University of California Transportation Center
UCTC- FR- 2010- 28
Simulation Evaluation of Green Driving Strategies Based on
Inter- Vehicle Communications
Hao Yang, Daji Yuan, Wen- Long Jin,
and Jean- Daniel Saphores
University of California, Irvine
August 2010
1 SIMULATION EVALUATION OF GREEN DRIVING STRATEGIES BASED ON
2 INTER- VEHICLE COMMUNICATIONS
3 HAO YANG
4 Ph. D. Student
5 Department of Civil and Environmental Engineering
6 Institute of Transportation Studies
7 University of California, Irvine
8 Irvine, CA 92697- 3600
9 Email: hyang5@ uci. edu
10 DAJI YUAN
11 Ph. D Student
12 Civil and Environmental Engineering
13 Institute of Transportation Study
14 University of California, Irvine
15 Irvine, CA 92697- 3600
16 Email: dajiy@ uci. edu
WEN- LONG JIN1 17
18 Assistant Professor
19 Civil and Environmental Engineering
20 Institute of Transportation Study
21 University of California, Irvine
22 Irvine, CA 92697- 3600
23 Email: wjin@ uci. edu
24 JEAN- DANIEL SAPHORES
25 Associate Professor
26 Civil and Environmental Engineering
27 Institute of Transportation Study
28 University of California, Irvine
29 Irvine, CA 92697- 3600
30 saphores@ uci. edu
31 Word Count: 5000+ 250  9= 7250
32 August 1, 2010
33 SUBMITTED TO 2011 TRB ANNUAL MEETING
1Author for correspondence
1
34 ABSTRACT
35 Transportation system produces a large percentage of local pollutants including hydrocarbons
36 ( HC), carbon monoxide ( CO), carbon dioxide ( CO2), and oxides of nitrogen ( NOx), etc. Apart
37 from switching to alternative fuels, one measure would be to apply information and communi-38
cation technologies to help us drive more smoothly so as to decrease pollutants emissions. This
39 paper studies potential benefits of two green driving strategies based on inter- vehicle communica-40
tion ( IVC). Here green driving strategies are similar to intelligent speed adaptation , but we assume
41 that an IVC- equipped vehicle is able to receive detailed trajectory information from other such ve-42
hicles with the help of IVC. For the purpose of evaluation, we integrate Newell’s car- following
43 model and VT- Micro to establish a simulation platform. Market penetration rates of IVC- equipped
44 vehicles and delivery delays of messages are two prominent features of IVC systems. We simulate
45 stop- and- go traffic to calculate potential reductions in air pollutant emissions and fuel consumption
46 under different market penetration rates and delivery delays. Results show that significant savings
47 under frequent stop- and- go traffic conditions may be obtained with our strategies ( HC: - 88.3%,
48 CO: - 95.8%, NOx: - 91.5%, CO2: - 36.3%, Fuel Consumption: - 71.3%) for the same travel time
49 and almost the same overall travel distance. It is also shown that relatively large savings can be
50 achieved even for a market penetration rate as low as 1% and communication delays larger than
51 2 minutes. In the future we will investigate environmental benefits of green driving strategies for
52 more traffic scenarios and realistic communication scenarios.
53 Keywords: Green Driving, Emissions, Fuel Consumption, VT- micro Emission Model, Newell
54 Car- following Model, Intelligent Adaptation System, Inter- vehicle Communication
2
55 1 INTRODUCTION
56 According to the Energy Information Administration, the transportation sector in the U. S. is re-57
sponsible for one third of CO2 emissions, over one half of NO2 emissions, and over three quarters
58 of CO emissions. Globally, the situation is worsening with the rapid development of motor vehicle
59 transportation in developing countries. It is well known that excessive speed ( defined as exceeding
60 the posted speed limit or driving too fast for ambient conditions [ 1]) and stop- and- go traffic on road
61 can significantly increase fuel consumption and vehicle emissions [ 2]. Many strategies have been
62 proposed to address this problem. In [ 3], effects of speed bumps on road were investigated to con-63
trol speed. In [ 4], effects of police enforcement were studied to monitor speed in road. However,
64 both of these two traditional methods have proven to have only moderate effects on controlling
65 excessive speed [ 2].
66 In the past few years, many telecommunications and information technologies have been adopted
67 by drivers to improve their daily driving experience. For example, the sales of global positioning
68 systems ( GPS) units are up 488% during the holiday season of 2008 [ 5], and adaptive cruise control
69 systems have helped drivers reduce their workload and its associated stress [ 6]. In the near future,
70 with the development of IntelliDrive technologies, especially inter- vehicle communications ( IVC),
71 including vehicle- to- vehicle and vehicle- to- infrastructure communications, will be available to re-72
lay time- critical and location- based traffic information between vehicles so that people can drive
73 more smoothly and more safely. As the number of cars equipped with these technologies increases,
74 we expect that drivers will adapt their behaviors accordingly [ 7]. Such collective behavior changes
75 will result in different traffic flow characteristics, transportation systems performance, and envi-76
ronmental impacts. Therefore instead of using alternative fuels[ 8] and traditional methods, new
77 technologies, such as IVC and cooperative autonomous cruise control ( CACC) can also be used
78 to improve traffic flow, fuel consumption economy, and reduce emissions.
79 Inter- vehicle Communications ( IVC) can help establish self- organized, decentralized, real-
3
time 80 traffic information systems. Many studies are underway to investigate IVC based on mobile ad
81 hoc networking technology as a mean of developing the ” internet on the road” [ 9, 10]. In [ 11, 12,
82 13], researchers describe potential applications and properties of Autonet, including connectivity,
83 adthe impacts of market penetration rate ( MPR), and delivery delay, and the effect of IVC on
84 vehicle travel time. However, there have been no systematic studies of potential environmental
85 benefits of IVC.
86 In the literature there have been studies on intelligent speed adaptation ( ISA) strategies to
87 smooth traffic. ISA systems use aggregate- level road congestion information to adjust speed limits
88 of vehicles on specific road sections [ 2]. Moreover, ISA systems monitor vehicle speeds and
89 current traffic conditions, and, based on these, they provide corrective actions or optimal process
90 for drivers. The information collected for ISA is usually obtained from loop- detectors or on-91
board sensors. In ISA systems, there are three basic methods to adjust speed limits [ 14, 2]:
92 fixed, variable and dynamic. Such speed limits could be implemented in advisory, voluntary, or
93 mandatory fashions [ 14, 2].
94 Recently, a lot of energy has been devoted to testing the impacts of ISA systems, including on
95 road injuries, the release of air pollutants and fuel consumption. These studies have shown that
96 ISA can effectively improve safety and reduce traffic congestion. In [ 2], a set of speed limits and
97 corrective actions are communicated to drivers. In addition, ISA has potential to mitigate conges-98
tion by smoothing the dynamics of congested traffic. ISA equipped vehicle has a much smoother
99 trajectory ( with smaller speed variation), which leads to fuel savings and emission reductions. Re-100
sults show that ISA can reduce fuel consumption up to 70 percent, and cut emissions of CO, HC
101 and NOx by 93 percent, 90 percent and 86 percent respectively. Even for realistic traffic adjust-102
ment, fuel consumption can decrease by 13 percent. But it was shown that ISA systems could
103 increase travel time by a small percentage ( 6 percent). In addition, many field implementations of
104 ISA systems have been done to test the influence of ISA on traffic safety and the environment. ISA
105 experiments in Tilburg ( Netherlands) show that with speed limits control, driving is much safer
4
106 [ 15, 16, 17]. Road injuries are reduced by 15 to 20 percent, and carbon dioxide emissions are
107 reduced by approximately 11 percent. Moreover, in [ 18], with optimal speed limits adjustment,
108 freeway traffic conditions are more stable, which also benefits to driving safety and reduces air
109 pollutants emissions. Another concern with ISA systems is that they may worsen road congestion.
110 However, in [ 19, 20], Liu and Tate show that under very congested levels, ISA does not change
111 traffic conditions.
112 In this paper, we will study green driving strategies based on IVC and, in particular, their effects
113 in emission reductions and fuel consumption savings in different traffic conditions. We assume
114 that IVC- equipped vehicles share their trajectories with each other. With detailed information of
115 vehicles’ trajectories, we propose two green driving strategies to maximize energy efficiency of
116 vehicles in road. These green driving strategies are similar to ISA schemes, but, with the help of
117 IVC, vehicles can get information from other vehicles directly. One advantage of such IVC- based
118 green driving strategies is that vehicles can get more relevant information. But such strategies could
119 be restricted by limited market penetration rates of IVC devices and communication delays. To
120 evaluate the potential benefits of such green driving strategies, Newell’s car- following model [ 21]
121 and the VT- Micro emission model [ 22] are integrated into a simulation platform.. Newell’s model
122 is a trajectory translation model, which is based on vehicle locations. The model is simple and
123 straightforward to describe vehicle movement in traffic streams. Moreover, studies in [ 23] show
124 that Newell’s model matches realistic vehicle trajectories very well. So, with proper parameter
125 values, using Newell model can lead to well description of traffic streams. Moreover, we can
126 change the desired speed of individual drivers based on green driving strategies. With the integrated
127 simulation model we then study environmental benefits of green driving strategies for different
128 market penetration rates and communication delays.
129 This paper is organized as follows. In Section 2, We first describe green driving strategies to
130 smooth traffic streams. In Section 3, we combine a microscopic traffic model with an emission
131 model to study green driving strategies. In Section 4, we summarize results from our simulations
5
132 that study strategies described in Section 2. Section 5 presents concluding remarks.
133 2 GREEN DRIVING STRATEGIES BASED ON INTER- VEHICLE
134 COMMUNICATIONS
135 In this section, we first present two IVC- based green- driving strategies and then introduce commu-136
nication delays to the strategies.
137 2.1 Two Green Driving Strategies
138 With IVC- based green driving strategies, speed limits of allIVC- equipped vehicles are set based
139 the information they gather from other IVC- equipped vehicles. In this study we propose to set an
140 IVC- equipped vehicle’s desired speed to the average speed calculated from other IVC- equipped
141 vehicles.
142 Suppose that N vehicles run in a selected region of one traffic network and market penetra-143
tion rate of IVC equipped vehicles is p, the number of IVC- equipped vehicles is n ( where expected
144 value of n is Np). For all vehicles with new technologies, we have two methods to set the speed lim-145
its under global and sectional traffic information. Then, based on messages broadcasted among ve-146
hicles, speed limits are modeled in the following ways. Suppose that spacing of all IVC- equipped
vehicles are sIVC
1 ( t ) ; sIVC
2 ( t ) ; ; sIVC
n ( t ) , and their speeds at time t are vIVC
1 ( t ) ; vIVC
2 ( t ) ; ; vIVC
147 n ( t ) , then
148 dynamic speed limits at time t can be modeled in the following two models. The first model is
149 based on average value of speeds collected by IVC- equipped vehicles.
vlk ( t ) =
å ni
= 1 vIVC
i ( t )
n
( 1)
150 where vlk ( t ) is the speed limit set for equipped vehicle k at time t. The second model is based on
151 desired average speed of all IVC- equipped vehicles. The speed limit of equipped vehicle i is set as
6
152 following equation.
vlk ( t ) =
å ni
= 1 ( sIVC
i ( t ) 􀀀 dj )
å n i = 1 t i
( 2)
153
154 Both of the two strategies have their own advantages. For average speed adjustment, only
155 speed information is necessary to be delivered, and speed can be obtained directly from vehicle
156 engines or GPS devices. While for desired speed adjustment, we need more information, including
157 time gap and jam spacing, which are only available when distance sensors are installed. However,
158 the first model only considers global information, and the second strategy also incorporates local
159 information. Therefore theoretically the second strategy should be more robust.
160 Additionally, with different considerations of network scales, speed limit adjustments are dif-161
ferent because of various numbers of equipped vehicles. In this paper, we set two different levels
162 of network scales: whole network ( global), downstream section ( sectional). When we study the
163 whole network, all equipped vehicles in transportation system communicate with each other. They
164 share traffic information to improve the entire network conditions. The more advanced strategy is
165 identifying dynamic speed limits for individual vehicles. The settings are based on downstream
166 flow for a given vehicle. All IVC- equipped vehicles in selected region deliver information to the
167 chosen vehicle and speed limits for this individual vehicle are calculated based on this informa-168
tion. Those two adjustments work under the two different network scales ( global and sectional)
169 and satisfy different planning targets.
170 2.2 Communication Delays
171 Using information provided by IVC, many interesting and useful applications have recently
172 been studied, such as information warning system, traffic control, or cooperative assistance systems
173 [ 24]. The most general platforms to apply IVC systems are cellular networks and mobile wireless
7
174 networks. Different applications of ICC depend differently on properties of the vehicular network.
175 For example, communication delays affect the quality of message delivery. For cellular networks,
176 the accuracy of traffic system delay, caused by communication is treated as constant value [ 25].
177 By contrast, for mobile ad hoc wireless network, communication delays dependent on routing
178 protocols and vehicle distributions [ 26]. . In [ 12], it was shown that delivery delay is highly
179 related to routing protocols, flow- rates, and market penetration rates. The conclusion of delay for
180 IVC system is that delay = 1 = ( f lowrate  IVC ( % ) ) . For environmental applications, delay is an
181 important consideration because the amount of pollutants released is sensitive to vehicles speed
182 and acceleration, which may change a lot during a non- trivial delay. Green driving strategies in
183 this paper process delay as a key parameter.
184 In this study, we consider two types of delays: constant delays, and delays linearly proportional
185 to distances between vehicles. The first type of delays could occur when IVC are enabled with
186 cellular networks, where communication delays are not sensitive to distance in a relatively small
187 region for applying green driving strategies. The second type of delays could occur when IVC are
188 enabled with instantaneous or delay- tolerant multi- hop ad hoc communications.
189 If we denote Di ! k ( t ) as the delay of the information from vehicle i to vehicle k at t, then the
190 two green driving strategies can be written as
v ( 1 )
lk ( t ) =
å ni = 1 vIVC
i ( t 􀀀 Di ! k ( t ) )
n
; ( 3)
v ( 2 )
lk ( t ) =
å ni
= 1 ( sIVC
i ( t 􀀀 Di ! k ( t ) ) 􀀀 di )
å n i = 1 t i
: ( 4)
191 For constant delays, Di = D ; i = 1 ; 2 ;    ; n : For delays linearly proportional to the distance,
Di ! k ( t ) = d  j xIVC
i ( t ) 􀀀 xIVC
k ( t ) j 8 i = 1 ; 2 ;    ; n ( 5)
8
192 where d is the coefficient of delay with distance between two equipped vehicles.
193 3 AN INTEGRATED SIMULATION MODEL
194 3.1 Traffic Flow Model
195 Newell’s car- following model [ 21] is the simplest car- following model as it focuses on predicting
196 vehicle trajectories. It assumes that a following vehicle tries to minimize its distance from its
197 leading vehicle in congested traffic. And in free traffic, vehicle always keeps the free flow speed.
198 Equation 6 describes this driving rule in congested and free traffic. We set the speed limit as vli
199 for vehicle i, then from [ 27], we get Newell- Daganzo Car- following Model.
xi ( t + t i ) = min f xi 􀀀 1 ( t ) 􀀀 di ; xi ( t ) + vli  t i g ( 6)
200 where, vehicle i 􀀀 1 is the leader of vehicle i.
201 3.2 Emission Model
202 In 2004, Rakha et al [ 22] presented the Virginia Tech Microscopic energy and emission model
203 ( VT- Micro), which was developed to predict the emissions of different air pollutants for different
204 vehicle classes using statistical models that relay on speed and acceleration. Their typical model,
205 which was estimated via linear regression, linked the logarithm of a emission rate ( or a fuel con-206
sumption rate) with a simple polynomial that contains vehicle speed and acceleration.
logMOEe =
3 å
i = 0
3 å
j = 0
( Ke
i juiaj ) ( 7)
where MOE is an instantaneous fuel consumption or emission rate ( mg/ s), Ke
207 i j’s are regression
208 coefficients, u is a vehicle’s instantaneous speed ( km/ h), and a is its instantaneous acceleration rate
9
209 ( km/ h/ s).
210 3.3 Integrated Model
211 In this subsection, one integrated model with traffic flow and emissions models is described. Since
212 both Newell model and VT- Micro emission model are microscopic models, the connection be-213
tween these two models are straightforward. Figure 1 describes the flow chart of applying Newell
214 model and VT- Micro emission model to estimate emissions and fuel consumption under different
215 green driving strategies.
Figure 1: Flow Chart of Integrating Newell Model and VT- Micro Emission Model
216 The integrated model has four basic components: Initial traffic stream setting, traffic stream
217 simulation, speed limit adjustment with green driving strategies and emission estimation. In initial
10
218 traffic stream setting, distribution of initial vehicle speeds is arbitrarily provided, which is applied
219 to set the initial vehicle locations in road. Secondly, for traffic stream simulation, all vehicle trajec-220
tories are simulated with proper parameter settings ( e. g. time gap, jam spacing, speed limit, etc). In
221 the third components, historical trajectories are packed to be communicated between informed ve-222
hicles, which are used to adjust speed limits based on the strategies described in section 2. Finally,
223 with well adjusted vehicle trajectories, VT- Micro emission model helps to estimate emissions and
224 fuel consumption. With different green driving strategies, we compare emissions and fuel usage to
225 study effects of these new technologies.
226 4 SIMULATION
Table 1 Simulation Settings
227 In this study, we simulate traffic on a one- lane ring road with settings in Table 1. In our simu-228
lation, we set boundary of initial speed for all vehicles. All initial speed are randomly distributed
229 in the region of [ 1 􀀀 e ; 1 + e ]  v ¯ desired ( where v ¯ desired is the average speed calculated from overall
230 vehicle density in road). This initial setting is applied to make the traffic scenarios more reason-231
able and reduce extreme accelerations. In the simulation runs, e = 0 : 5. In addition, we assume that
11
232 all IVC- equipped vehicles rigorously comply with suggested speed limits through green driving
233 strategies.
234 4.1 Effect of Different Strategies and Network Scales
235 In section 2, we proposed two different strategies to maintain speed limits. These two strategies
236 make the speed variation smaller than that in non- Green Driving system. Figure 2 shows the
237 speed trajectory of one vehicle during half hour. In this figure, velocity trajectory of Green Driving
238 system applying desired speed ( red) is much smoother than that of normal non- Green Driving
239 system. Moreover, the actual average speeds of both scenarios are approximately same ( Green
240 Driving: 31 : 0km = hr, non- Green Driving: 31 : 1km = hr).
0 200 400 600 800 1000 1200 1400 1600 1800
15
20
25
30
35
40
45
50
Time ( seconds)
Speed ( km/ hr)
Non− Green Driving
Green Driving
Figure 2: Speed trajectories of non- Green Driving system ( blue) and Green Driving system
( red)
12
241 Furthermore, Table 2 lists the emissions and fuel consumption savings from these strategies.
242 The table indicates that applying green driving strategies, definitely, emissions and fuel consump-243
tion are reduced. In this example, all vehicles are IVC- equipped. We find that savings for different
244 types of emissions: HC: 87.90% - 88.33%, CO: 93.08% - 95.79%, NOx: 88.92% - 91.52%, CO2:
245 28.66% - 36.32%, FUEL USE: 68.47% - 71.31%. With random initial traffic, emission and fuel
246 consumption savings are large and green driving strategies are all efficient.
Table 2 Emissions and Fuel Consumption Obtained from Non- delay System with 100% MPR
247 Moreover, considering total vehicle distance traveled, desired speed adjustment has much bet-248
ter performance than that of average speed adjustment. Under average speed adjustment, total
249 distance traveled decreases more than 22%; while for desired speed adjustment, it is less than
250 0.7%. This difference comes from the properties of the strategy. As we explained in section 3,
251 average speed adjustment can leads large gap ahead of IVC- equipped vehicles. Figure 3 shows
252 several trajectories picked from our simulation. In Figure 3( a), when global average speed adjust-253
ment is applied, in front of IVC- equipped vehicles, large gap exists, but it does not accelerate to
254 approach its leader due to its lower speed limit. This gap does not appear on the second graph.
255 Since traveling speed takes an important role in transportation study, people will not accept any
13
new 256 strategies if they reduce traveling speed significantly. So, we claim that using desired speed
257 adjustment has better effect on transportation system.
a
0 50 100 150 200 250 300
0
500
1000
1500
2000
2500
3000
Time ( seconds)
Position ( meters)
nonIVC− equipped
IVC− equipped
nonIVC− equipped
IVC− equipped
IVC− equipped
b
0 50 100 150 200 250 300
0
500
1000
1500
2000
2500
3000
Time ( seconds)
Position ( meters)
nonIVC− equipped
IVC− equipped
nonIVC− equipped
IVC− equipped
IVC− equipped
Figure 3: Vehicle Trajectories ( a) Global Average Speed Adjustment, ( b) Global Desired
Speed Adjustment
258 4.2 Effect of MPR
259 In this subsection, the same initial traffic condition is set, but different market penetration rates
260 are proposed. Traffic scenarios based on global desired speed adjustment are simulated. Figure
261 4 shows speed and acceleration histograms of all vehicles during half hour. Figure 4( a) comes
262 from non- Green Driving system, while Figure 4( b) describes Green Driving system with 50% of
263 IVC- equipped vehicles. Comparing these two graphs, speed concentrates on a narrow region after
264 using global desired speed adjustment. It seems that the speed control scheme really works for
265 traffic stream.
266 Results of emission and fuel consumption savings are shown in Figure 5. As expected, emis-267
sion savings can gradually increase when we apply green driving strategy with global desired speed
268 adjustment. For HC, its reduction increases from 63.45% at 1% MPR to 89.0% at 100% MPR; for
269 CO, it increases from 67.32% to 94.0%; for NOx, it is from 60.60% to 89.9%; for CO2, it is from
14
a Velocity ( km/ hr)
Acceleration ( km/ hr/ sec)
0 20 40 60 80 100 120
− 5
− 4
− 2
0
2
4
6
8
10
12
13
0
1000
2000
3000
4000
5000
6000
7000
8000
b Velocity ( km/ hr)
Acceleration ( km/ hr/ sec)
0 20 40 60 80 100 120
− 5
− 4
− 2
0
2
4
6
8
10
12
13
0
1000
2000
3000
4000
5000
6000
7000
8000
Figure 4: Speed- Acceleration Histograms ( a) non- Green Driving system, ( b) Green Driving
system with 50% MPR
270 20.18% to 35.8%; and for fuel consumption, it is from 51.78% to 71.0%. All these reduction are
271 huge, but the improvements of reduction with MPR are not significant when MPR is greater than
272 20%. The cause of this observation is that we only apply car- following behaviors in our simula-273
tion, which leads to the situation that one IVC- equipped vehicle not only adjusts its own driving
274 behavior, but also affects its followers in road. After a while, location trajectories of both equipped
275 vehicle and its followers are all smoothed due to green driving strategies and car- following rule.
276 An important observation is that, even with an MPR as low as 1%, savings of emission and fuel
277 consumption can still be huge. We expect that, in real world, due to lane- changing and other activ-278
ities, savings at MPR’s may not be as high as 60%. But reasonable savings are still possible due to
279 car- following behaviors.
280 4.3 Effect of Communication Delay
281 In this subsection, MPR of IVC- equipped vehicles is 50% and constant delivery delay are assigned
282 for all vehicles. Considering delay, when larger delay in the communication system exists, the
15
0 10 20 30 40 50 60 70 80 90 100
− 100
− 90
− 80
− 70
− 60
− 50
− 40
− 30
− 20
− 10
0
IVC Penetration Rate (%)
Percent Change
HC
CO
NOx
CO2
FUEL USE
Figure 5: Emission/ Fuel Consumption Reductions at different MPR with Global Desired
Speed Adjustment
283 information vehicles receive is older and less useful for current speed limit adjustment. So, it
284 is straightforward to predict that high delay can reduce the effect of green driving strategies. In
285 these simulations, various delivery delays are assigned, and their effects on emission and fuel
286 consumption reductions are studied. Figure 6 verifies our prediction. With higher delivery delay,
287 all emissions and fuel consumption savings are increasing ( approximately 3- 5% for 150 seconds
288 delay).
289 Besides of effect of constant delay, linear delivery delay is another reasonable assumption. In
290 section 3, we assume a simple linear relationship between delay and distance ( Equation 5). It
291 is obvious that higher delay causes less emissions and fuel consumption reductions. We expect
292 that with larger coefficient d , savings under sectional desired speed adjustment is smaller, because
16
0 20 40 60 80 100 120 140 160
− 100
− 90
− 80
− 70
− 60
− 50
− 40
− 30
− 20
− 10
0
Time Delay ( seconds)
Percent Change
HC
CO
NOx
CO2
FUEL USE
Figure 6: Emission/ Fuel Consumption Reductions at different MPR with Global Desired
Speed Adjustment
293 higher coefficient value leads higher delay for all IVC- equipped vehicles. In simulations, various
294 d values are set: f 0 ; 0 : 005 ; 0 : 01 ; 0 : 02 ; 0 : 05 g second = meter, and savings of all fives emissions and
295 fuel consumption are calculated. From Figure 7, we observe the decreasing trend of reductions
296 with coefficient. Combining with constant delay analysis, we claim that larger delay actually
297 makes the reduction smaller.
298 5 CONCLUSION
299 In this paper, we investigated the effect of green driving strategies based on inter- vehicle communi-300
cation system. Two important factors of IVC systems, market penetration rate and communication
17
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
− 100
− 90
− 80
− 70
− 60
− 50
− 40
− 30
− 20
− 10
0
Coefficient d
Percent Change
HC
CO
NOx
CO2
FUEL USE
Figure 7: Emissions/ Fuel Consumption Reduction at Different Coefficients with Sectional
Desired Speed Adjustment
delay, 301 were studied. We made two major conclusions from this work. Firstly, with higher market
302 penetration rate ( MPR) of IVC- equipped vehicles, reduction of emissions and fuel consumption
303 were larger. This conclusion was reasonable, since higher MPR leaded to more communication
304 and large amount of information, which could help us to find even more accurate and optimal ad-305
justment of speed limits to achieve less emissions and fuel consumption. The second conclusion
306 was that with the effect of communication delay, savings of emissions and fuel consumption were
307 reduced. Larger delay made more information useless. Then, speed adjustment would not be ac-308
curate enough, and this leaded to less smoother traffic, which was equivalent to less reduction of
309 emissions and fuel consumption.
18
310 But the more important insight with the impacts of market penetration rates and communication
311 delays is that, even with a very low market penetration rate ( 1%) and a large communication delay
312 ( > 2min), we can still achieve significant savings for frequent stop- and- go traffic. This feature
313 is very promising, since it means that such green driving strategies can work even with a small
314 adoption rate. This is different from traditional approaches, e. g., with alternative fuels, which
315 require high market penetration rate to achieve significant savings.
316 In the future, we will investigate the potential benefits of such green driving strategies for dif-317
ferent traffic conditions. In this study, as shown in Figure 2, the frequency of stop- and- go traffic
318 is very high, but in reality it is usually smaller. We will investigate impacts of the frequencies on
319 emission savings in future studies. In this paper, homogeneous traffic is modeled. We want to
320 extend our strategies to non- homogeneous traffic and evaluate savings of emission and fuel con-321
sumption. And, also when we apply ISA system, 100% acceptance rate are assigned to equipped
322 vehicles, which is not obtainable in realistic world. So, in future, we can simulate the traffic and
323 communication system with reasonable acceptance rate. Furthermore, since only arbitrary com-324
munication properties are assigned in this paper, the result may not match to realistic situation. So,
325 it is important to simulate transportation system with some reasonable communication settings.
326 Finally, the application of these green driving strategies should be tested in real world situation.
327 ACKNOWLEDGEMENT
328 This study is supported in part by a grant from University of California Transportation Center.
329 References
330 [ 1] A. Clayton. An accident- based analysis of road- user errors, 1972.
331 [ 2] O. Servin, K. Boriboonsomsin, and M. Barth. An energy and emissions impact evaluation of
19
332 intelligent speed adaptation. In IEEE Intelligent Transportation Systems Conference, 2006.
333 ITSC’ 06, pages 1257– 1262, 2006.
334 [ 3] M. Pau and S. Angius. Do speed bumps really decrease traffic speed? An Italian experience.
335 Accident Analysis & Prevention, 33( 5): 585– 597, 2001.
336 [ 4] T. Vaa. Increased police enforcement: effects on speed. Accident Analysis & Prevention, 29
337 ( 3): 373– 385, 1997.
338 [ 5] Wolfe, J. . Breakthrough Technologies for 2008. URL http:// www. forbes.
339 com/ 2008/ 01/ 17/ nanosphere- honda- qualcomm- lead- clayto% n- in_ jw_
340 0116claytonchristensen_ inl. html. Accessed on February 2, 2009.
341 [ 6] N. A. Stanton and M. S. Young. Driver behaviour with adaptive cruise control. Ergonomics,
342 48( 10): 1294– 1313, 2005.
343 [ 7] M. Hoedemaeker and KA Brookhuis. Behavioural adaptation to driving with an adaptive
344 cruise control ( ACC). Transportation Research Part F: Psychology and Behaviour, 1( 2): 95–
345 106, 1998.
346 [ 8] T. A. Semelsberger, R. L. Borup, and H. L. Greene. Dimethyl ether ( DME) as an alternative
347 fuel. Journal of Power Sources, 156( 2): 497– 511, 2006.
348 [ 9] P. Morsink, R. Hallouzi, I. Dagli, C. Cseh, L. Sch
349 ” afers, M. Nelisse, and D. de Bruin. Cartalk 2000: Development of a cooperative adas based
350 on vehicle- to- vehicle communication. Intelligent Transport Systems and Services, 2003.
351 [ 10] W. Franz, R. Eberhardt, and T. Luckenbach. Fleetnet- internet on the road. In Proc. 8thWorld
352 Congress on Intelligent Transport Systems, 2001.
20
[ 11] 353 W. Recker, W. L. Jin, X. Yang, and J. Marca. Autonet: inter- vehicle communication and
354 network vehicular traffic. International Journal of Vehicle Information and Communication
355 Systems, 1( 3): 306– 319, 2008.
356 [ 12] X. Yang and W. Recker. Simulation studies of information propagation in a self-357
organizing distributed traffic information system. Transportation Research Part C: Emerging
358 Technologies, 13( 5- 6): 370– 390, 2005.
359 [ 13] W. L. Jin andW. W. Recker. Instantaneous information propagation in a traffic stream through
360 inter- vehicle communication. Transportation Research Part B: Methodological, 40( 3): 230–
361 250, 2006.
362 [ 14] OMJ Carsten and FN Tate. Intelligent speed adaptation: accident savings and cost- benefit
363 analysis. Accident Analysis & Prevention, 37( 3): 407– 416, 2005.
364 [ 15] L. Duynstee and G. Martens. Effects of intelligent speed adaptation on driving behaviour. In
365 Proceedings of the 7th World Congress on Intelligent Transport Systems, 2000.
366 [ 16] HL Oei and PH Polak. Intelligent speed adaptation ( isa) and road safety. IATSS RESEARCH,
367 26( 2): 45– 51, 2002.
368 [ 17] A. van Loon and L. Duynstee. Intelligent Speed Adaptation ( ISA): A successful test in the
369 Netherlands. In Proceedings of the Canadian Multidisciplinary Road Safety Conference XII,
370 London, Ontario, pages 10– 13, 2001.
371 [ 18] OH Jun- Seok and OH Cheol. Dynamic speed control strategy for freeway traffic congestion
372 management. Journal of the Eastern Asia Society for Transportation Studies, 6: 595– 607,
373 2005.
374 [ 19] R. Liu and J. Tate. Microsimulation modelling of intelligent speed adaptation system.
375 PTRC- PUBLICATIONS- P, pages 89– 100, 2000.
21
376 [ 20] R. Liu and J. Tate. Network effects of intelligent speed adaptation systems. Transportation,
377 31( 3): 297– 325, 2004.
378 [ 21] G. F. Newell. A simplified car- following theory: a lower order model. Transportation
379 Research Part B: Methodological, 36( 3): 195– 205, 2002.
380 [ 22] H. Rakha, K. Ahn, and A. Trani. Development of VT- Micro model for estimating hot stabi-381
lized light duty vehicle and truck emissions. Transportation Research Part D: Transport and
382 Environment, 9( 1): 49– 74, 2004.
383 [ 23] S. Ahn, M. J. Cassidy, and J. Laval. Verification of a simplified car- following theory.
384 Transportation Research Part B: Methodological, 38( 5): 431– 440, 2004.
385 [ 24] J. Luo and J. P. Hubaux. A survey of inter- vehicle communication. EPFL, Lausanne,
386 Switzerland, Tech. Rep, 2004.
387 [ 25] D. Shah, J. Mammen, AE Gamal, and B. Prabhakar. Throughput- delay trade- off in wireless
388 networks. In Proceedings of IEEE Infocom, volume 4, 2004.
389 [ 26] J. Zhao and G. Cao. VADD: Vehicle- assisted data delivery in vehicular ad hoc networks.
390 IEEE Transactions on Vehicular Technology, 57( 3): 1910– 1922, 2008.
391 [ 27] C. F. Daganzo. In traffic flow, cellular automata = kinematic waves. Transportation Research
392 Part B: Methodological, 40( 5): 396– 403, 2006.
22