline, propane or diesel has also been strongly linked to average speed of microtrips (Zhang et al., 2014). Thus, based on these reported observations, the results of Table 7 were plotted against the PKE and average vehicle speed of each driving cycle. The data correlate more strongly with the driving cycle average speed and a such only these results are presented in Fig. 5 for CO2 and fuel consumption as they show the highest adjusted coefficient of determination (R2) of 0.98 and 0.97 respectively using power law with a 95% prediction bounds as Zhang et al.
Table 6 Representative index between experimental data and driving cycles. Driving cycle
Representative index Dcycle PETIA
SdV
PETIA SdV
0.149 –
– 0.486
UDDS HWFET US06 SC03 NYCC
2.64 49.38 7.16 2.5 3.77
4.21 7.16 10.83 4.48 4.99
Table 7 Emissions and fuel-consumption comparison. Pollutants
FTP-75
HWFCT
PETIA
SdV
CO (g/km) NOx (g/km) THC (g/km) Fuel consumption (L/100 km) CO2 equivalent (g/km) Cycle average speed (km/h)
0.814 0.019 0.037 15.69 360.4 34.1
0.23 0.006 0.00 10.13 233 77.6
4.822 .217 0.062 20.57 472 21.6
2.063 0.087 0.11 21.20 493 17.4
384
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385 3
CO2 emissions (g/km) / fuel consumption (l/100 km)
10
Fuel consumption
2
10
CO2 Fitted curve fuel cons. Fitted curve CO2
1
10
10
20
30
40
50
60
70
80
Driving cycle average speed (km/h) Fig. 5. Pollutant emissions and fuel consumption as a function of driving cycle average speed (data from Table 7).
(2014). Because THC emissions were below the exhaust gas analyser lower limit for the HWFCT cycle, only 3 measurements could be used but they also show a strong correlation with driving cycle average speed but the results are not shown due to the small amount of experimental measurements. On the other hand, CO and NOx poorly correlate with both cycle characteristics for the vehicle tested herein. These results suggest that for a given vehicle and fuel, knowing its driving cycle average speed could allow predicting CO2 emissions and fuel consumption, however, due to the limit number of cycle tested herein, more results for the same vehicle over other driving cycles are needed to confirm this trend. Conclusion This paper presents two different driving cycles based on the measurement of pickup truck speed over time based on onboard data logging that characterized each fleet vehicle’s speed behavior. Based on the experimental data, a methodology for assembling specific driving cycles was presented and applied. Initial characterization of the experimental data showed the uniqueness of each fleet with high PKE, high idling, and high maximum vehicle speed, confirming that standard driving cycles would not be representative of their driving patterns. Therefore, specific cycles were proposed based on the experimental data and the proposed cycles were then implemented in standard dynamometer measurements. The aggressive driving characterizing both of the proposed cycles yield higher fuel consumption and pollutant emissions than conventional driving cycles. The results for a pick-up truck evaluated over 4 different driving cycles have shown that there exists a strong relationship between THC, CO2, fuel consumption and the driving cycle average speed. Based on the results estimations of fleet’s CO2 equivalent emissions are now possible allowing decision makers evaluating different vehicle’s technologies or fuels to reduce this pollutant. Finally, governmental environmental agencies are now better tool to estimate fuel consumption and pollutant emissions from municipal and airport pick-up truck fleets. Acknowledgement The authors would like to express their gratitude to Stephane Deschamps of Budget Propane. References Amirjamshidi, G., Roorda, M.J., 2015. Development of simulated driving cycles for light, medium and heavy duty trucks: case of the Toronto waterfront area. Transport. Res. Part D 34, 255–266. Booth, A.E., Muneer, T., Kirby, J., Kubie, J., Hunter, J., 2001. The measurement of vehicular driving cycle within the city of Edinburgh. Transport. Res. Part D 6, 209–220. Grieshop, A.P., Boland, D., Reynolds, C.C.O., Gouge, B., Apte, J.S., Rogak, S.N., Kandlikar, M., 2012. Modeling air pollutant emissions from Indian autorickshaws: model development and implications for fleet emission rate estimates. Atmos. Environ. 50, 148–156. Han, D.S., Choi, N.W., Cho, S.L., Yang, J.S., Kim, K.S., Yoo, W.S., Jeon, C.H., 2012. Characterization of driving patterns and development of a driving cycle in a military area. Transport. Res. Part D – Transp. Environ. 17 (7), 519–524.
P. Seers et al. / Transportation Research Part D 41 (2015) 377–385
385
Hung, W.-T., Tam, K.-M., Lee, C.-P., Chan, L.-Y., Cheung, C.-S., 2005. Comparison of driving characteristics in cities of Pearl River Delta, China. Atmos. Environ. 39, 615–625. Hung, W.T., Tong, H.Y., Lee, C.P., Ha, K., Pao, L.Y., 2007. Development of a practical driving cycle construction methodology: a case study in Hong Kong. Transport. Res. Part D – Transp. Environ. 12 (2), 115–128. Kamble, S.H., Mathew, T.V., Sharma, G.K., 2009. Development of real-world driving cycle: case study of Pune, India. Transport. Res. Part D 14, 132–140. Lin, J., Niemeier, D.A., 2003. Estimating regional air quality vehicle emission inventories: constructing robust driving cycles. Transport. Sci. 37 (3), 330–346. Ma, H., Riera-Palou, X., Harrison, A., 2011. Development of a new tank-to-wheels methodology for energy use and greenhouse gas emissions analysis based on vehicle fleet modeling. Int. J. Life Cycle Assess. 16, 285–296. Manuel, J.B., Biona, M., Culaba, A.B., 2006. Drive cycle development for tricycles. Clean Technol. Environ. Policy 8 (2), 131–137. Nesamani, K.S., Subramanian, K.P., 2006. Impact of real-world driving characteristics on vehicular emissions. JSME Int. J. Ser. B – Fluids Therm. Eng. 49 (1), 19–26. Saleh, W., Kumar, R., Kirby, H., Kumar, P., 2009. Real world driving cycle for motorcycles in Edinburgh. Transport. Res. Part D 14, 326–333. Schifter, I., Díaz, L., Rodríguez, R., López-Salinas, E., 2005. A driving cycle for vehicle emissions estimation in the Metropolitan area of Mexico city. Environ. Technol. 26 (2), 145–154. Tamsanya, S., Chungpaibulpatana, S., Limmeechokchai, B., 2009. Development of a driving cycle for the measurement of fuel consumption and exhaust emissions of automobiles in Bangkok during peak periods. Int. J. Automot. Technol. 10 (2), 251–264. U.S. Environmental Protection Agencya, Dynamometer Drive Schedules (last updated June 2, 2013). (accessed 10.06.15). U.S. Environmental Protection Agencyb, Vehicle Testing Regulations (last updated May 11, 2015). . Wang, B.H., Luo, Y.G., Zhang, J.W., 2008. Simulation of city bus performance based on actual urban driving cycle in China. Int. J. Automot. Technol. 9 (4), 501– 507. Watson, H.C., 1995. Effects of a Wide Range of Drive Cycles on the Emissions from Vehicles of Three Levels of Technology. SAE Paper 950221. Watson, H.C., Milkins, E.E., Braunsteins, J., 1982. The development of the Melbourne Peak Drive Cycle. SAE/ARRB Second Conference on Traffic Energy and Emissions, Melbourne, 19–21 May 1982, Paper 82148. Zhang, S., Wu, Y., Liu, H., Huang, R., Un, P., Zhou, Y., Fu, L., Hao, J., 2014. Real-world fuel consumption and CO2 (carbon dioxide) emissions by driving conditions for light-duty passenger vehicles in China. Energy 69, 247–257.