design of an efficient vehicle-actuated signal control...

1

Design of an Efficient Vehicle-Actuated Signal Control Logic for Signalized 1

Intersections 2

3

Ziya CAKICIa,*, Yetis Sazi MURATb, Metin Mutlu AYDINc 4

a Bayburt University, Faculty of Engineering, Department of Civil Engineering, Bayburt 69000, Turkey 5

b Pamukkale University, Faculty of Engineering, Department of Civil Engineering, Denizli 20160, Turkey 6

c Ondokuz Mayıs University, Faculty of Engineering, Department of Civil Engineering, Samsun 55000, Turkey 7

8

* Corresponding author: 9

E-mail: [email protected]; Phone no: +90 554 204 5924 10

11

Abstract 12

The effectiveness of fixed-time management systems dramatically decreases in case of 13

fluctuated traffic demands at signalized intersection approaches. This leads to the waste of time 14

in traffic and may cause material, psychological and ecological problems. Especially in recent 15

years, to minimize the negative impacts of these problems, many researchers focus on 16

Intelligent Transportation Systems (ITS). As it is known, one of the Intelligent Transportation 17

Systems applications is called vehicle-actuated (traffic-actuated) management systems. 18

Because the success of these types of applications is directly related to the created control logic, 19

the selection of the most proper control parameters for the created control logic is an important 20

issue. This study aims to create an effective control logic and flow chart for the vehicle-actuated 21

management system. At the end of the analyses, it is seen that the created vehicle-actuated 22

management system can adapt to fluctuations in traffic demands at signalized intersection 23

approaches. Average vehicle delays, fuel consumptions and exhaust emissions can be reduced 24

significantly by the created system. Especially, in case of fluctuations in traffic demands at 25

2

intersection approaches exist, it is concluded that the performance of the intersection can 26

increase enormously with the created vehicle-actuated management system. 27

Keywords: Fluctuated traffic demand; Vehicle-actuated management, Delay, Fixed-time 28

management, Simulation. 29

1. Introduction 30

Economic and technological developments all over the world make life easier. However, this 31

situation may cause several negative effects in many areas. Highways are undoubtedly one of 32

the areas where these problems are mostly seen. In parallel with the increase in demand for 33

passenger and freight transportation, the number of vehicles has increased. This increase can 34

cause many economic and environmental problems [1-3]. To minimize or eliminate the 35

negative effects of these problems on humans and the environment, geometric-topographic and 36

operational arrangements at road networks can be pointed out as important parts of 37

environment-economy-friendly measures and reasonable solutions. At this stage, scientific-38

based analysis and solution approaches should be preferred. Minimizing the stop numbers and 39

delays is an important step to solve this problem at signalized intersections [4]. The 40

minimization of delays and stops at signalized intersections can be achieved in two different 41

ways: with most proper designing of intersection geometry (geometric and topographic 42

arrangements) and with determining of signal timings considering traffic demands at 43

intersection approaches (operational arrangement) [5-8]. Hence, effective signal management 44

is a quite important factor to reduce the delay, fuel consumption, noise pollution and exhaust 45

emissions and to increase the intersection capacity and level of service of intersection [9-12]. 46

In the absence of the fluctuations in traffic demands at intersection approaches, the optimum 47

fixed-time management system, which is widely used today, maybe the right and appropriate 48

choice to manage the traffic flows at intersections [13]. But sometimes sports events, concerts, 49

traffic accidents, bad weather conditions can cause short or long-term fluctuations in traffic 50

3

demands at some intersection approaches. In such cases, the operating efficiency of the 51

optimum fixed-time management system decreases especially at intersection approaches where 52

the fluctuations in traffic demands exist. This situation may cause long-term delays, long 53

queues, excessive fuel consumption and exhaust emissions [13-16]. It has been demonstrated 54

in many studies by researchers that in the event of fluctuations in traffic demands, optimum 55

fixed-time signal management systems cannot provide efficient results [17, 18]. 56

For the last 60 years, a lot of studies have been conducted on signal control systems. And, the 57

studies have continued today to improve their performances [19-21]. Since the last quarter of 58

the 20th century, it has been also aimed to make these systems more effective for high traffic 59

demands at signalized intersections and thus it has been aimed to increase the operational 60

efficiency further [22-26]. Especially in recent years, there has been a significant increase in 61

the number of studies on the signal management systems and control algorithms that take the 62

instantaneous traffic flows into account. These systems may appear in different ways either 63

optimization or created control algorithm based [27, 28]. 64

Traffic management systems that are operated based on any created control algorithm are 65

referred to as vehicle-actuated (traffic-actuated) signal management systems. Vehicle-actuated 66

management systems determine the green and red signal timings according to the current 67

dynamics of vehicle arrivals and queuing information obtained by detectors placed on the lanes 68

at intersection approaches [28]. Because vehicle-actuated signal management systems generally 69

have positive effects on the intersection performance, many researchers have worked on this 70

issue [26]. Some of these studies can be summarized as follows: 71

Akçelik [22] proposed an approach to estimate the average green times and cycle time for 72

vehicle-actuated management systems. He made an evaluation on the prediction of vehicle 73

arrival headways and tried to determine the optimum cycle time. Trabia et al. [29] worked on 74

fuzzy logic-based vehicle-actuated management systems. As a result of the study, they 75

4

determined that delays can be reduced by 10% with a fuzzy-based approach. Kim and Courage 76

[30] proposed a maximum green time design method for a vehicle-actuated management system 77

that minimizes delay. Guo et al. [31] developed the two analytical models to optimize the 78

maximum green times which are applied on vehicle-actuated management systems. They 79

specified that these models could adjust maximum green time systematically and cyclically by 80

adapting to the changes in traffic demands at intersections. Zheng et al. [32] worked on a control 81

model which aims to provide adaptive functionality of vehicle-actuated management systems. 82

As a result of the study, it was found that the model has a great potential to improve the 83

performance of the signalized network. Jiang et al. [33] developed a method to optimize the 84

parameters of the vehicle-actuated signal management systems. In another study, Grether et al. 85

[13] evaluated the vehicle-actuated signal control strategy. As a result, they concluded that 86

travel times can be significantly reduced with the vehicle-actuated management system, 87

especially in the event of excessive increases in traffic demands. Swaminathan et al. [34] 88

developed a vehicle-actuated management model. According to analysis results, it was 89

determined that the average delays can be reduced by 28% compared to the current situation 90

with the developed control model. In a similar study, Guo and Ma [35] aimed to develop a new 91

vehicle-actuated signal control model, which resulted in lower average vehicle delays compared 92

to both fixed-time and standard actuated control systems. Ribeiro and Simoes [36] used global 93

optimization and complementarity to determine the green and red signal timings at each cycle 94

for vehicle-actuated management systems. At the end of the analysis, it was concluded that the 95

suggested methodology can be used for the determination of the green and red signal timings, 96

as well as cycle lengths. Lee and Maleki [37] studied on maximum green time settings for 97

vehicle-actuated signal control at isolated intersections. They aimed to determine maximum 98

green times using fuzzy logic approach, dynamically. As a result, it was determined that fuzzy 99

logic is a useful approach to overcome the difficulty of setting suitable maximum green time 100

5

for any phase. In a different study, Wang et al. [38] investigated the green time extension for 101

vehicle-actuated signal control. As a result, they found that the optimum critical headway value, 102

which is an important parameter for green time extension, decreases with an increase in traffic 103

demand and in some cases, this value may be more than 2-3 seconds as a common assumption. 104

Promraksa et al. [39] studied on vehicle-actuated signal management system for reducing CO2 105

emission. At the end of the analyses, it was concluded that a fully-actuated management system 106

can provide better and more encouraging results than the fixed-time and semi-actuated 107

management systems. 108

As can be seen from the previous studies, the selection of the most proper design parameters 109

(such as minimum critical headway, placement of detectors, minimum green signal timings, 110

maximum green signal timings) for the vehicle-actuated management systems has a direct and 111

significant effect on the system performance. This study guides for designing an efficient 112

vehicle-actuated signal control logic for signalized intersections. In the scope of the study, the 113

selection of the most proper design parameters was evaluated, separately. Then, a new vehicle-114

actuated management system for a four-leg signalized intersection was developed in VISSIM 115

Vehicle Actuated Programming (VISVAP). The efficiency of the developed system was tested 116

on 20 different traffic scenarios considering the different performance criteria (vehicle delay, 117

fuel consumption and exhaust emission) in VISSIM. To determine the success of the developed 118

system, unlike the current studies in the literature, short and long-term fluctuations in traffic 119

demands at the signalized intersection approaches were also taken into account in the analysis. 120

It is seen that the developed system can adapt to the fluctuations in traffic demands and 121

enormously reduce the average vehicle delay, fuel consumption and exhaust emission at the 122

signalized intersection. Thus, it is determined that the developed system can be used for 123

improving the performance of signalized intersections. 124

2. Method 125

6

2.1.Vehicle-Actuated Management Model for Four-Leg Signalized Intersections: VAM4 126

In vehicle-actuated management, traffic flows at an intersection are managed using data and 127

information from the detectors which are located at the intersection approaches [18]. Vehicle-128

actuated management systems can be classified in two ways as semi-actuated and fully-129

actuated. In semi-actuated management, only one or a few of the intersection approaches have 130

detectors. In fully-actuated management, all intersection approaches have detectors and traffic 131

flow data and information are provided from all approaches. In this type of management, the 132

green signal timings and the cycle times are determined by considering the traffic demands 133

detected. In fully-actuated management systems, the order of phases can be fixed or flexible. 134

Furthermore, if there is no demand for a phase, that phase can be skipped directly. In vehicle-135

actuated management systems, the effective operation of an intersection generally depends on 136

the following factors: minimum green times (sec.), critical arrival headways of traffic flow, 137

placement of detectors (detector locations), maximum green times (sec.). Therefore, choosing 138

reasonable and appropriate values for these factors is crucial to improve signalized 139

intersections’ performance [14, 35]. In this study, the performance of the vehicle-actuated 140

signal management system (VAM4) developed for a four-leg signalized intersection which 141

includes scenarios, where short or long-term fluctuations exist in traffic demands at intersection 142

approaches, was examined. For this purpose, firstly, the control parameters and criteria related 143

to the control algorithm of the proposed VAM4 were determined. In the analysis, considering 144

the high traffic volumes, the minimum critical arrival headway is selected as approximately 2 145

seconds. In this case, as long as the critical arrival headway determined by the detectors does 146

not exceed 2 seconds, the green signal timing of that phase is extended by 2 seconds. When the 147

current studies in the literature are examined, it can be seen that this value is generally selected 148

between 1.5 and 4 seconds. [14, 40]. Therefore, the selected green extension value (2 seconds) 149

was appropriate for the analysis. Besides, when the previous studies are examined carefully, 150

7

there is no clear and precise information about the placement of detectors for vehicle-actuated 151

management systems. However, the placement of the detectors was directly related to the traffic 152

volume as a common practice in the literature [14, 15, 40]. According to previous studies, the 153

detectors were placed generally 15 m to 45 m before the stop lines. However, there is no any 154

precise information about the placement of the detectors in the literature. In this study, the 155

locations of the detectors were determined by trial and error method for each traffic scenario. 156

According to the results obtained by trial and error, the relationship between the distance of the 157

detectors to the stop line at the intersection approaches and the average delay was determined 158

as given in Fig. 1. This relationship is similar to the relationship depicted by Bullen [14]. 159

When the previous studies are examined, the minimum green time is generally selected as 4-5 160

seconds or above depending on the distance of the detector to the stop line [14, 15]. As 161

mentioned before, especially when the traffic volumes at intersection approaches are high, the 162

distance of the detectors to the stop line can be up to 45 meters. For this reason, the minimum 163

green time value must be determined considering locations (placements) of the detectors at the 164

intersection approaches. Determined minimum green time must allow to clearing of the queues 165

at intersection approaches. Depending on the locations of the detectors, the minimum green 166

times for the vehicle-actuated management systems can be calculated by using Eq. (1) [15]: 167

min L

lG t h integer

e

= +

(1) 168

Where; 169

minG : Minimum green time (sec), 170

Lt : Assumed start-up lost time (sec), 171

h : Assumed saturation headway (sec), 172

l : Distance between detectors and stop line (m), 173

e : Distance between storaged vehicles (average 6-7 m). 174

8

In the scope of this study, minimum green times are computed by using Eq. (1). Depending on 175

the distance between detectors and stop line, the obtained minimum green times are presented 176

in Table 1. 177

As can be seen from the previous studies, the maximum green time for vehicle-actuated 178

management systems is generally selected between 45 and 60 seconds [41]. In this study, the 179

maximum green time of each phase is selected as 60 seconds and this value is used in the 180

analysis. General information about the proposed VAM4 for the four-leg signalized intersection 181

model is shown in Fig. 2. 182

As can be seen from Fig. 2, the modeled four-leg signalized intersection has a total of 14 183

detectors for 14 lanes (one detector for each lane). These detectors are used for both detecting 184

the queuing at intersection approaches and determining the vehicle arrival headways. In the 185

developed VAM4 model, it is assumed that the modeled four-leg signalized intersection is 186

operated with four phases. Right of way for each intersection approach is provided in separate 187

phases. Right of ways for the West, North, East and South intersection approaches are provided 188

in Phase I, Phase II, Phase III and Phase IV, respectively. Besides, in the developed VAM4 189

model, when excitation is not detected from the detectors which are at one of the intersection 190

approaches, phase transition can be actualized (the order of phases can be changed). The 191

simulation period for the analysis is determined as 3600 sec. The flow chart of the developed 192

VAM4 control logic is presented in Fig. 3. 193

2.2.Traffic Scenarios and Fluctuations in Traffic Demands 194

In the scope of this study, firstly, a movement direction-based reference demand matrix has 195

been created considering the four-leg signalized intersection model shown in Fig. 2. This 196

created reference demand matrix for the analysis is given in Table 2. 197

In the next step, a total of 20 different traffic scenarios are created by using the reference 198

demand matrix given in Table 2 for the modeled four-leg signalized intersection. For each 199

9

created scenario, movement direction-based traffic volumes showed an increase or decrease 200

based on the reference demand matrix. Besides, some of the movement direction-based traffic 201

volumes are constant in some scenarios. The increase and decrease rates of the movement 202

direction-based traffic volumes for all scenarios are shown in Fig. 4. 203

As can be seen from Fig. 4, when the scenario number increases, general increase trends are 204

seen in movement direction-based traffic volumes. Therefore, the total traffic volumes at the 205

intersection increase when the scenario number increases as shown in Fig. 5. As can be seen in 206

Fig. 5, created traffic scenarios are divided into four groups considering total traffic volumes at 207

the intersection. According to their demand levels, these groups are named as low, moderate, 208

high and very high. When the hourly total traffic demand is between 2000 veh. and 2500 veh., 209

the demand is considered as low. If the hourly total traffic demand is between 2500 veh. and 210

3000 veh., it can be said that the demand is moderate. When the hourly total traffic demand is 211

between 3000 veh. and 3500 veh., the demand is regarded as high. Finally, if the hourly total 212

traffic demand is between 3500 veh. and 4000 veh., it can be said that the demand is very high. 213

Thus, the performance of the VAM4 model developed for a four-leg signalized intersection 214

could be evaluated for the cases of low, moderate, high and very high traffic demands, 215

separately. 216

As it is known, in the VISSIM simulation program, the time intervals when the fluctuations in 217

traffic demands occur and the quantities (proportions) of the fluctuations in these time intervals 218

can be determined by the users. Thus, as a result of the short or long-term fluctuations that may 219

be seen due to the reasons such as concerts, sports events, traffic accidents, adverse weather 220

conditions, changes that occur in the performance of intersection can be analyzed in VISSIM. 221

As mentioned above, in this study, it is aimed to examine the performance and effectiveness of 222

the developed VAM4 in case of fluctuations with variable quantities in variable time intervals 223

for traffic demands at different intersection approaches. For this aim, sample cases related to 224

10

the quantities of the fluctuations in traffic demands and the time intervals when the fluctuations 225

occur in traffic demands are created for each traffic scenario. The created sample cases are 226

presented in Fig. 6. In Fig. 6, for each traffic scenario, the time intervals when the fluctuations 227

occur in traffic demands at different intersection approaches and the quantities of fluctuations 228

during 60 minutes of the analysis period can be seen in detail. As an example; for Scenario 6, 229

it is assumed that the fluctuations in traffic demands occur twice at the West and North 230

approaches and only once at the East approach during 60 minutes of the analysis period. 231

Besides, for Scenario 6, the following matters can be said about the quantities of fluctuations, 232

and the time intervals when the fluctuations occur in traffic demands: 233

• It is assumed that approximately 33% of the vehicles at the west approach arrive to the 234

intersection between the 20th and 33rd minutes, and approximately 28% arrive to the 235

intersection between the 47th and 60th minutes. 236

• It is assumed that approximately 22% of the vehicles at the north approach arrive to the 237

intersection between the 20th and 27th minutes and 22% arrive to the intersection between 238

the 40th and 47th minutes. 239

• It is assumed that approximately 50% of the vehicles at the east approach arrive to the 240

intersection between the 13th and 33rd minutes. 241

3. Analysis 242

Webster, Highway Capacity Manual (HCM) and Akcelik methods are the most used delay 243

calculation approaches for predicting the average vehicle delay which occurs at signalized 244

intersections [20, 40, 42]. While the Webster and HCM methods are based on the phase-related 245

design, the Akcelik method is based on the movement-related design. In the scope of the study, 246

when the modeled intersection was examined carefully, it was seen that all lanes at intersection 247

approaches were designed based on the traffic flows’ movement directions. Thus, the Akcelik 248

delay model which aims movement-related design was used for optimizing the signal timings 249

11

in this study. According to the Akcelik method, the approximate value of total delay for a 250

movement can be calculated by using Eq. (2) [42]: 251

( )

( )

21

2 1o

qC uD N x

y

−= +

− (2) 252

Where; 253

D : Total Delay (sec), 254

q : Flow (veh/sec), 255

C : Cycle time (sec), 256

qC : Average number of arrivals in vehicles per cycle, 257

u : Green time ratio ( g C ), 258

y : Flow ratio ( q s ), 259

oN : Average overflow queue (veh) 260

x : Degree of saturation ( q Q ) (veh). 261

At a signalized intersection approach, to predict the overflow queues for both undersaturated 262

( 1x ) and oversaturated ( 1x ) conditions, the following Eq. (3) which was developed by 263

Akcelik can be used: 264

( )212

4

0

f o

o

O f

o

QT x xz z , x x

N QT

, x x

− + +

=

(3) 265

Where; 266

oN : Average overflow queue (veh), 267

Q : Capacity in vehicle in per hour (veh/hr), 268

fT : Flow period, 269

fQT : Maximum number of vehicles which can be discharged during the interval fT , 270

12

x : Degree of saturation ( q Q ), 271

z : 1x− , 272

ox : The degree of saturation below which the average overflow queue is approximately 273

zero. 274

The degree of saturation below which the average overflow queue is approximately zero ( ox ) 275

can be calculated by Eq. (4): 276

0 67600

o

sgx .= + (4) 277

Where; 278

s : Saturation flow (veh/sec), 279

g : Effective green time (sec). 280

Average vehicle delay can be calculated by Eq. (5): 281

Dd

q= (5) 282

Where; 283

D : Total Delay (sec), 284

q : Flow (veh). 285

With the analysis studies, the determination of the effectiveness of VAM4 which was created 286

in VISVAP was aimed. Therefore, firstly, it was assumed that the modeled signalized 287

intersection was managed with an optimum fixed-time signal control system and it was operated 288

with four phases. The signal timings were optimized by minimizing of average vehicle delay at 289

the intersection. The objective function ( )f x , the decision variables and the set of constraints 290

for the signal timing optimization problem are shown in Table 3. 291

As can be seen from Table 3, for the modeled intersection, the optimization problem consists 292

of 1 objective function, 4 decision variables and 18 constraints. When the previous studies were 293

13

investigated carefully, it was seen that green signal timings for each phase should be within a 294

certain range [43]. Thus, green signal timings were constrained between 7 seconds and 45 295

seconds considering the previous studies in the literature [43, 44]. In addition to this, the degree 296

of saturation for each lane was constrained with a maximum of 1.2. Differential Evolution (DE) 297

algorithm which is a powerful and simple meta-heuristic optimization algorithm is preferred to 298

solve the signal timing optimization problem [45-49]. A signal timing optimization program 299

was developed in MATLAB for this purpose. For created traffic scenarios, obtained signal 300

timings and average vehicle delays are given in Table 4 in detail. 301

Because all of the scenarios which are created in the scope of the study are analyzed in VISSIM, 302

Akcelik average vehicle delay results and average vehicle delay results obtained from VISSIM 303

should be similar to each other for the same scenarios. Rational and suitable evaluation cannot 304

be possible, otherwise. Due to differences between the results, firstly, an evaluation area was 305

created considering the points which are 100 meters away from all approaches stop lines. Then, 306

driving behaviors and the safety factors in VISSIM software were revised considering the 307

results which are obtained from Akcelik delay equation. Thus, average vehicle delay results for 308

each scenario are likened to each other. For created scenarios, the comparison of the average 309

vehicle delay values is shown in Table 5. 310

As can be seen from Table 5, for created scenarios, the differences are below about 5% in 311

generally. The accuracy of the simulation is determined by using the Sum of Squared Error 312

value [50]. This value is calculated as 33.56. These results show that the VISSIM can provide 313

similar average delay values with the Akcelik delay equation and can be used for analysis 314

studies. The average vehicle delay values which are obtained from VISSIM and by using the 315

Akcelik delay equation are shown in Fig. 7(a). The percentage differences between Akcelik and 316

VISSIM are also shown in Fig. 7(b). 317

14

After the completion of the revision process in the VISSIM simulation program, to determine 318

the effectiveness of VAM4, created vehicle actuated signal management control algorithm is 319

tested on all traffic scenarios. At this stage, firstly, the most proper placements of detectors for 320

each scenario were determined with the trial and error method. For Scenario 5 and Scenario 17, 321

average vehicle delay values which were obtained depending on the detector placement are as 322

shown in Fig. 8. 323

As seen in Fig. 8, a comprehensive study was carried out to determine the most proper 324

placement of the detectors. Because the distance between detectors and stop lines is varied from 325

15 meters to 45 meters, detectors must be placed within this range at all intersection approaches. 326

At this stage, created scenarios were analyzed considering different distances between detectors 327

and stop lines. At the end of the analyses, the distance which corresponds to the minimum 328

average vehicle delay value was considered as the most proper distance between detectors and 329

stop lines. Near optimum detector placements (distances) which are obtained from the analysis 330

for all created scenarios are presented in Table 6. 331

As can be seen from Table 6, when the total traffic volume at the intersection increases, the 332

distance between detectors and stop line also shows an increase. While the total traffic demand 333

at the intersection is low or moderate, near optimum distances between detectors and stop line 334

vary in the range of 15 meters and 25 meters in generally. Besides, while the total traffic demand 335

at the intersection is high or very high, near optimum distances between detectors and stop line 336

vary in the range of between 30 meters and 45 meters. 337

4. Results 338

After the determination of the most proper placements of detectors for all created scenarios, 339

two different types of traffic management approaches were evaluated separately considering 340

whether the fluctuations in traffic demands exist or not for each scenario. These four different 341

cases which are taken into account for the analysis can be summarized as follows: 342

15

• Optimum Fixed-Time Management / Fluctuations do not exist in traffic demands (OFTM 343

/ Not-Fluctuated): In this case, fluctuations do not exist in traffic demands. Movement-344

based traffic volumes and optimum signal timings are transferred to VISSIM. At the end 345

of the analysis, average vehicle delay values for each scenario were obtained. Obtained 346

results can be seen in Table 5. 347

• Vehicle-Actuated Management / Fluctuations do not exist in traffic demands (VAM4 / 348

Not-Fluctuated): In the second case, fluctuations do not exist in traffic demands. All 349

scenarios were analyzed considering the VAM4 model which is created in VISVAP. 350

• Optimum Fixed-Time Management / Fluctuations exist in traffic demands (OFTM / 351

Fluctuated): In the third case, fluctuations exist in traffic demands for each scenario (in 352

Fig. 6). The only difference of this case from the first case is existing fluctuations in traffic 353

demands. Movement-based traffic volumes and total traffic volumes are the same for the 354

first case and third case. In this case, the investigation of the success of optimum fixed-355

time management (OFTM) was aimed for fluctuated traffic demands. 356

• Vehicle-Actuated Management / Fluctuations exist in traffic demands (VAM4 / 357

Fluctuated): In the fourth case, fluctuations exist in traffic demands for each scenario (in 358

Fig. 6). The only difference of this case from the second case is existing fluctuations in 359

demands. Movement-based traffic volumes and total traffic volumes are the same for the 360

second case and fourth case. In this case, the investigation of the success of VAM4 was 361

aimed for fluctuated traffic demands. 362

At the end of the analysis which is made considering four different cases described above, 363

average vehicle delay-fuel consumption and exhaust emission (CO+NOx) values were obtained 364

for each scenario. Obtained average vehicle delay, fuel consumption and exhaust emission 365

values are shown in Fig. 9(a), Fig. 9(b) and Fig. 9(c), respectively. 366

16

As seen in Fig. 9, especially in case of fluctuated traffic demands, the effectiveness of the 367

optimum fixed-time signal management system decreases dramatically. Besides, the average 368

vehicle delays, fuel consumptions and exhaust emissions can be significantly reduced by 369

implementing vehicle-actuated signal control at the intersection. For Scenario 3, in the time 370

interval when the fluctuations occur in traffic demands, sample intersection views which are 371

obtained from VISSIM in case of implementing the optimum fixed-time management or/and 372

vehicle actuated management are presented in Fig. 10. 373

In Fig. 10, for Scenario 3, in the time interval when the fluctuations occur in the traffic demands, 374

sample intersection views in case of implementing the different types of management 375

approaches are shown. As shown in Fig. 6, the fluctuations in the traffic demands for Scenario 376

3 occur in between the 12th and 36th minutes. Therefore, simulation views are taken in the 25th 377

minutes (1475 sec.) of the simulation. As can be seen from Fig. 10, in case of implementing of 378

optimum-fixed time traffic management at modeled signalized intersection, excessive queuing 379

(queues) are seen at the intersection approaches where fluctuations in traffic demands occur 380

(Fig. 10 (a)). When Fig. 10(b) is examined carefully, in case of implementing vehicle-actuated 381

management instead of optimum fixed-time management at the modeled intersection, queuing 382

(queues) at intersection approaches where fluctuations in traffic demands occur can be 383

significantly reduced. Thus, the performance of the intersection can also be improved in this 384

way. 385

When the fluctuations do not exist in traffic demands, although vehicle-actuated signal 386

management systems have more advantages than optimum fixed-time signal management 387

systems, obtained success rates become low. For the sample scenario (Scenario 3) which is 388

shown in Fig. 10, sample intersection views in the same time intervals in case of implementing 389

the optimum fixed-time management and vehicle-actuated management are presented in Fig. 390

11. 391

17

In Fig. 11, for Scenario 3, when the fluctuations do not exist in the traffic demands, sample 392

intersection views in the same time interval in case of implementing the different types of 393

management approaches are shown. As can be seen from Fig. 11, when the fluctuations do not 394

exist in traffic demands, excessive queuing is not seen at the intersection approaches in case of 395

implementing of both optimum fixed-time management and vehicle-actuated management. But, 396

more queuing occurs in case of implementing of optimum fixed-time management (in Fig. 397

11(a)) than in case of implementing of vehicle-actuated management (in Fig. 11(b)). When Fig. 398

11 is investigated carefully, the queuing at intersection approaches can be reduced in case of 399

implementing of vehicle-actuated management. As a result, it was concluded that simulation 400

views are compatible with obtained performance results. The simulation views support and 401

verify the obtained results. 402

For created traffic scenarios, reduction rates (success rates) were evaluated separately 403

considering whether fluctuations in traffic demands exist or not. In case of implementing of 404

vehicle-actuated management instead of optimum fixed-time management, obtained reduction 405

rates (success rates) for each performance criteria are presented in Fig. 12. 406

When Fig. 12 is examined carefully, the obtained results can be summarized as follows: 407

• For the fluctuated traffic demands, when the total traffic demand at the intersection is 408

low, moderate or high, average vehicle delay can be reduced between about 20% and 409

60%, fuel consumption and exhaust emission can be reduced between about 15% and 410

40%. 411

• For the fluctuated traffic demands, when the total traffic demand at the intersection is 412

very high, the reduction rates for average vehicle delay decrease down to 15%, reduction 413

rates for fuel consumption and exhaust emission decrease down to 10%. These results 414

show that the vehicle-actuated management systems are not efficient systems for 415

specified traffic demand conditions. 416

18

• For not-fluctuated traffic demands, when the total traffic demand at the intersection is 417

low, moderate or high, average vehicle delay can be reduced between about 5% and 20%, 418

fuel consumption and exhaust emission can be reduced between about 3% and 9%. 419

• For the not-fluctuated traffic demands, when the total traffic demand at the intersection 420

is very high, it can be seen that the reduction rates for average vehicle delay increase up 421

to 30%, reduction rates for fuel consumption and exhaust emission increase up to 18%. 422

These results show that the vehicle-actuated management systems are very effective 423

systems for specified traffic demand conditions. 424

When the cases of whether the fluctuations in traffic demands exist or not are investigated 425

separately, the obtained results are presented in Table 7 in detail. 426

As seen in Table 7, for the created scenarios; 427

• For not-fluctuated traffic demands, in case of implementing of VAM4 instead of OFTM, 428

the obtained reduction rates for average vehicle delay are about 4.5% of minimum, 29.2% 429

of maximum and 14.2% of average. For the fuel consumption and exhaust emission, these 430

rates are about 2.8% of minimum, 17.7% of maximum and 7.3% of average. 431

• For fluctuated traffic demands, in case of implementing of VAM4 instead of OFTM, the 432

obtained reduction rates for average vehicle delay are about 15.6% of minimum, 58.6% of 433

maximum and 32.5% of average. For the fuel consumption and exhaust emission, these 434

rates are about 9.6% of minimum, 38.4% of maximum and 21.8% of average. 435

As can be seen from the obtained results, developed VAM4 is more effective in case of 436

fluctuated traffic demands. For fluctuated traffic demands, reduction rates for all performance 437

criteria increase about three times compared to not-fluctuated traffic demands in case of 438

implementing of VAM4 instead of OFTM. 439

5. Discussion and Conclusion 440

19

In this study, a VAM4 that can adapt to fluctuations in traffic demands was developed. The 441

effectiveness of the developed system was tested considering whether the fluctuations in traffic 442

demands exist or not. Obtained results from this study can be summarized as follows: 443

• When the fluctuations in traffic demands did not exist, lower average vehicle delay, fuel 444

consumption and exhaust emission values were obtained with VAM4 compared to 445

OFTM. When Fig. 12 is examined carefully, especially in case of traffic demand is very 446

high at the intersection, reduction rates for all performance criteria increase. One of the 447

main reasons for this situation is that the constraint of maximum green time of 45 seconds 448

for OFTM is lower than the constraint of maximum green time of 60 seconds for VAM4. 449

The increase of maximum green time length for VAM4 enabled to lower waste of time 450

for vehicles at intersection approaches where the traffic demand is very high. Thus, 451

average vehicle delays, fuel consumption and exhaust emissions were significantly 452

reduced. 453

• When the fluctuations in traffic demands existed, the effectiveness of OFTM decreased 454

and average vehicle delay, fuel consumptions and exhaust emissions significantly 455

increased. Average vehicle delay, fuel consumption and exhaust emissions were 456

significantly reduced with the implementation of VAM4, especially in case of traffic 457

demand at the intersection was low, moderate and high. In case of traffic demand at the 458

intersection was very high, reduction rates decreased. This situation has arisen from the 459

increase of the usage of maximum green time by different intersection approaches. It was 460

determined that the density of the usage of maximum green time was an important factor 461

that encourages the increase of delay, fuel consumption and exhaust emission. 462

• For created scenarios, when the fluctuations in traffic demands did not exist, in case of 463

implementation of VAM4 instead of OFTM, the average vehicle delay was reduced about 464

the average of 14.2%. Besides, fuel consumption and exhaust emissions were reduced by 465

20

about an average of 7.3%. When the fluctuations in traffic demand existed, these rates 466

increased 2.5-3 times. 467

• When the fluctuations in traffic demands did not exist, in case of implementation of 468

VAM4 instead of OFTM, the level of service of only five scenarios was improved. When 469

the fluctuations in traffic demands existed, the level of service of fourteen scenarios was 470

improved. 471

All obtained results show that VAM4 can effectively adapt to fluctuations in traffic demands. 472

In addition to this, it is thought that the effectiveness and the performance of VAM4 can be 473

more improved with the different computational approaches. 474

Acknowledgments 475

The authors of this article would like to thank to PTV (Planung Transport Verkehr AG) for 476

providing of VISSIM simulation software. 477

Funding 478

The authors of this article received no financial support for the research, authorship, and 479

publication of this article. 480

Availability of Data and Materials 481

The data used to support the findings of this study are available from the corresponding author 482

upon request. 483

6. References 484

[1] Li, J-Q., Wu, G. and Zou, N. “Investigation of The Impacts of Signal Timing on Vehicle 485

Emissions at an Isolated Intersection”, Transp. Res. D Transp. Environ., 16 (2011), pp. 409-486

414 (2011). 487

[2] Asadi, H., Tavakkoli Moghaddam, R. and Shahsavari Pour, N. et al., “A New 488

Nondominated Sorting Genetic Algorithm based to The Regression Line for Fuzzy Traffic 489

Signal Optimization Problem”, Sci. Iran., 25 (3), pp. 1712-1723 (2018). 490

21

[3] Yang, L., Zhang, L. and Stettler, M. E. J. et al., “Supporting An Integrated Transportation 491

Infrastructure and Public Space Design: A Coupled Simulation Method for Evaluating 492

Traffic Pollution and Microclimate”, Sustain. Cities Soc., 52 (2020), pp. 101796 (2020). 493

[4] Barzegar, S., Davoudpour, M. and Meybodi, M. R. et al., “Formalized Learning Automata 494

with Adaptive Fuzzy Coloured Petri net; An Application Specific to Managing Traffic 495

Signals”, Sci. Iran., 18 (3), pp. 554-565 (2011). 496

[5] Samadi, S., Rad, A. P. and Kazemi, F. M. et al., “Performance Evaluation of Intelligent 497

Adaptive Traffic Control Systems: A Case Study”, J. Transp. Technol., 2, pp. 248-259 498

(2012). 499

[6] McKenny, D. and White, T. “Distributed and Adaptive Traffic Signal Control Within a 500

Realistic Traffic Simulation”, Eng. Appl. Artif. Intell., 26, pp. 574-583 (2012). 501

[7] Vilarinho, C. and Tavares, J. P. “Real-Time Traffic Signal Settings at an Isolated Signal 502

Control Intersection”, Procedia Soc. Behav. Sci., 3, pp. 1021-1030 (2014). 503

[8] Dabiri, S. and Abbas, M. “Arterial Traffic Signal Optimization using Particle Swarm 504

Optimization in an Integrated VISSIM-MATLAB Simulation Environment”, 2016 IEEE 505

19th International Conference on Intelligent Transportation Systems, Rio de Janerio, Brazil, 506

pp. 766-771 (2016). 507

[9] Abdelghaffar, H. M., Yang, H. and Rakha, H. A. “Isolated Traffic Signal Control using 508

Nash Bargaining Optimization”, Glob. J. Res. Eng. B, 16 (1), pp. 1-11 (2016). 509

[10] Li, Z., Shahidehpour, M. and Bahramirad, S. et al., “Optimizing Traffic Signal Settings 510

in Smart Cities”, IEEE T. Smart Grid, 8 (5), pp. 2382-2393 (2017). 511

[11] Hashim, I., Ragab, M. and Asar, G. “Optimization of Vehicle Delay and Exhaust 512

Emissions at Signalized Intersections”, Adv. Transp. Stud. A, 46 (2018), 5-18 (2018). 513

22

[12] Yao, Z., Jiang, Y. and Zhao, B. et al., “A Dynamic Optimization Method for Adaptive 514

Signal Control in a Connected Vehicle Environment”, J. Intell. Transp. Syst., 24 (2), pp. 515

184-200 (2019). 516

[13] Grether, D., Bischoff, J. and Nagel, K. “Traffic Actuated Signal Control: Simulation of 517

The User Benefits in a Big Event Real-World Scenario”, 2nd International Conference on 518

Models and Technologies for Intelligent Transportation Systems, pp. 1-4 (2011). 519

[14] Bullen, A. G. R. “Effects of Actuated Signal Settings and Detector Placement on 520

Vehicle Delay”, Transp. Res. Rec., 1244, pp. 32-38 (1989). 521

[15] Mathew, T. V. Signalized Intersections Delay Models, Lecture Notes in Transportation 522

System Engineering (2017). 523

[16] Villiers, C., Nguyen, L. D. and Zalewski, J. “Evaluation of Traffic Management 524

Strategies for Special Events using Probe Data”, TRIP, 2 (2019), pp. 100052 (2019). 525

[17] Van Katwijk, R. T. Multi-Agent Look-Ahead Traffic-Adaptive Control, PhD Thesis, 526

Technische Universiteit Delft, Delft, (2008). 527

[18] Gundogan, F. Simplified Traffic Responsive Signal Control Method for Developing 528

Large Cities, PhD Thesis, Graz University of Technology, Graz, (2012). 529

[19] Webster, F. V. Traffic Signal Settings Report (1958). 530

[20] Mohajerpoor, R., Saberi, M. and Ramezani, M. “Analytical Derivation of the Optimal 531

Traffic Signal Timing: Minimizing Delay Variability and Spillback Probability for 532

Undersaturated Intersections”, Transp. Res. B Meth., 119 (2019), pp. 45-68 (2019). 533

[21] Lian, F., Chen, B. and Zhang, K. et al., “Adaptive Traffic Signal Control Algorithms 534

based on Probe Vehicle Data”, J. Intell. Transp. Syst., 25 (2021), pp. 41-57 (2021). 535

[22] Akcelik, R. “Estimation of Green Times and Cycle Time for Vehicle Actuated Signals”, 536

Transp. Res. Rec., 1457, 63-72 (1994). 537

23

[23] Mirchandani, P. and Head, L. “A Real-Time Traffic Signal Control System: 538

Architecture, Algorithms and Analysis”, Transp. Res. C Emerg. Technol., 9 (2001), pp. 539

415-432 (2001). 540

[24] Angulo, E., Romero, F. P. and Garcia, R. et al., “An Adaptive Approach to Enhanced 541

Traffic Signal Optimization by using Soft-Computing Techniques”, Expert Syst. Appl., 38, 542

2235-2247 (2011). 543

[25] Yu, D., Tian, X. and Xing, X. et al., “Signal Timing Optimization Based on Fuzzy 544

Compromise Programming for Isolated Signalized Intersection”, Math. Probl. Eng., 2016, 545

Article ID: 1682394, pp. 1-12 (2016). 546

[26] Hao, R., Wang, L. and Ma, V. et al., “Estimating Signal Timing of Actuated Signal 547

Control using Pattern Recognition under Connected Vehicle Environment”, Promet-548

Zagreb, 33 (1), pp. 153-163 (2021). 549

[27] Li, H. and Prevedouros, P. D. “Traffic Adaptive Control for Oversaturated Isolated 550

Intersections: Model Development and Simulation Testing”, J. Transp. Eng., 130 (5), pp. 551

594-601 (2004). 552

[28] Viti, F. and van Zuylen, H. J. “A Probabilistic Model for Traffic at Actuated Control 553

Signals”, Transp. Res. C Emerg. Technol., 18 (2010), pp. 299-310 (2010). 554

[29] Trabia, M. B., Kaseko, M. S. and Ande, M. “A Two-Stage Fuzzy Logic Controller for 555

Traffic Signals”, Transp. Res. C Emerg. Technol., 7 (1999), pp. 353 – 367 (1999). 556

[30] Kim, J. T. and Courage, K. G. “Evaluation and Design of Maximum Green Time 557

Settings for Traffic-Actuated Control”, Transp. Res. Rec., 3606, pp. 246-255 (2003). 558

[31] Guo, W., Yu, Z. and He, Z. et al., “Traffic-Actuated Signal Control based on Dynamic 559

Optimal Maximum Green Time”, International Conference on Transportation Engineering 560

2007 (ICTE 2007), pp. 582-587 (2007). 561

24

[32] Zheng, X., Recker, W. and Chu, L. “Optimization of Control Parameters for Adaptive 562

Traffic-Actuated Signal Control”, J. Intell. Transp. Syst., 14 (2), pp. 95-108 (2010). 563

[33] Jiang, X., Qiu, Y. and Ruan, S. “An Approach to Optimize the Settings of Actuated 564

Signals”, J. Mod. Transp., 19 (1), pp. 68-74 (2011). 565

[34] Swaminathan, N., Rathinawel, N. and Duraisamy, S. et al., “Design of Vehicle Actuated 566

Signal Using Simulation”, Gradevinar, 7 (2014), pp. 635-641 (2014). 567

[35] Guo, Y. and Ma, J. “An Improved Actuated Signal Control of Intersection based on 568

VISVAP”, 6th International Conference on Sensor Network and Computer Engineering, pp. 569

123-128 (2016). 570

[36] Ribeiro, I. M. and Simoes, M. L. O. “The Fully Actuated Traffic Control Problem 571

Solved by Global Optimization and Complementarity”, Eng. Optim., 48 (2), pp. 199-212 572

(2016). 573

[37] Lee, S. H. and Han, D. “Safety Impacts of The Actuated Signal Control at Urban 574

Intersections”, Promet-Zagreb, 28 (1), pp. 31-39 (2016). 575

[38] Wang, X.B, Yin, K. and Liu, H. “Vehicle Actuated Signal Performance Under General 576

Traffic at an Isolated Intersection”, Transp. Res. C Emerg. Technol., 95 (2018), pp. 582-577

598 (2018). 578

[39] Promraksa, T., Satiennam, T. and Satiennam, W. “Vehicle Actuated Signal Control for 579

Low Carbon Society”, Int. J. GEOMATE, 16 (55), pp. 86-91 (2019). 580

[40] Highway Capacity Manual (2010). National Research Council, Washington, D. C., 581

(2010). 582

[41] Dogan, E. Isolated Signalized Intersection Control by Optimized Fuzzy Logic Method, 583

PhD Thesis, Kirikkale University, Institute of Natural and Applied Sciences, Kırıkkale, 584

(2014). 585

25

[42] Akcelik, R. Traffic Signals: Capacity and Timing Analysis, Australian Road Research 586

Board, Research Report ARR No: 123 (1998). 587

[43] Murat, Y. S. and Kikuchi, S. “The Fuzzy Optimization Approach: A Comparison with 588

the Classical Optimization Approach Using the Problem of Timing a Traffic Signal”, 589

Transp. Res. Rec., 2024, pp. 82-91 (2007). 590

[44] Ceylan, H. and Bell, M. G. H. “Traffic Signal Timing Optimisation based on Genetic 591

Algorithm Approach including Drivers’ Routing”, Transp. Res. B Meth., 38 (4), pp. 329-592

342 (2004). 593

[45] Keskinturk, T. “Differential Evolution Algorithm”, J Ins. Nat. App. Sci. of Istanbul 594

Ticaret University, 9 (2006), pp. 85-99 (2007). 595

[46] Baskan, O. and Ceylan, H. “Differential Evolution Algorithm based Solution 596

Approaches for Solving Transportation Network Design Problems”, Pamukkale University 597

J. Eng. Sci., 20 (9), pp. 324-331 (2014). 598

[47] Dogan, E. and Akgungor, A. P. “Optimizing a Fuzzy Logic Traffic Signal Controller 599

via the Differential Evolution Algorithm Under Different Traffic Scenarios”, Simulation, 600

92 (11), pp. 1013-1023 (2016). 601

[48] Korkmaz, E. and Akgungor, A. P. “Delay Estimation Models for Signalized 602

Intersections Using Differential Evolution Algorithm”, Int. J. Eng. Res., 5 (3), pp. 16-29 603

(2017). 604

[49] Baskan, O. and Ozan, C. “Determining Optimum Configuration of One-Way and Two-605

Way Streets Using Shortest Path Travel Costs Based on Results of Traffic Assignment”, 606

Pamukkale University J. Eng. Sci., 24 (6), pp. 1087-1092 (2018). 607

[50] Yu, L., Yu, L. and Chen, X. et al., “Calibration of VISSIM for Bus Rapid Transit 608

Systems in Beijing using GPS Data”, J. Public Trans., 2006 BRT Special Edition, pp. 239-609

257 (2006). 610

26

Figure and Table Captions 611

List of Figures 612

Fig. 1 The relation between average vehicle delay and the placement of detectors 613

Fig. 2 Vehicle-actuated management application at a four-leg signalized intersection in VISSIM 614

Fig. 3 The flow chart of vehicle-actuated control logic created with VAP 615

Fig. 4 The change rates of movement direction-based traffic volumes for created scenarios 616

Fig. 5 Hourly total traffic volumes for all traffic scenarios 617

Fig. 6 Time intervals and quantities of fluctuations in traffic demands for each traffic scenario 618

Fig. 7 For created scenarios (a) the comparison of average vehicle delay values and (b) the 619

percentage (%) differences of average vehicle delay values. 620

Fig. 8 The effect of detector placements on average vehicle delay (a) for Scenario 5 and (b) 621

Scenario 17. 622

Fig. 9 At the end of the analysis, for created scenarios (a) obtained average vehicle delay results; 623

(b) obtained fuel consumption results; (c) obtained exhaust emission results. 624

Fig. 10 For the Scenario 3, in time interval when the fluctuations occur in traffic demands, 625

sample intersection view in case of implementing (a) the optimum-fixed time management 626

(OFTM) (b) vehicle-actuated management (VAM4) 627

Fig. 11 For Scenario 3, when the fluctuations do not exist in traffic demands, sample 628

intersection view in time same time interval in case of implementing (a) the optimum fixed-629

time management (OFTM), (b) vehicle-actuated management (VAM4) 630

Fig. 12 Obtained reduction rates in case of implementing of vehicle-actuated management 631

instead of optimum fixed-time management for each performance criteria 632

List of Tables 633

Table 1 Minimum green times for different detector locations 634

Table 2 Reference demand matrix created for analysis studies 635

27

Table 3 The objective function, the decision variables and the set of constraints for the signal 636

timing optimization problem 637

Table 4 Optimum signal timings and average vehicle delays for created traffic scenarios 638

Table 5 Scenario-based comparison of average vehicle delay values (sec/veh) 639

Table 6 Optimum detector placements for created scenarios 640

Table 7 For the cases of fluctuated and not-fluctuated traffic demands, obtained reduction rates 641

(success rates) 642

643

644

645

646

647

648

649

650

651

652

653

654

655

656

657

658

659

660

28

661

662

663

664

665

666

Fig. 1 The relation between average vehicle delay and the placement of detectors 667

668

669

670

671

672

673

674

675

676

677

678

679

Fig. 2 Vehicle-actuated management application at a four-leg signalized intersection in 680

VISSIM 681

682

683

684

685

Near minimum delay Near optimum distance

Av

era

ge

Del

ay

Distance between Detectors and Stop Line

North East

West

South

Phase I

Phase II

Phase III

Phase IV

Detector No 1 – 2 – 3 – 4

Detector No 5 – 6 – 7

Detector No 8 – 9 – 10 - 11

Detector No 12 – 13 – 14

VAM4

29

686

687

688

689

690

691

692

693

694

695

696

697

698

699

700

701

702

703

704

705

706

707

708

Fig. 3 The flow chart of vehicle-actuated control logic created with VAP 709

710

N

N

Y

Y

N

Y Y

N

N

N

Y Y

Y Y

Y Y

N

N

Y Y

N

Y

Y

N

N

N

N

N

N

Y Y

Y Y

Y Y

Y Y

N

Y Y

N

N

N

N

N

Y Y

N

Y Y

Y Y

Y Y

N

Y Y

N

N

N

N

N

N

Y Y

Y Y

Y Y

N

Y

Start (t=1)

Any interphase active?

Phase I active?

Phase II active?

Phase III active?

Intersection management period complete? (t=3600 sec.)

Obtain the average vehicle delay from VISSIM simulation program

Green Time

for Phase I

>=

Minimum

Green Time

for Phase I

Green Time

for Phase I

<

Maximum Green Time

for Phase I

Min. critical gap (1 or 2 or 3 or 4)

<=

⁓ 2 sec.

Occupancy (5 or 6 or 7)

>=

1 sec.

Occupancy (8 or 9 or 10 or 11)

>=

1 sec.


>=

1 sec.

From Phase I

to Phase II

From Phase I

to Phase III

From Phase I

to Phase IV

Green Time

for Phase II

>=

Minimum

Green Time

for Phase II

Green Time

for Phase II

<

Maximum Green Time

for Phase II

Min. critical gap (5 or 6 or 7)

<=

⁓ 2 sec.


>=

1 sec.


>=

1 sec.


>=

1 sec.

From Phase II

to Phase III

From Phase II

to Phase IV

From Phase II

to Phase I

Green Time

for Phase III

>=

Minimum

Green Time

for Phase III

Green Time

for Phase III

<

Maximum Green Time

for Phase III

Min. critical gap (8 or 9 or 10 or 11)

<=

⁓ 2 sec.


>=

1 sec.


>=

1 sec.


>=

1 sec.

Phase IV active?

From Phase III

to Phase IV

From Phase III

to Phase I

From Phase III

to Phase II

Green Time

for Phase IV

>=

Minimum

Green Time

for Phase IV

Green Time

for Phase IV

<

Maximum Green Time

for Phase IV

Min. critical gap (12 or 13 or 14)

<=

⁓ 2 sec.


>=

1 sec.


>=

1 sec.


>=

1 sec.

From Phase IV

to Phase I

From Phase IV

to Phase II

From Phase IV

to Phase III

t = t+1

30

711

Fig. 4 The change rates of movement direction-based traffic volumes for created scenarios 712

713

714

715

716

Fig. 5 Hourly total traffic volumes for all traffic scenarios 717

718

719

720

721

722

-50%

-40%

-30%

-20%

-10%

0%

10%

20%

30%

40%

50%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Ch

an

ge

Ra

tes

Scenerio No

The Change Rates of Movement Direction Based Traffic Volumes

Turning Right Going Straight Turning Left

Reference Demand Matrix

0

500

1000

1500

2000

2500

3000

3500

4000

4500

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

To

tal

Tra

ffic

Vo

lum

e a

t

Inte

rsec

tio

n (

veh

/hr)

Scenario No

Hourly Total Traffic Volumes for all Traffic Scenarios

31

Scenario

No

Intersection

Approach

Analysis Period (60 minutes)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scenario 1 West 44%

Scenario 2 West 33%

South 50%

Scenario 3 East 60%

Scenario 4 West 56%

East 56%

Scenario 5 North 33%

South 33%

Scenario 6

West 33% 28%

North 22% 22%

East 50%

Scenario 7

West 67%

East 42%

South 50%

Scenario 8

West 38%

North 38%

East 50%

South 38%

Scenario 9 West 50%

East 50%

Scenario 10 West 50%

South 50%


East 75%


East 50%


South 33% 33%


South 67%




South 38%


South 50%


South 33%

Scenario 20 East 50%

Fig. 6 Time intervals and quantities of fluctuations in traffic demands for each traffic scenario723

32

724

(a) (b) 725

Fig. 7 For created scenarios (a) the comparison of average vehicle delay values and (b) the 726

percentage (%) differences of average vehicle delay values 727

728

729

730

731

(a) (b) 732

Fig. 8 The effect of detector placements on average vehicle delay for (a) Scenario 5 and 733

(b) Scenario 17. 734

735

736

737

738

R² = 0.996

20

30

40

50

60

70

80

90

20 30 40 50 60 70 80 90

VIS

SIM

(se

c/veh

)

Akçelik (sec/veh)

VISSIM - Akçelik

The Comparison of Average Delay Values

-6%

-4%

-2%

0%

2%

4%

6%

1 2 3 4 5 6 7 8 9 1011121314151617181920

Per

cen

tag

e D

iffe

ren

ces

Scenario No

VISSIM - Akçelik

The Differences of Average Delay Values

21.76

20

22

24

26

28

30

32

15 20 25 30

Avera

ge V

eh

icle

Dela

y (

sec/v

eh

)

Distance between Detectors and Stop Line (m)

Scenario 5

63.762

66

70

74

78

35 40 45 50

Avera

ge V

eh

icle

Dela

y (

sec/v

eh

)

Distance between Detectors and Stop Line (m)

Scenario 17

33

739

740

(a) 741

742

743

744

745

746

747

748

0

15

30

45

60

1 2 3 4 5

Aver

ag

e V

ehic

le D

elay

(se

c/veh

)

Scenario

Average Vehicle Delay - Demand: Low

Not-Fluctuated (OFTM) Not-Fluctuated (VAM4)

Fluctuated (OFTM) Fluctuated (VAM4)

0

25

50

75

100

1 2 3 4 5

Aver

ag

e V

ehic

le D

elay

(se

c/veh

)

Scenario

Average Vehicle Delay - Demand: Moderate



0

25

50

75

100

1 2 3 4 5

Aver

ag

e V

ehic

le D

elay

(se

c/veh

)

Scenario

Average Vehicle Delay - Demand: High


Fluctuated (OFTM) Fluctuated (VAM4)20

40

60

80

100

120

1 2 3 4 5

Aver

ag

e V

ehic

le D

elay

(se

c/veh

)

Scenario

Average Vehicle Delay - Demand: Very High



34

749

750

(b) 751

752

753

754

755

756

757

758

759

760

761

80

120

160

200

240

280

1 2 3 4 5

Fu

el C

on

sum

pti

on

(l/

hr)

Scenario

Fuel Consumption - Demand: Low



150

200

250

300

350

400

450

1 2 3 4 5

Fu

el C

on

sum

pti

on

(l/

hr)

Scenario

Fuel Consumption - Demand: Moderate



150

200

250

300

350

400

450

500

1 2 3 4 5

Fu

el C

on

sum

pti

on

(l/

hr)

Scenario

Fuel Consumption - Demand: High



300

350

400

450

500

550

1 2 3 4 5

Fu

el C

on

sum

pti

on

(l/

hr)

Scenario

Fuel Consumption - Demand: Very High



35

762

763

(c) 764

Fig. 9 At the end of the analysis, for created scenarios (a) obtained average vehicle delay 765

results; (b) obtained fuel consumption results; (c) obtained exhaust emission results 766

767

768

769

770

771

772

773

774

775

776

2000

3000

4000

5000

6000

1 2 3 4 5

Ex

hau

st E

mis

sion

(g

/hr)

Scenario

Exhaust Emission - Demand: Low



2000

4000

6000

8000

10000

1 2 3 4 5

Ex

hau

st E

mss

ion

(g

/hr)

Scenario

Exhaust Emission - Demand: Moderate



2000

4000

6000

8000

10000

12000

1 2 3 4 5

Ex

hau

st E

mis

sion

(g

/hr)

Scenario

Exhaust Emission - Demand: High



7000

9000

11000

13000

1 2 3 4 5

Ex

hau

st E

mis

sion

(g

/hr)

Scenario

Exhaust Emission - Demand: Very High



36

777

778

779

780

781

782

783

(a) (b) 784

Fig. 10 For the Scenario 3, in time interval when the fluctuations occur in traffic demands, 785

sample intersection view in case of implementing (a) the optimum-fixed time management 786

(OFTM) (b) vehicle-actuated management (VAM4). 787

788

789

790

791

792

793

794

795

796

797

(a) (b) 798

Fig. 11 For Scenario 3, when the fluctuations do not exist in traffic demands, sample 799

intersection view in time same time interval in case of implementing (a) the optimum fixed-800

time management (OFTM), (b) vehicle-actuated management (VAM4). 801

802

803

Scenario 3

West West

North North

East East

South South

VAM4 OFTM

Simulation Time 1475 sec.


VAM4 OFTM

West West

South South

North North

East East



Scenario 3

37

804

805

806

Fig. 12 Obtained reduction rates in case of implementing of vehicle-actuated management 807

instead of optimum fixed-time management for each performance criteria. 808

809

810

811

0%

10%

20%

30%

40%

50%

60%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Red

uct

ion

Ra

tes

Scenario No

Average Vehicle Delay

Demand: Not-Fluctuated

Demand: Fluctuated

0%

10%

20%

30%

40%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Red

uct

ion

Ra

tes

Scenario No

Fuel Consumption


Demand: Fluctuated

0%

10%

20%

30%

40%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Red

uct

ion

Ra

tes

Scenario No

Exhaust Emission


Demand: Fluctuated

38

Table 1 Minimum green times for different detector locations 812

Option O-1 O-2 O-3 O-4 O-5 O-6 O-7

Distance between detectors and stop line (m) 15 20 25 30 35 40 45

Minimum green times (sec) 5 7 9 11 13 15 17

*O-1, 2, …..7 shows the options of distance between detector and stop line. 813

814

815

816

Table 2 Reference demand matrix created for analysis studies. 817

Volumes (veh/hr) Flow Rates

Intersection

Approach

(O-D)

West North East South

Intersection

Approach

(O-D)

West North East South

West - 200 700 100 West - 0.200 0.700 0.100

North 150 - 100 150 North 0.375 - 0.250 0.375

East 600 100 - 150 East 0.706 0.118 - 0.176

South 200 250 200 - South 0.308 0.384 0.308

West Intersection Approach: 1000 veh/hr

North Intersection Approach: 400 veh/hr

East Intersection Approach: 850 veh/hr

South Intersection: 650 veh/hr

818

819

820

Table 3 The objective function, the decision variables and the set of constraints for the signal 821

timing optimization problem. 822

Objective

Function ( )f x

n = 1, 2, …,14

( n = Total number of lanes at

intersection)

( )1 1

n n

i i

i i

f x min d D q= =

= =

Decision

Variables

i = 1, 2, …,4

( i = Phase no) ig : Green signal timing (green

split) for phase i (sec)

The Set of

Constraints

Constraints for minimum and maximum

green signal timing ( ig )

i = 1, 2, …,4

( i = Phase no)

7 45ig

Constraints for degree of saturation ( nx )

n = 1, 2, …,14

( n = Total number of lanes at

intersection)

0 1 2nx .

Total Traffic Flow at Intersection: 2900

veh/hr

39

Table 4 Optimum signal timings and average vehicle delays for created traffic scenarios 823 D

em

an

d

Scen

ario

No

Ph

ase

I (

sec)

Ph

ase

II

(sec

)

Ph

ase

III

(s

ec)

Ph

ase

IV

(se

c)

Cy

cle

Tim

e

(sec

)

Av

era

ge

Veh

icle

Dela

y

(sec

/veh

)

Dem

an

d

Scen

ario

No

Ph

ase

I (

sec)

Ph

ase

II

(sec

)

Ph

ase

III

(s

ec)

Ph

ase

IV

(se

c)

Cy

cle

Tim

e

(sec

)

Av

era

ge

Veh

icle

Dela

y

(sec

/veh

)

Lo

w

1 10 9 9 12 60 25.00

Hig

h

11 19 9 17 14 79 34.00

2 9 7 8 9 53 20.82 12 25 11 22 17 95 43.82

3 19 8 16 13 76 33.25 13 24 11 21 17 93 44.63

4 13 9 11 11 64 26.89 14 25 12 22 18 97 45.55

5 12 7 10 9 58 23.60 15 31 14 27 22 114 55.59

Mo

dera

te 6 14 8 12 11 65 28.32

Very H

igh

16 45 19 39 31 154 78.83

7 19 8 16 13 76 32.84 17 45 20 39 31 155 82.37

8 19 9 17 13 78 33.30 18 31 15 27 22 115 57.56

9 (R) 19 9 16 13 77 33.20 19 45 20 39 31 155 81.21

10 18 9 16 13 76 34.51 20 45 20 39 31 155 80.26

Phase I: West ; Phase II: North / Phase III: East / Phase IV: South

Inter Green Time: 5 sec.

(R): Reference Demand

824

825

Table 5 Scenario-based comparison of average vehicle delay values (sec/veh) 826

Dem

an

d

Scen

ario

No

Ak

çeli

k

(sec

/veh

)

VIS

SIM

(sec

/veh

)

Dem

an

d

Scen

ario

No

Ak

çeli

k

(sec

/veh

)

VIS

SIM

(sec

/veh

)

Dem

an

d

Scen

ario

No

Ak

çeli

k

(sec

/veh

)

VIS

SIM

(sec

/veh

)

Dem

an

d

Scen

ario

No

Ak

çeli

k

(sec

/veh

)

VIS

SIM

(sec

/veh

)

Lo

w

1 25.00 25.51

Mo

dera

te

6 28.32 27.52

Hig

h

11 34.00 34.47

Very H

igh

16 78.83 82.44

2 20.82 21.68 7 32.84 33.38 12 43.82 42.32 17 82.37 84.21

3 33.25 33.91 8 33.30 31.74 13 44.63 42.73 18 57.56 56.16

4 26.89 26.60 9 33.20 32.23 14 45.55 44.82 19 81.21 79.88

5 23.60 23.76 10 34.51 34.94 15 55.59 56.10 20 80.26 79.55

827

828

Table 6 Optimum detector placements for created scenarios 829

Scenario

No

Distance between Detectors

and Stop Line (m) Scenario

No

Distance between Detectors

and Stop Line (m)

West

Approach

North

Approach

East

Approach

South

Approach

West

Approach

North

Approach

East

Approach

South

Approach

1 15 15 15 15 11 20 20 20 20

2 15 15 15 15 12 30 30 30 30

3 15 15 15 15 13 30 30 30 30

4 15 15 15 15 14 35 35 35 35

5 15 15 15 15 15 45 45 45 45

6 15 15 15 15 16 45 45 45 45

7 20 20 20 20 17 45 45 45 45

8 20 20 20 20 18 45 45 45 45

9 25 25 25 25 19 45 45 45 45

10 20 20 20 20 20 45 45 45 45

830

40

Table 7 For the cases of fluctuated and not-fluctuated traffic demands, obtained 831

reduction rates (success rates) 832

Performance Criteria

For created scenarios, obtained reduction rates (%)

in case of implementing of VAM4 instead of OFTM

Demand: Not - Fluctuated Demand: Fluctuated

Minimum Maximum Average Minimum Maximum Average

Average vehicle delay 4.51 29.16 14.20 15.58 58.55 32.54

Fuel Consumption 2.80 17.68 7.29 9.62 38.41 21.83

Exhaust Emission 2.79 17.68 7.29 9.60 38.40 21.82

833

834

A Brief Technical Biography of Authors 835

Ziya CAKICI received B.Sc. degree from Civil Engineering Department of Celal Bayar 836

University, Manisa, Turkey in 2010. He received the M.Sc. and Ph.D. degrees in transportation 837

engineering from Graduate School of Natural and Applied Sciences at Pamukkale University 838

(Turkey), in 2014 and 2020, respectively. He has been working as Research Assistant in 839

Department of Civil Engineering at Bayburt University. He studies in the fields of traffic 840

management (especially Intelligent Transportation Systems), traffic and transportation 841

planning and traffic optimization. He has published several papers in different journals and 842

international conference proceedings. 843

844

Yetis Sazi MURAT received B.Sc. degree from Civil Engineering Department of Dokuz Eylul 845

University, Izmir, Turkey in 1992. He has received M.Sc. degree from Pamukkale University 846

in 1996 and Ph.D. degree from Istanbul Technical University in 2001. He has been working as 847

Professor in Department of Civil Engineering at Pamukkale University. He worked as visiting 848

scholar at Virginia Tech. University, Falls Church, USA in 2006 (6 months) and University of 849

Nevada, Reno, USA in 2017. He has published many technical-scientific research papers and 850

conducted many projects on different subjects of Transportation Engineering. His main 851

research interests include fuzzy logic and neural network based traffic management approaches. 852

He studies in the fields of traffic management (especially Intelligent Transportation Systems), 853

traffic and transportation planning, traffic accidents and etc. 854

855

Metin Mutlu AYDIN received B.Sc. degree from Civil Engineering Department of Dokuz 856

Eylul University, Izmir, Turkey in 2010. He received the M.Sc. degree from Dokuz Eylul 857

University in 2012 and Ph.D. degree from Akdeniz University, Antalya, Turkey in 2017. He 858

41

has been working Associate Professor in Department of Civil Engineering at Ondokuz Mayıs 859

University. He studies in the fields of traffic and transportation engineering. His research 860

interests include modelling of the lane discipline, optimizing of the intersection geometry, 861

simulating of driver behaviours, in generally. He has published many research papers in 862

different journals and international conference proceedings. 863

864

design of an efficient vehicle-actuated signal control...

Documents