Vous êtes sur la page 1sur 29

1 Acceleration-Deceleration Behaviour of Various Vehicle

2 Types

3 Dr. P.S.Bokare (corresponding author)


Principal, RSR Rungta College of Engineering and Technology,
Kohka Road, Kurud, Bhilai,
Chhattisgarh, India 490024
E Mail: psbokare@gmail.com
Tel:91-788-6459564/65
Fax:91-788-2286481

Dr. Akhilesh Kumar Maurya


Associate Professor, Department of Civil Engineering,
Indian Institute of Technology Guwahati, India 781039
E Mail: maurya@iitg.ernet.in
Tel:0361-2582426
Fax:91-361-2582440

Word Count (Text = –, Tables = –, Figures = – Total = – )

4 July 31, 2014


5 Abstract

6 Acceleration/deceleration (A/D) behaviour of vehicles is important for various applications


7 like length of yellow light at intersection, determination of sight distances at intersection, deter-
8 mination of length of A/D lanes, ramp design, traffic simulation modelling, vehicular emission
9 modelling, instantaneous fuel consumption rate modelling, etc. Literature reports A/D studies for
10 cars in lane disciplined homogeneous traffic. However, Indian traffic stream is weak lane dis-
11 ciplined and heterogenous, containing various vehicle types like truck, motorized three and two
12 wheeler and diesel and petrol driven cars. Also, the reported studies are based on out of date data,
13 collected using traditional and less accurate methods. Hence, this work aims to study the A/D be-
14 haviour of various vehicle types using modern instruments like Global Positioning System (GPS)
15 in controlled manner including maximum A/D envelop.
16 It is observed that acceleration rates of vehicles observed in this study, differed from accel-
17 eration rates reported in literature. On finding existing acceleration models insufficient to explain
18 the acceleration behaviour of vehicles observed in this study, new models have been proposed and
19 validated using statistical tools. Acceleration behaviour of cars varied with the change in gears,
20 though the pattern remained similar in all gears.
21

22 Key words – Acceleration/Deceleration Models, GPS, maximum and mean acceleration/deceleration.


Bokare-Maurya 1

23 INTRODUCTION
EHICLE acceleration/deceleration (A/D) behaviour has significant impact on several traffic
24

25

26
V related factors like intersection design, deceleration lane design, ramp design, traffic sim-
ulation modelling, vehicular emission modelling, instantaneous fuel consumption rate modelling
27 etc.
28 At signalized intersections, vehicles decelerate, stop, queue up (during red signal) and then
29 accelerate during green and amber signal. The way vehicles decelerate, stop, crawl in queue, and
30 accelerate to leave the intersection depends on number of factors such as, individual vehicle type,
31 driver behaviour, number and type of vehicles at intersection etc. Stronger accelerations may be
32 applied by vehicles with high A/D capability (like cars) to clear the intersection during amber and
33 green phases. Aggressive or young drivers may also showcase similar behaviour.
34 In India (and in many Asian countries) the traffic is heterogenous (having vehicles with
35 varying size, A/D capabilities and weight to power ratios, sharing same right of way) and weak
36 lane disciplined. Hence, at signalized intersection, the movement of vehicles with high accelera-
37 tion capability, like car, is restricted by presence of vehicles with low acceleration capability like
38 motorized three wheelers, trucks etc. Therefore, to model such phenomena and to understand the
39 effect of such restricted movements on design of duration of red, green and amber at signal inter-
40 section, fuel consumption and tailpipe emission, the precise knowledge of A/D behaviour of these
41 vehicles is needed. Many other authors, (like 2, 3, 19, 26, 29, 30) also stated the importance of
42 modelling A/D.
43 In past, A/D profile models are developed for different applications. Most of the A/D mod-
44 els reported in literature refer maximum acceleration capabilities of vehicles. Experimental reports
45 indicate that vehicle drivers use maximum A/D capabilities at intersection and in the incidence of
46 crash. Acceleration rate most frequently occurring at signalized intersection depends on, vehicle
47 characteristics, geometry of intersection, nature and extent of hindrance to A/D of subject vehicle,
48 driver attitude to speeding, etc. (21). Due to numerous such factors influencing A/D of vehicles,
49 less number of studies are available in this area (15, 30). Hence, there is a great need to study A/D
50 behaviour of various vehicle types.
51 Furthermore, most of the existing models are based on outdated and limited data (for ex-
52 ample - Bham and Benekohal (8) model uses 1985 data, Akcelik and Biggs (3) uses data prior to
53 1987, etc.) which are insufficient to describe the acceleration behavior of current fleet and driver
54 behavior. Many researchers (5, 7, 12, 17, 24, 29) used traditional methods for vehicle’s speed mea-
55 surements which may not provide precise speed measurement. In few recent studies, like Wang
56 et al. (32), GPS is used to measure the vehicle’s speed and acceleration accurately.
57 Though, deceleration rate is an important input for duration of amber light at signalized
58 intersection, limited work (3, 7, 33)) is reported in the past on deceleration modelling vehicles in
59 comparison to acceleration modelling. Bennet and Dunn (7) reported second order polynomial
60 deceleration model for vehicles in New Zealand. With rapid change in engine technology and new
61 generation drivers taking seat of older ones, it has become imperative to have a fresh look at the
62 A/D behaviour.
63 After an extensive review of literature related to A/D, it is found that

64 1. Most of the existing A/D models described above are developed for a particular vehicle
65 type (passenger car). In India however, the traffic stream consists of vehicles differing
66 in size and shape and weight to power ratio (engine characteristics), (5). Their share
Bokare-Maurya 2

67 in traffic stream is 36.5%, 19%, 6.8% and 16.5% respectively for truck, car, motorized
68 three and two wheeler (11).

69 2. Some of the reported studies (3, 8) formulated acceleration models based on old data
70 sets (of 1968 and 1985). The vehicle technology and driver response to control devices
71 have changed since then.

72 3. Limited studies (11, 24) related to acceleration behaviour for Indian traffic streams are
73 found in literature. In these studies, speed data are collected using manual measurement
74 of travel time to traverse a fixed distance. This method limits scope of experiment to get
75 complete acceleration profile of vehicle.

76 Therefore, present study aims at developing A/D models for various vehicle types found
77 on Indian roads using modern device like Global Position System (GPS).

78 EXPERIMENTAL METHODS
79 To study the A/D behaviour of a vehicle one needs to observe the speed and position data of vehicle
80 with time. Vehicle approaching to a signalized intersection has to decelerate and stop during red
81 phase of signal and has to accelerate at the onset of green phase. Figure 1 presents the time-distance
82 and speed-time diagram of a vehicle movement at signalized intersection showing cruising, decel-
eration, stopping, acceleration and cruising stages of a vehicle at signalized intersection.

FIGURE 1 : Time-distance and speed-time diagrams showing the deceleration, waiting and ac-
celeration manoeuver of a vehicle at signalized intersection.
83
84 Therefore, data on A/D pattern of vehicles can be well collected by observing the vehicle
85 movement at signalized intersection. However, heterogeneous and weak lane disciplined traffic at
86 intersection in India often results in data that is inconsistent (in A/D manoeuver) and difficult to
87 analyze. At signalized intersections, generally smaller vehicles (like motorized and non-motorized
88 two wheelers and three wheelers) creep through the gaps between other queued vehicles (like cars,
89 trucks) and stop in front of the queues at intersection (see the Figure 2 showing a typical scenario
90 at intersection).
Bokare-Maurya 3

FIGURE 2 : A photograph showing heterogeneity and congestion condition in front of the queues
at urban signalized intersection at Nagpur, India)

91 The congestion condition at the front of queues can be seen from the Figure 2, which leads
92 to inconsistent A/D behaviour of vehicles. Therefore, an alternative is, to observe driver behaviour
93 over short stretch and under controlled conditions (replicating signalized intersection lead vehicle
94 acceleration-deceleration) as an acceptable surrogate for actual behaviour. Such alternative pro-
95 cedures are also used by earlier researchers like (6, 10, 14, 23, 25, 34). Therefore, the present
96 study also collected the vehicle A/D data over selected stretch of road under controlled conditions,
97 replicating lead vehicles at signalized intersection.

98 Data Collection
99 Road stretches for the study were so selected that vehicle travels under free flow traffic, the facility
100 is access controlled so that there is no obstruction to speeding. Road geometry is fairly straight
101 to have constant effect of geometry on A/D and entire road surface is smooth to ensure constant
102 effect of rolling resistance.
103 Accordingly, the study was conducted on 1.5 km stretch of a two lane Nagpur-Mumbai
104 Highway on outskirts of Wardha Town, about 70 km from Nagpur (India) which satisfies above
105 mentioned conditions. All kinds of vehicles (like truck, car, motorized three-wheeler and motor-
106 ized two-wheeler) generally observed plying over this facility. GPS device with 1 Hz data logging
107 (data logged once in a second) is used to collect speed and position data of vehicles.
108 Most of the cars are owner driven. They did not allow the volunteers to board the car and
109 collect data using GPS device. Hence data collection of cars is undertaken at a link road joining
110 Indian Institute of Technology Guwahati, India and National Highway No. 37 in India. Hired cars
111 were used for experimentation. V-Box device, capable of logging data at 10 Hz. frequency was
112 used for data collection. A similar device could not be used for other vehicle types since this device
113 requires more time to install its antennae.
114 All the drivers were asked to speed up their vehicles from stop condition to achieve their
115 desired speed (maximum speed at which driver feel safe for a given road geometry and environ-
116 mental condition; hereafter referred as maximum speed) as early as possible. After allowing them
117 to cruise at the maximum speed for some time, they were asked to decelerate to stop condition in
118 shortest possible time (30). This replicated movement of queue leaders at signalized intersection.
Bokare-Maurya 4

119 Analysis Methods


120 Position and speed data of different vehicles were collected using GPS during the designed ex-
121 periment (i.e. acceleration, cruising and deceleration). All experiments (trips) were made during
122 free flow traffic condition. Similar to Rakha et al. (27), exponential smoothing algorithm was used
123 for smoothing of GPS data. Following steps were used in analysis of collected GPS data for A/D
124 study:

125 1. Acceleration was computed from second by second GPS speed data collected during
126 acceleration manoeuver data, using Equation 1:
v2 − v1
at2 = (1)
t2 − t1
127 where, at2 is acceleration at time t2 and v1 and v2 are the speeds at time t1 and t2
128 respectively. It is assumed that acceleration process ended when the increment in speed
129 between two successive data points is less than 0.1 m/s, for next five seconds (refer
130 Wang et al. (32)).

131 2. Similar to acceleration calculation, deceleration was computed from second by second
132 speed data obtained from GPS during deceleration manoeuver using Equation 2:
v1 − v2
dt2 = (2)
t2 − t1
133 where, dt2 is deceleration at time t2 and v1 and v2 are the speeds at time t1 and t2
134 respectively. Starting of deceleration process was defined from the time onwards where
135 deceleration values calculated from Equation 2 are greater or equal to 0.1 m/s2 for five
136 consecutive seconds (refer Wang et al. (33)). At the end of deceleration process vehicle
137 speed become zero.

138 3. For all vehicle classes, speeds were averaged over every second. The average speed
139 profile so obtained was termed as idealized speed profile.

140 4. For A/D behaviour modelling, idealized A/D - speed data were obtained by averaging
141 A/D values over 1 m/s speed interval for each vehicle class. This idealized A/D - speed
142 data is used to develop the A/D model.

143 5. Developed A/D models were evaluated using various statistical methods.

144 Summary of data collected i.e. number of trips and number of data points for each vehicle
145 type along with their weight to horsepower ratio is presented in Table 1.
146 It can be observed from the Table 1 that number of data points for cars (petrol and diesel,
147 both) are significantly higher than number of data points collected for other vehicles. This was due
148 to higher frequency (i.e. 10 Hz) of data logging in V-Box. Analysis results of collected field data
149 are presented in next section.
Bokare-Maurya 5

TABLE 1 : Number of trips and data points collected for various vehicle types

Serial No Type of vehicle Weight to No. of trips∗∗


horsepower Ratio
lb/hp∗ Acceleration Deceleration
1 Trucks 300 (183) 114(3231) 42(902)
2 Motorized three 100 (61) 116(3159) 67(1740)
wheeler
3 Motorized two 31 (18.91) 59(1050) 29(594)
wheeler
4 Car
• Diesel car 33 (20.13) 110(43862) 53(3982)
• Petrol car 30 (18.3) 115(50152) 75(5808)
∗ In bracket kg/kW
∗∗ Figures in bracket are number of data points

FIGURE 3 : Speed profiles of various vehicle types (a) Truck (b) Motorized three wheeler (c)
Motorized two wheeler (d) Diesel car (e) Petrol car during acceleration manoeuver

150 DATA ANALYSIS


151 Figure 3 presents speed profiles (speed-time scatter) of truck, car, motorized three-wheeler and
152 motorized two-wheeler during acceleration maneuver.
153 It is observed from Figure 3 that the maximum desired speed time to achieve maximum
154 speed varies with vehicle type and driver. Slope of speed time scatter is more in the beginning of
155 acceleration maneuver and less at the end of acceleration maneuver. This indicates higher acceler-
156 ation at the beginning of acceleration maneuver and lower accelerations at the end of acceleration
157 maneuvers.
158 Figure 4 presents speed profiles (speed-time scatter) of truck, car, motorized three-wheeler
Bokare-Maurya 6

and motorized two-wheeler during deceleration maneuver. Figure 4 indicates that the decelera-

FIGURE 4 : Speed profiles of various vehicle types(a) Truck (b) Motorized three wheeler (c)
Motorized two wheeler (d) Diesel car (e) Petrol car during deceleration manoeuver
159
160 tion time varies with vehicle type, driver and speed at which driver starts decelerating (approach
161 speed). Higher approach speed is associated with lesser deceleration time and vice versa. Ben-
162 net and Dunn (7), Wang et al. (33) also reported similarly. The reason is that the faster vehicles
163 decelerate at higher rate and hence take less time to complete deceleration maneuver.
164 The speed-time scatters during A/D maneuver indicate that there is dependence of A/D
165 parameters like, A/D time and distance, maximum and mean A/D rates and speed at maximum
166 A/D (referred hereafter as critical speed) on maximum speed of trip. A similar dependence was
167 also reported by Bennet and Dunn (7), Samuels and Jarvis (28), Wang et al. (33). Therefore, the
168 trips are segregated as per their maximum speed range and these parameters are evaluated within
169 each range of maximum speed.
170 Table 2 presents these parameters during acceleration manoeuver for all vehicle types.
171

172 Acceleration time and acceleration distance


173 During acceleration manoeuver, acceleration time and acceleration distance for all vehicle type
174 increase with increase in maximum speed (driver desired speed) in all speed ranges of vehicle
175 (refer Table 2). This is because with more speed, driver needs more time to achieve maximum
176 acceleration rate. This time and distance is also related to the acceleration capability of vehicles. In
177 similar speed range, acceleration distance (and time) of vehicle with lower acceleration capability
178 (like motorized three wheeler) is more as compared to other vehicle types with higher acceleration
179 capability (such as truck and motorized two-wheeler). A similar observation is reported by (24)
180 for passenger car.
181
Bokare-Maurya 7

TABLE 2 : Various parameters corresponding to different maximum speed ranges of all vehicle
classes during acceleration manoeuver
Vehicle Max. Accel. Accel. Critical Max. Mean
Type Speed Time Distance Speed∗ Accel. Accel.
Range Rate Rate
km/h (m/s) (sec) (m) (m/s) (m/s2 ) (m/s2 )
20-30 (5.55-8.33) 11 56.98 2.77 0.75 0.28
30-40 (8.33-11.11) 17 98.26 1.53 1.00 0.29
Truck
40-50 (11.11-13.89) 34 259.08 1.27 0.96 0.24
50-60 (13.89-16.67) 35 361.20 1.08 0.87 0.24
15-25 (4.17-6.94) 27 94.50 2.04 0.54 0.21
Motorized
25-32 (6.94-8.88) 36 156.24 2.30 0.45 0.22
three wheeler
32-36 (8.88-10.0) 40 220.80 1.53 0.60 0.22
36-43 (10.0-11.94) 50 308.50 2.53 0.64 0.20
30-40 (8.39-11.11) 22 167.24 4.21 0.94 0.47
Motorized
40-50 (11.11-13.89) 34 337.68 3.27 1.08 0.39
two wheeler
50-60 (13.89-16.67) 35 374.80 3.97 1.96 0.52
68-76 (18.88-21.11) 34.80 519.18 1.46 1.89 0.55
76-84 (21.11-23.33) 45.70 766.22 1.34 2.23 0.47
Diesel Car
84-92 (23.33-25.55) 52.50 923.64 1.21 1.97 0.52
80-84(22.22-23.33) 28.80 425.99 2.4 2.24 0.82
84-88(23.33-24.44) 31.60 545.01 2.78 2.47 0.64
Petrol Car
88-92(24.44-25.25) 34.80 620.90 3.74 2.87 0.70
Max.: Maximum, Accel: Acceleration, ∗ : Speed at maximum acceleration

182 Critical speed


183 The speed at which the maximum acceleration rate occurs (referred as critical speed) varies with
184 vehicle type. The critical speed range is 1.08 m/s to 2.77 m/s for truck, 1.53 m/s to 2.53 m/s
185 for motorized three-wheeler, 3.27 m/s to 4.21 m/s for motorized two- wheeler, 1.46 m/s to 1.21
186 m/s for diesel car, 2.40 m/s to 3.74 m/s for petrol car. Trucks achieve maximum acceleration
187 quickly, whereas other vehicle types take more time to achieve their maximum acceleration rate.
188 Hence, more data points could be recorded before vehicles’ critical speed for other vehicle types
189 as compared to truck. Also it is observed that since trucks achieve highest acceleration quickly,
190 the proportion of time spend before achieving maximum acceleration is negligible as compared to
191 total acceleration time (only 2 sec out of 60 sec acceleration manoeuver for truck). Further, it is
192 observed that critical speed reduces (in most of the cases) with increase in driver desired speed.
193 This implies that driver accelerates quickly when he/she plans to drive with higher speed.
194

195 Maximum and mean acceleration rate


196 It can be observed from Table 2 that the maximum acceleration rate varies with vehicle type and
197 maximum speed of vehicle. The maximum and mean acceleration rates are higher at higher maxi-
198 mum speed, in majority of cases, of all vehicle types.
199 The maximum acceleration rate observed for truck is 1.0 m/s2 , for motorized three-wheeler
200 0.64 m/s2 , for motorized two-wheeler 1.95 m/s2 , for diesel car 2.23 m/s2 and for petrol car 2.87
201 m/s2 (refer Table 2). This indicates that petrol car employs highest acceleration rate among all
202 vehicle types while motorized three-wheeler employs lowest maximum acceleration rate. Mean
203 acceleration rates of different vehicles also follow similar trends.
204 Maximum acceleration rates of different vehicles at various speed reported in literature are
205 presented in Table 3. It can be observed from Table 2 that maximum acceleration values of cars
206 observed in this study are higher than reported acceleration value by previous Indian researchers
207 (5, 11). However, it is comparable with some other international studies (8, 32). Previous Indian
Bokare-Maurya 8

208 studies used travel time method for speed measurement which may under estimate the speed (hence
209 acceleration) of vehicles. Similar observations holds true for acceleration values of truck. In
210 case of motorized two wheelers, observed acceleration value is little lower than reported values in
211 previous Indian studies. This underestimation of two wheeler’s acceleration may due to driver’s
212 cautiousness during data collection in the present study.

TABLE 3 : Maximum and Mean Acceleration Rates Observed by Various Researchers

Author Country Vehicle Speed Max. Accel


Type Range Rate
km/h m/s2
Dey et al. (11) India Truck 0-43 0.47
Car 0-72 1.24
Motorized 0-54 1.52
Two-wheeler
Arasan and Koshy (5) India Truck 0-20 0.79
20-40 0.45
> 40 0.35
Car 0-20 1.50
20-40 1.30
> 40 1.00
Motorized 0-20 1.01
Three-wheeler 20-40 0.58
> 40 0.34
Motorized 0-20 1.35
Two-wheeler 20-40 1.03
> 40 0.37
RaiChowdhury and Rao (24) India Car 0-45 2.25
Dockerty (13) USA Car – 2.02
Glauz et al. (18) USA Car – 2.50
St.John (31) USA Car – 3.36
Bonesson (9) USA Car – 2.02
Bham and Benekohal (8) USA Car 0-60 2.00
Wang et al. (32) USA Car 0-60 1.86

213 Various parameters of deceleration maneuver are presented in Table 4


214 Following observations were made from Table 4.
215

216 Deceleration time and deceleration distance


217 It can be observed from Table 3 that deceleration distance and time increases with increase in max-
218 imum (desired) speed (speed at which driver starts decelerating) in most of the speed ranges of all
219 vehicle types. This implies that during deceleration manoeuver from higher speed to stop condition
220 drivers require more distance (or time) as compared to deceleration manoeuver from lower speed
221 ranges. Further vehicle with lower deceleration capability (like motorized three wheeler) requires
222 more distance and time to complete the deceleration manoeuver in comparison to other vehicles
223 with higher deceleration capability (like truck and motorized two wheeler) at a particular maxi-
224 mum speed range. These observations are in agreement with the observations made by researchers
225 like Akcelik and Biggs (3), Wang et al. (33) but contradict with the findings of Bennet and Dunn
226 (7).
227 The deceleration time of diesel and petrol car is more or less similar irrespective of desired
228 speed of car. It can be seen from Table 3 that the variation in deceleration time of petrol car is more
229 with desired speed as compared to diesel car.
Bokare-Maurya 9

TABLE 4 : Various parameters corresponding to different desired speed ranges of all vehicle
classes during deceleration manoeuver
Vehicle Maximum Deceleration Deceleration Speed Maximum Mean
Category Speed Time Distance at Maximum Deceleration Deceleration
Range Deceleration Rate Rate
km/h(m/s) (sec) (m) (m/s) (m/s2 ) (m/s2 )
20-30 (5.55-8.33) 16.00 70.88 3.75 0.72 0.47
30-40 (8.33-11.11) 21.30 124.39 3.82 0.75 0.46
Truck
40-50 (11.11-13.88) 20.33 148.81 3.85 0.88 0.52
50-60 (13.88-16.66) 30.75 243.54 3.93 0.88 0.51
27-31(7.5-8.61) 19.85 107.52 3.15 0.85 0.35
Motorized
31-35 (8.61-9.72) 27.33 159.33 3.21 1.12 0.31
three
35-39 (9.72-10.83) 26.45 172.31 3.63 1.14 0.36
wheeler
39-43 (10.83-11.94) 28.42 201.05 3.21 1.06 0.36
40-50 (11.11-13.88) 18.30 152.01 7.52 1.60 0.58
Motorized
50-60 (13.88-16.66) 21.21 214.82 7.27 1.33 0.47
two wheeler
60-65 (16.66-18.05) 23.00 292.79 9.65 0.59 0.41
92-94 (25.55-26.11) 8.08 83.38 10.28 4.30 3.19
94-96 (26.11-26.66 8.52 108.80 16.17 4.33 3.11
Diesel Car
96-98 (26.26-27.22) 8.60 113.04 23.28 5.00 3.36
98-100 (27.22-27.77) 8.87 129.59 24.21 4.52 3.72
61-72 (17-20) 7.61 85 2.97 3.36 2.42
Petrol Car 72-83 (20-23) 9.96 129 3.79 3.97 2.52
83-91 (23-25) 10.27 134 5.69 4.33 2.59

230

231 Maximum and mean deceleration rate


232 Maximum deceleration rates generally increases with increase in maximum speed of all vehicle
233 types observed in this study. This observation is in agreement with the observation reported by
234 Wang et al. (33) and Bennet and Dunn (7). The average maximum deceleration rates recommended
235 by ITE (20) are 3.0 m/s2 and by AASHTO (1) are 3.4 m/s2 for cars. In present study the average
236 maximum deceleration rates observed for petrol car is 3.88 m/s2 and for diesel car is 4.53 m/s2 . In
237 case of petrol cars the average maximum deceleration rates exceed the recommended rates where
238 as for diesel cars the rates are within the limit recommended by ITE (20) and AASHTO (1). For
239 all other vehicle types the average maximum deceleration rates are well within the recommended
240 maximum rates.
241 Wang et al. (33) reported that there is as such no relation between approach speed (speed at
242 which driver starts decelerating) and maximum and mean deceleration rates. In this study, however,
243 in case of cars (petrol and diesel) the maximum and mean deceleration rates are found to increase
244 with increase in approach speed in most of the cases. This is because at higher approach speed
245 driver is in hurry to stop within a particular distance. Hence they apply higher deceleration rate.
246 Petrol car employ highest deceleration rates while truck use the lowest among the vehicle
247 types considered in this study. Deceleration rates reported by Bennet and Dunn (7) for vehicles on
248 free motor way in New Zealand, are similar to the maximum deceleration values observed at speed
249 range 60 − 70 km/h in present study.

250 STUDY OF ACCELERATION BEHAVIOUR


251 This section presents the acceleration behaviour of all vehicle types considered in this study. The
252 speed data obtained from GPS are smoothed using exponential smoothing (27). The resulting
253 speed data for all trips is further used to compute acceleration as per Equation 1. The acceleration
254 is then plotted against speed and presented in Figure 5.
Bokare-Maurya 10

1.2 1.2

Acceleration, m/s
Acceleration, m/s2
0.9
0.8

0.6 0.6

0.4
0.3
0.2

0 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12
Speed, m/s Speed, m/s

(a) Truck (b) Motorized three wheeler

1.4 3
1.2
2.5
Acceleration, m/s2

Acceleration, m/s2
1
2
0.8
1.5
0.6
0.4 1

0.2 0.5

0 0
0 2 4 6 8 10 12 14 16 0 5 10 15 20 25
Speed, m/s Speed, m/s

(c) Motorized two wheeler (d) Diesel Car

(e) Petrol Car

FIGURE 5 : Variation of acceleration distance and time with vehicle type and average maximum
trip speed

255 It is observed from the plots in Figure 5a, 5b, 5c, 5d and 5e that;

256 1. Acceleration has a strong relationship with speed in all vehicle types. Bham and Beneko-
257 hal (8) reported that while modelling acceleration of vehicles, speed is preferred over
258 distance since speed provides better fit than distance. Distance is a cumulative measure
259 and hence errors accumulate over time. Small initial error in distance profile magnifies
260 over time. This results in not so good fit and errors are unrealistic, Bham and Benekohal
261 (8). Therefore, appropriate model selection becomes difficult. Bham and Benekohal (8)
262 also found that shape of distance profile of vehicle look similar for different speed mod-
263 els. Moreover, many authors (4, 8, 21) reported that vehicle acceleration changes over
264 the vehicle speed. At lower speed, acceleration is high and at higher speed acceleration
Bokare-Maurya 11

265 is low, indicating strong relationship between acceleration and speed.

266 2. In case of truck, maximum acceleration is quickly achieved (due to high weight to horse-
267 power ratio). There are very few data points before achieving maximum acceleration.
268 Hence the regime before achieving maximum acceleration is negligible.

269 3. In case of motorized three and two wheeler, rate of acceleration initially increases up to
270 a maximum value and then decreases to become zero at cruising speed (driver desired
271 speed).

272 4. In case of cars, acceleration initially increases in first gear, achieves maximum value
273 and decreases during gear change. Similar behaviour is repeated while changing second
274 and third gears. Maximum acceleration is achieved in first gear. This behaviour fades in
275 fourth and fifth gear change. The reason is, while shifting gears from third to fourth or
276 from fourth to fifth gear, the change in speed is negligible. Hence the rate of change of
277 acceleration is also not significant.

278 5. The speed at which maximum acceleration is achieved by vehicle and time taken by
279 vehicle to achieve maximum acceleration are presented in Table 5, for all vehicle types.

TABLE 5 : Speed at maximum acceleration and Time Taken to Achieve Maximum Acceleration

Vehicle Type Speed at max. Time taken to


acceleration, m/s achieve max.
acceleration, sec.
Truck 1.45 2
Motorized 2.5 4
Three-wheeler
Motorized 2.97 3
Two-wheeler
Car
Diesel car 1.38 0.8
Petrol Car 1.2 1

280

281 It is observed that the speed at which the maximum acceleration is achieved and time
282 taken to achieve it, are different for different vehicles types due to their different weight
283 to horsepower ratio and engine capacity. In case of car and truck, maximum accelera-
284 tion is quickly achieved at the lower speed whereas in case of other vehicles maximum
285 acceleration is achieved late.

286 Acceleration Models


287 This section presents acceleration modeling for various vehicle types. The existing acceleration
288 models are evaluated for the suitability to present data set and then fresh models are proposed.
289 For modelling average behaviour of vehicles, acceleration and speed are averaged over
290 every 1 m/s interval, (32) and the resulting idealized acceleration-speed plot for all vehicle types
291 is presented in Figure 6.
292 The nature of idealized plot depicts similar behaviour like that of scatter plots of acceleration-
293 speed presented in Figure 5.
Bokare-Maurya 12

0.7 1.05
0.6

Acceleration, m/s2
Acceleration, m/s2
0.5
0.7
0.4

0.3

0.2 0.35

0.1

0 0
0 2 4 6 8 10 12 14 16
0 10 20 30 40 50
Speed, m/s Time, s

(a) Truck (b) Motorized three wheeler

1.4
Average Acceleration, m/s2

1.2

0.8

0.6

0.4

0.2

0
0 10 20 30 40 50
Time, s

(c) Motorized two wheeler (d) Diesel Car

2.5
Maximum acceleration in
2 various gears
Acceleration, m/s2

1.5

0.5

0
0 5 10 15 20 25 30
Speed, m/s

(e) Petrol Car

FIGURE 6 : Idealized plot of acceleration with speed

294 The single regime model proposed by Samuels and Jarvis (28) assumes constant accelera-
295 tion throughout acceleration maneuver. This is the simplest model form that can be used to describe
296 acceleration-speed relationship. However, the idealized plot of acceleration-speed for motorized
297 three-wheeler and motorized two-wheeler (Figure 6b and 6c) indicate that there exist two separate
298 regimes; regime-I before attaining maximum acceleration and regime-II after attaining maximum
299 acceleration. Slope of acceleration-speed curve is opposite before and after a point of maximum
300 acceleration. Hence, in case of motorized three and two wheelers the acceleration-speed is mod-
301 elled in two separate regimes (dual regime) unlike Samuels and Jarvis (28). It is assumed that the
302 point of maximum acceleration will act as a separator point between two regimes. The dual regime
303 model of acceleration-speed offers simplicity and ease of calculations as compared to polynomial
304 model. This results in reduction in simulation time. Dual regime model contains only one point
305 of discontinuity unlike several in linear model in the form of step function, (8, 22, 32). For car
Bokare-Maurya 13

306 (diesel and petrol) and truck, however, the part of regime-I is very small (though it exists), since
307 maximum acceleration is attained very fast at lower speed hence can be neglected. Hence for car,
308 the acceleration-speed relationship is modeled as a single regime. The shape suggests non linear
309 form of model, hence non-linear form is further evaluated.

310 Acceleration Model Formulation


311 Various model forms such as linear, polynomial and exponential are calibrated using linear regres-
312 sion for all vehicle types. To choose among various model forms, Residual Sum of Squares (RSS),
313 is used, Equation 3, (16).
n
X  2
RSS = yi − ŷ (3)
i=1

314 where, RSS is Residual Sum of Squares, yi is observed value of response, ŷ is estimated value of
315 response.
316 The RSS values for various vehicle types and various model forms are presented in Table 6

TABLE 6 : RSS Values for various acceleration model forms for various vehicle types

Vehicle Type Regime-I Regime-II


Linear Exponential Polynomial Linear Exponential Polynomial
Model Model Model Model Model Model
Motorized 0.014 0.023 0.004 0.11 0.02 0.023
three wheeler
Motorized 0.011 0.001 0.0007 0.001 0.0008 0.05
two wheeler
Truck1 0.03 0.01 0.17 – – –
Diesel Car1 0.59 0.55 0.81 – – –
Petrol Car1 2.78 2.14 12.86 – – –
1 -Single Regime Model

317 It is observed from Table 6 that for motorized three-wheeler and motorized two-wheeler,
318 minimum values of RSS are for linear form for regime-I and exponential form for regime-II.
319 For truck, diesel and petrol cars minimum RSS values are for exponential form.
320 Linear regression (16) is used to decide the model coefficients for all vehicle types. The
321 general form of models and model coefficients are presented in Table 7.

322 Model Diagnostic


323 The linear regression model (which is used for formulating model) is based on several assumptions
324 (16). These assumptions are used to diagnose the correctness of model. The diagnostic procedures
325 used are graphical (residual plots, observed and predicted value plots) and numerical (hypothesis
326 test), (16), are used for testing models.

327 Graphical Procedure

328 Residues versus Predicted Value Plot Residues are the differences between observed ac-
329 celeration values and predicted (modeled) acceleration values. The plot of residues vs. predicted
330 values should show no trend for assumption of independence of errors to be true (16). The residual
331 plots (plot of residuals versus predicted values) are presented in Figure 9.
Bokare-Maurya 14

TABLE 7 : Model parameters for various vehicle types

Vehicle Type Regime-I Regime-II


a = k1 × v 2 + k2 × v + k3 a = k4 × ek5 ×v
k1 k2 k3 r2 k4 k5 r2
Motorized -0.23 0.98 -0.29 0.99 1.47 -0.26 0.95
three wheeler
Motorized -0.39 0.007 0.58 1 1.62 -0.15 0.98
two wheeler
a = k6 × ek7 ×v
k6 k7 r2
Truck1 0.666 -0.13 0.92
Diesel Car1 2.38 -0.1 0.88
1
Petrol Car 2.5 -0.09 0.83
1
-Single regime model

332 The plots in Figure 7 indicates that the residues are not correlated to predicted values (null
333 plot, Freund and Wilson (16)) and also errors are not dependent on predicted values. This rein-
334 forces the compliance of assumption of ordinary least square method, that the errors are indepen-
335 dent.

336 Observed versus Predicted Acceleration Plot The observed acceleration (computed from
337 observed speed values using Equation 1) is then plotted with predicted acceleration for all vehicle
338 types and presented in Figure 8.
339 Figure 10 indicates that the points are clustered around a 45 ◦ straight line. For a nor-
340 mally distributed variable, this is required. Hence the assumption ‘error terms are normally
341 distributed’ is valid (refer Pg. 358 and 359, (16)).

342 Numerical Procedure

343 Hypothesis test Paired ‘t’ test is used to test the means of observed acceleration and pre-
344 dicted acceleration. Two hypothesis are tested −(i) null hypothesis: µ̄ = µo − µm = 0, where µo
345 and µp are mean of observed and predicted acceleration respectively and (ii) alternate hypothesis:
346 µ̄ 6= 0. The test statistic is calculated using Equation 4 (16),
µ̄
|t| = | √ | (4)
sd / n

347 where, µ̄ is mean of the difference between observed and predicted acceleration, sd is standard
348 deviation of difference in paired data, n is number of data points. The hypothesis is tested for 95%
349 confidence interval (α = 0.05), where α is significance level. One cannot reject null hypothesis if
350 |t| ≤ tα/2 (= t0.025 ). Table 8 presents values of |t| and tα/2 for various vehicle types.
Bokare-Maurya 15

0.2 0.4
0.15 0.3
0.1 0.2
Residues

Residues
0.05 0.1
0 0
-0.05 0.1 0.2 0.3 0.4 0.5 0.6 -0.1 0.2 0.4 0.6 0.8

-0.1 -0.2
-0.15 -0.3
-0.2 -0.4
Predicted Acceleration, m/s2 Predicted Acceleration, m/s2

(a) Truck (b) Motorized three wheeler, Regime-I

0.4 0.6
0.3
0.4
0.2
0.2
Residues

0.1

Residues
0 0
-0.10.07 0.17 0.27 0.37 0.47 0.57 0.67 0.77 0.4 0.5 0.6 0.7 0.8 0.9
-0.2
-0.2
-0.3 -0.4
-0.4
Predicted Acceleration, m/s2 -0.6
Predicted Acceleration, m/s2

(c) Motorized three wheeler, Regime-II (d) Motorized two wheeler, Regime-I

0.6 0.5
0.4
0.4 0.3
0.2
0.2
Residues

0.1
Residues

0 0
0.1 0.3 0.5 0.7 0.9 1.1 -0.1 0 0.5 1 1.5 2 2.5 3
-0.2 -0.2
-0.3
-0.4 -0.4
-0.5
-0.6 Predicted acceleration, m/s2
Predicted Acceleration, m/s2

(e) Motorized two wheeler, Regime-II (f) Diesel car

0.5
0.4
0.3
0.2
Residues

0.1
0
-0.1 0 0.5 1 1.5 2 2.5 3
-0.2
-0.3
-0.4
-0.5
Predicted acceleration, m/s2

(g) Petrol car

FIGURE 7 : Plot of acceleration residuals versus predicted acceleration values

351 For all vehicle types, |t| ≤ tα/2 , hence, null hypothesis that µ = µo − µm = 0 cannot be
352 rejected. This implies that there is no statistically significant difference between means of observed
353 and predicted acceleration values.
Bokare-Maurya 16

0.7 1

0.6
0.8

Acceleration, m/s2
Acceleration, m/s2
0.5

Predicted
Predicted
0.4 0.6

0.3 0.4
0.2
0.2
0.1

0 0
0 0.4 0.8 1.2 0 0.2 0.4 0.6 0.8 1
Observed Acceleration, m/s2 Observed Acceleration, m/s2

(a) Truck (b) Motorized three wheeler, Regime-I

0.8 1

Predicted Acceleration, m/s2


0.9
Acceleration, m/s2

0.6
0.8
Predicted

0.7
0.4
0.6
0.2 0.5

0.4
0 0.4 0.6 0.8 1 1.2 1.4
0 0.2 0.4 0.6 0.8 Observed Acceleration, m/s2
Observed Acceleration, m/s2

(c) Motorized three wheeler, Regime-II (d) Motorized two wheeler, Regime-I

1.4 3
Predicted Acceleration, m/s2

1.2
Predicted accelertion

1
2
0.8

0.6
1
0.4

0.2

0
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3
Observed Acceleration, m/s2 Observed acceleration

(e) Motorized two wheeler, Regime-II (f) Diesel car

3
Predicted acceleration, m/s2

2.5

1.5

0.5

0
0 0.5 1 1.5 2 2.5 3
Observed acceleration, m/s2

(g) Petrol car

FIGURE 8 : Plot of observed acceleration versus predicted acceleration

354 Therefore, it can be concluded that proposed models of acceleration-speed relationship are
355 statistically robust.
Bokare-Maurya 17

TABLE 8 : Results of Hypothesis Test

Vehicle Type |t| tα/2 Remark


Truck 1.48 2.13 |t| < tα/2
Motorized three wheeler
Regime-I 0.03 1.96 |t| < tα/2
Regime-II 0.99 1.96 |t| < tα/2
Motorized two wheeler
Regime-I 0.24 2.16 |t| < tα/2
Regime-II 0.39 2.16 |t| < tα/2
Car
Diesel 0.73 2.6 |t| < tα/2
Petrol 0.093 1.7 |t| < tα/2

356 STUDY OF DECELERATION BEHAVIOUR


357 This section presents the deceleration behaviour of all vehicle types considered in this study. The
358 speed data obtained from GPS are smoothed using exponential smoothing (27). The resulting
359 speed data for all trips is further used to compute deceleration as per Equation 2. The acceleration
360 is then plotted against speed and presented in Figure 9.

(a) Truck (b) Motorized three wheeler

1.6
6
1.4
5
Deceleration, m/s2

1.2
Deceleration m/s2

1 4
0.8 3
0.6
2
0.4
0.2 1

0 0
0 3 6 9 12 15 18 0 5 10 15 20 25 30
Speed, m/s Speed m/s

(c) Motorized two wheeler (d) Diesel Car

5
Deceleration, m/s2

0
0 5 10 15 20 25
Speed, m/s

(e) Petrol car

FIGURE 9 : Scatter plots of deceleration-speed

361 It is observed from the deceleration-speed scatter plots that deceleration is strongly related
Bokare-Maurya 18

362 to speed. To further reinforce this observation, the deceleration is averaged over 1 m/s speed
363 interval and idealized deceleration and speed plots for all vehicle types are presented in Figure 10.
364 While presenting idealized plots, cars are separated from other vehicles since the magnitude of
365 deceleration is more in cars as compared to other vehicles.

(a) Truck, Motorized three wheeler and (b) Diesel car and Petrol car
two wheeler

FIGURE 10 : Idealized plots of deceleration-speed

366 It is observed from Figure 9 and 10 that critical speed (speed where deceleration is max-
367 imum) depends on vehicle type. Critical speed is highest for petrol car and lowest for motorized
368 three wheeler which has lowest deceleration capability in comparison to other vehicle types.

369 Deceleration Models


370 The nature of idealized deceleration-speed curve (more specifically slope) is opposite before and
371 after the critical speed, except in case of cars. Hence, a dual regime is proposed for all vehicle
372 types except cars. Regime-I before attaining maximum deceleration and Regime-II after attaining
373 maximum deceleration. For cars, single regime model is proposed (as depicted in Figure 11a). In
374 order to decide the form of model before and after a critical speed, Residual Sum of Squares (RSS)
375 is used.
376 Residual Sum of Squares for three proposed forms of model such as linear, second order
377 polynomial and exponential are evaluated and results are presented in Table 9.

TABLE 9 : Residual Sum of Squares (RSS) Values for Different Model forms for deceleration-
speed relationship of various vehicle types
Vehicle Category Regime I Regime II
Linear Exponential Second order Linear Exponential Second order
Polynomial Polynomial
Truck 0.066 0.031 0.038 0.006 0.10 0.086
Motorized Three Wheeler 0.007 0.004 0.005 0.00045 0.006 0.002
Motorized Two Wheeler 0.021 0.016 0.09 0.017 0.028 0.19
Petrol Car 1.62 0.931 0.891∗ NA NA NA
Diesel Car 2.03 0.751 0.611 NA NA NA
1
-Single regime model
∗-RSS for third order polynomial is also evaluated and found to be lower than for RSS for second order polynomial

378 It is observed from Table 9 that for all vehicle types (except for cars) negative exponential
379 model for Regime I and linear model for Regime II are found suitable (minimum RSS values are
Bokare-Maurya 19

380 shown by bold face letters). For diesel car, RSS values are minimum for second order polynomial
381 and for petrol car RSS values are minimum for third order polynomial model.
382 Linear regression is used to obtain model parameters from the idealized deceleration and
383 speed relationships of vehicles and presented in Table 10. Model forms for various models are also
384 presented in Table 10.

TABLE 10 : Model parameters and r2 for deceleration-speed models of various vehicle types

Vehicle Calibrated parameter values Critical


Category Regime I Regime II Speed
d1 = k1 × e−k2 ×v d2 = α + β × v m/s
k1 k2 r2 α β r2
Truck 1.58 0.017 0.86 0.104 0.225 0.92 3.83
Motorized three 0.806 0.130 0.90 0.163 0.152 0.94 2.09
wheeler
Motorized two 1.106 0.080 0.95 0.342 0.087 0.86 11.46
wheeler
Cars
Diesel Car dc = −k3 × v 2 + k4 × v + k5
k3 k4 k5 r2
-0.005 +0.15 +0.50 0.92
Petrol Car dc = k6 × v 3 − k7 × v 2 + k8 × v + k9
k6 k7 k8 k9 r2
0.001 -0.52 +0.62 1.47 0.97
where, dc is deceleration (m/s2 ) at speed v (m/s)
k1 , k2 , k3 , k4 , k5 ,k6 ,k7 ,k8 ,k9 , α, and β are the model parameters

385 The calibrated model plots along with idealized deceleration-speed relationship for differ-
386 ent vehicles are presented in Figure 11. It can be seen from Figure 11 that modelled deceleration-
387 speed relationship can satisfactorily reproduces the idealized deceleration-speed relationship.

(a) Truck, Motorized three and two wheeler (b) Diesel car and petrol car

FIGURE 11 : Model deceleration plots of motorized three wheelers, motorized two wheelers,
diesel car and petrol car

388 Model Diagnostic


389 Similar to acceleration modelling, proposed models for both regimes (single regime in case of
390 diesel and petrol car) are statistically evaluated using various diagnostic tools like graphical (resid-
391 ual plots, observed versus predicted deceleration plots) and numerical (hypothesis testing).
Bokare-Maurya 20

392 Graphical Procedure

393 Residual versus predicted plots The residues are calculated using observed and predicted
394 deceleration for both regimes (single regime in case of diesel and petrol cars). These residues are
395 then plotted against predicted deceleration values. For satisfying the assumption of normality of
396 residues (which is an important assumption in linear regression). Figure 12 presents such residual
397 plots of motorized three and two wheelers and diesel and petrol car.

0.2 0.4
0.3
0.1 0.2
Residues

Residues
0.1
0 0
0.1 0.3 0.5 0.7 -0.1 0.1 0.3 0.5 0.7 0.9 1.1

-0.1 -0.2
-0.3
-0.2 -0.4
Predicted deceleration, m/s2 Predicted deceleration, m/s2

(a) Truck (b) Motorized three wheeler

0.2 1
0.8
0.15
0.6
0.1 0.4
0.2
Residues

0.05
Residues

0
0 -0.2 0.5 1.1 1.7
-0.05 0.2 0.4 0.6 0.8 -0.4
-0.6
-0.1
-0.8
-0.15 -1
-0.2 -1.2
Predicted deceleration, m/s2 Predicted deceleration, m/s2

(c) Motorized two wheeler (d) Diesel car

1.1
0.9
0.7
0.5
0.3
0.1
-0.1 0 0.5 1 1.5 2 2.5
-0.3
-0.5

(e) Petrol car

FIGURE 12 : Residual plots of deceleration for motorized three and two wheeler and diesel and
petrol car

398 Residual plots of all vehicle types indicate that the residues are more or less uniformly dis-
399 tributed over the predicted values, depicting uniform variance. In case of motorized three wheeler
400 the residues associated with lower deceleration values are densely located indicating that drivers
401 use lower deceleration more often than higher deceleration. Also the residues associated with
402 higher deceleration values are somewhat higher than that associated with lower deceleration val-
403 ues. This shows that the model performs well at the lower deceleration than at higher deceleration.
404 The residual plots for motorized two wheeler are somewhat nearer to null plot ((16)) indi-
405 cating satisfactory performance of model throughout deceleration manoeuver.
406 In case of diesel car the model has mixed performance at middle and lower deceleration
407 values. The model however performs well for higher deceleration values indicated by near null
Bokare-Maurya 21

408 plot at higher deceleration values. Same argument can also be associated to petrol cars.

409 Plot of observed and predicted deceleration Figure 13 presents the plots of observed and
410 predicted deceleration. Since the data points are clustered around the 45 degree line indicating that
411 residuals are approximately normally distributed.

0.8 0.6
Predicted Deceleration (m/s2)

0.7 0.5
0.6

Predicted Values
0.4
0.5
0.4 0.3
0.3
0.2
0.2
0.1 0.1
0 0
0 0.2 0.4 0.6 0.8 0 0.1 0.2 0.3 0.4 0.5 0.6
Observed Avg. Deceleration (m/s2) Observed Values

(a) Motorized three wheeler (b) Motorized three wheeler

3 0.9
0.8
0.7

Predicted Value
Predicted value

2 0.6
0.5
0.4
1 0.3
0.2
0.1
0
0
0 0.3 0.6 0.9
0 0.5 1 1.5 2 2.5 3
Observed value Observed Value

(c) Motorized two wheeler (d) Diesel car (e) Petrol car

FIGURE 13 : Observed versus predicted deceleration for motorized three and two wheeler and
diesel and petrol car

412 The plots indicates close clustering around 45◦ line indicating that observed and predicted
413 values of deceleration have good amount of agreement.

414 Plot of observed and predicted trajectories The observed and modeled trajectories are
415 plotted and the resulting plots are presented in Figure 14. The plot shows satisfactory matching of
416 observed and modeled trajectories in all vehicles.

417 Plot of observed and predicted speed The observed and modeled speeds are plotted and
418 the resulting plots are presented in Figure 15. The plot is indicative that the observed and modeled
419 speed are matching each other sufficiently.
420 It is seen from the above illustrations and discussions that the models satisfactorily replicate
421 the observed values of speed, deceleration and position (trajectory). Hence it is concluded that the
422 proposed models (negative exponential for regime-I and linear for regime-II and second order
423 polynomial) can be used as a surrogate to physical system.

424 CONCLUSIONS
425 This study reported the acceleration and deceleration behaviour of various vehicle types running
426 on roads in India. Following are the salient features of this study;
Bokare-Maurya 22

16 14 Observed Position, m
Observed Position, m Modeled Position, m
14 12
12
10

Position, m

Position, m
10
8
8
6
6
4 4

2 2

0 0
0 5 10 15 0 5 Time, s 10 15
Time, s

(a) Truck (b) Motorized three wheeler

18 Observed Position, m Modeled Position, m 60 Observed position


16
50 Predicted
14 position, diesel car
Position, m

12

Distance, m
40
10
8 30
6 20
4
2 10
0 0
0 5 10 15 20 0 5 10 15
Time, s Time. S

(c) Motorized two wheeler (d) Diesel car

180
Predicted position

petrol car
120
Distance, s

Observed position

60

0
0 2 4 6 8 10 12
Time, s

(e) Petrol car car

FIGURE 14 : Observed versus predicted trajectories during deceleration for motorized three and
two wheeler and diesel and petrol car

427 Acceleration Behaviour


428 1. The rate of acceleration was found to increase from lowest value to maximum value with
429 increase in initial speed. After attaining maximum value, acceleration rate decreased
430 with further increase in speed. This nature is witnessed for all vehicles but trucks. Since
431 trucks achieve maximum acceleration quickly, the initial behaviour of acceleration rate
432 couldn’t be observed with GPS device having 1 second logging interval. The device
433 with higher data logging interval may capture this behaviour.

434 2. The maximum acceleration rate observed for various vehicle types are; for truck 1.0
435 m/s2 , for motorized three-wheeler 0.64 m/s2 , for motorized two-wheeler 1.95 m/s2 ,
436 for diesel car 2.23 m/s2 and for petrol car 2.87 m/s2 , Petrol car posted highest max-
437 imum acceleration rate. These rates are comparable with the rates reported by Bham
438 and Benekohal (8), Wang et al. (32), whereas these rates are higher than that reported
439 by Arasan and Koshy (5), Dey and Biswas (12), RaiChowdhury and Rao (24).

440 3. The acceleration-speed relationship is modelled as a dual regime relationship (negative


441 exponential in regime-I before attaining maximum acceleration and second order poly-
442 nomial for regime-II after attaining maximum acceleration rate) for motorized three and
Bokare-Maurya 23

16
Observed speed, m/s Modeled speed, m/s
14 Observed speed, m/s Modeled speed, m/s
14 12
12
10

Speed, m/s

Speed, m/s
10
8
8
6
6
4 4
2 2
0 0
0 2 4 6 8 10 12 14 16
0 5 Time, s 10 15
Time, s

(a) Truck (b) Motorized three wheeler

20
Observed Speed, m/s Modeled speed, m/s 30
18 Diesel car
16 predicted speed
25
14 Diesel car
Speed, m/s

20

Speed, m/s
12 Observed speed
10
15
8
6 10
4
2 5
0 0
0 5 10 15 20 0 2 4 6 8 10 12
Time, s Time, s

(c) Motorized two wheeler (d) Diesel car


30
petrol car predicted
speed
25
petrol car observed
speed
20
Speed, m/s

15

10

0
0 2 4 6 8 10 12
Time, s

(e) Petrol car car

FIGURE 15 : Observed versus predicted speed plots during deceleration for motorized three and
two wheeler and diesel and petrol car

443 two wheeler. For truck, diesel car and petrol car a single regime negative exponential
444 model is proposed.

445 4. Models proved fairly accurate when various statistical model diagnostic tests are ap-
446 plied.

447 5. In cars, in a particular gear initially, acceleration increases with speed till acceleration
448 attains its maximum value, afterwards it decreases with further increase in speed. Simi-
449 lar acceleration behaviour is observed in every gear during driving.

450 Deceleration Behaviour


451 1. The distance travelled and time taken to complete deceleration maneuver by various
452 vehicle types during deceleration manoeuver is different and is found to vary with the
453 speed at which driver start decelerating (approach speed). Driver takes more distance
454 for decelerating if the approach speed is high.

455 2. The speed at which driver attains maximum deceleration (referred as critical speed),
456 changes with vehicle type and approach speed. Critical speed also increases with ap-
Bokare-Maurya 24

457 proach speed. This indicates that at higher approach speed, the drivers achieve their
458 maximum deceleration rate quickly to stop at the earliest.

459 3. The proposed models are dual regime models for truck, motorized three wheeler and
460 motorized two wheelers. For diesel and petrol cars the single regime polynomial models
461 are proposed.

462 4. Various statistical tests are applied to check the effectiveness of models for observed
463 deceleration-speed data. It was found that the models tested well for all vehicle types.
Bokare-Maurya 25

464 REFERENCES
465 [1] AASHTO (2004). A Policy on Geometric Design of Highways and Streets, Green Book. Amer-
466 ican Association of State Highway and Transportation Officials.

467 [2] AASHTO (2011). A Policy on Geometric Design of Highways and Streets, Green Book. Amer-
468 ican Association of State Highway and Transportation Officials.

469 [3] Akcelik, R. and D. Biggs (1987). Acceleration profile models for vehicles in road traffic.
470 Transportation Science 21, No.1, 36–54.

471 [4] Akçelik, R. and M. Besley (2001). Acceleration and deceleration models. In 23rd Conference
472 of Australian Institutes of Transport Research (CAITR 2001), Monash University, Melbourne,
473 Australia, 10-12 December.

474 [5] Arasan, V. T. and R. Z. Koshy (2005). Methodology for modeling highly heterogeneous traffic
475 flow. Journal of Transportation Engineering, ASCE 131, 544–551.

476 [6] Belz, N. and L. Aultman-Hall (2011). Analyzing the effect of driver age on operating speed
477 and acceleration noise using on-board second-by-second driving data. Transportation Research
478 Board Annual Meeting, 1–17.

479 [7] Bennet, C. R. and R. Dunn (1995). Driver deceleration behaviour on a freeway in new zealand.
480 Transportation Research Record: Journal of Transportation Research Board 1510, 70–74.

481 [8] Bham, G. and R. Benekohal (2002). Development evaluation and comparison of acceleration
482 models. In Annual Conference of Transportation Research Board, Washington D.C. USA, pp.
483 1–42.

484 [9] Bonesson, J. (1992). Modeling queued driver behaviour at signalized junction. Transportation
485 Research Record 1365, 99–107.

486 [10] Carcary, W. and F. Murray (2001). A new driver project. Technical report, Scottish Executive
487 Central Research Unit, Edinburgh, UK.

488 [11] Dey, P., S. Chandra, and S.Gangopadhyay (2008). Speed studies on two lane indian highways.
489 Indian Highways 36(6), 9–18.

490 [12] Dey, P. P. and P. Biswas (2011). Acceleration of queue leaders at signalized intersections.
491 Indian Highways No. 3(1), 49–54.

492 [13] Dockerty, A. (1966). Accelerations of queue leaders from stop lines. Traffic Engineering and
493 Control Vol. 8, No. 3, 150Ű155.

494 [14] El-Shawarby, I., H.Rakha, A.Amer, and C. McGhee (2011). Driver and surrounding traf-
495 fic impact on vehicle deceleration behavior at the onset of a yellow indication. TRB Annual
496 Meeting, 1–20.

497 [15] Fildes, B. (1991). Older drivers safety and mobility: Current and future issues. Transporta-
498 tion Research Record, Part F Journal of Transportation Research Board 9, 305–312.
Bokare-Maurya 26

499 [16] Freund, R. and W. Wilson (2011). Statistical Methods (3rd ed.). ACADEMIC PRESS,
500 California, USA.

501 [17] Gattis, J., M. A. Bryant, and L. K. Duncan (2010). Truck acceleration speeds and distances
502 at weigh stations. Transportation Research Record: Journal of the Transportation Research
503 Board, National Academies, Washington, D.C. USA. 2195, 20Ű26.

504 [18] Glauz, W., S.Cobet, and D. Harwood (1980). Projected vehicle characteristics. Transporta-
505 tion Research Re 772, 37–44.

506 [19] Haas, R., V. Inman, A. Dixson, and D. Warren (2004). Use of intelligent transportation sys-
507 tem data to determine driver deceleration and acceleration behavior. Transportation Research
508 Record 1899, 3–10.

509 [20] ITE (2009). Traffic Engineering Handbook. Institute of Transportation Engineers.

510 [21] Long, G. (2000). Acceleration characteristics of starting vehicles. Transportation Research
511 Record 1737, 58–70.

512 [22] Maurya, A. and P. Bokare (2012). Study of deceleration behaviour of different vehicle types.
513 International Journal of Traffic and Tra 2(3), 253–270.

514 [23] Mehmood, A. (2009). Determinants of speeding behaviour of drivers in al ain (united arab
515 amirates). ASCE Journal of Transportation Engg. 135, No 10, 721–729.

516 [24] RaiChowdhury, M. and S. Rao (1989). Acceleration characteristics of vehicles at signalized
517 intersection. Journal of Institution of Engineers (India) 70, 35–38.

518 [25] Rakha, H. and Y. Ding (2003). Impct of stops on vehicle fuel consumption and emissions.
519 Journal of Transportation Engineering ASCE Vol. 129, N0-1, 23–32.

520 [26] Rakha, H., F. Dion, and H. Sin (2001). Global positioning system data for field evaluation of
521 energy and emission impact of traffic flow improvement projects. issues and proposed solutions.
522 Transportation Research Record 1768, 221–230.

523 [27] Rakha, H., M. Snare, and F. Dion (2004). Vehicle dynamics model for estimating maximum
524 light-duty vehicle acceleration levels. Transportation Research Record No. 1883, 40–49.

525 [28] Samuels, S. and J. Jarvis (1978). Acceleration and deceleration of modern vehicles. Technical
526 report, Australian road research institute, Report 86.

527 [29] Searle, J. (1999). Equations of speed, time and distance for vehicles under maximum accel-
528 eration. Advances in Safety Technology SP-1433, 1–7.

529 [30] Snare, M. (2002). Dynamic model for predicting maximum and typical acceleration rates of
530 passenger vehicles. Master’s thesis, Virginia Polytechnic Institute and State University.

531 [31] St.John, A. (1989). Nchrp report::-grade effect on traffic flow stability and capacity. Technical
532 report, Transportation Research Board.
Bokare-Maurya 27

533 [32] Wang, J., K. Dixon, H. Li, and J. Ogle (2004). Normal acceleration behavior of passenger
534 vehicles starting from rest at all-way stop-controlled intersections. Transportation Research
535 Record 1883, 158–166.

536 [33] Wang, J., K. Dixon, H. Li, and J. Ogle (2005). Normal deceleration behavior of passenger ve-
537 hicles at stop signŰcontrolled intersections evaluated with in-vehicle global positioning system
538 data. Transportation Research Record No. 1937, 120Ű127.

539 [34] West, R., D. French, D. Kemp, and R. Elander (1993). Direct observation of driving, self
540 reports of driver behaviour and accident involvement. Ergonomics 36(5), 557–567.