Traffic Flow Theory 2001

Traffic Flow Theory
A State-of-the-Art Report
Revised
2001

Traffic Flow Theory
A State-of-the-Art Report
Revised
2001
Organized by the Committee on Traffic Flow Theory and Characteristics (AHB45)

TABLE OF CONTENTS
1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1
1.1 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4
2. TRAFFIC STREAM CHARACTERISTICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1

2.1 Definitions and Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
2.1.1 The Time-Space Diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
2.1.2 Definitions of Some Traffic Stream Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
2.1.3 Time-Mean and Space-Mean Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-4
2.1.4 Generalized Definitions of Traffic Stream Properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-4
2.1.5 The Relation Between Density and Occupancy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-6
2..1.6 Three-Dimensional Representation of Vehicle Streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-8
2.2 Measurement Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9
2.2.1 Measurement Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9
2.2.2 Error Caused by the Mismatch Between Definitions and Usual Measurements .. . . . . . . . . . . 2-12
2.2.3 Importance of Location to the Nature of the Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-13
2.2.4 Selecting intervals from which to extract data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-14
2.3 Bivariate Models .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-15
2.3.1 Speed-Flow Models .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-16
2.3.2 Speed-Concentration Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-21
2.3.3 Flow-Concentration Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-26
2.4 Three-Dimensional Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-29
2.5 Summary and Llinks to Other Chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-31
3. HUMAN FACTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1
3.1.1 The Driving Task . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1
3.2 Discrete Driver Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-3
3.2.1 Perception-Response Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-3
3.3 Control Movement Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7
3.3.1 Braking Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7
3.3.2 Steering Response Times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.4 Response Distances and Times to Traffic Control Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.4.1 Traffic Signal Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.4.2 Sign Visibility and Legibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
3.4.3 Real-Time Displays and Signs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
3.4.4 Reading Time Allowance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13
3.5 Response to Other Vehicle Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13
3.5.1 The Vehicle Ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13
3.5.2 The Vehicle Alongside . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14
3.6 Obstacle and Hazard Detection, Recognition, and Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
3.6.1 Obstacle and Hazard Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
3.6.2 Obstacle and Hazard Recognition and Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
3.7 Individual Differences in Driver Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16
3.7.1 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16
3.7.2 Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16
3.7.3 Driver Impairment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
i
3.8 Continuous Driver Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-18
3.8.1 Steering Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-18
3.8.1.1 Human Transfer Function for Steering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-18
3.8.1.2 Performance Characteristics Based on Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19
3.9 Braking Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-20
3.9.1 Open-Loop Braking Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-20
3.9.2 Closed-Loop Braking Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
3.9.3 Less-Than-Maximum Braking Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
3.10 Speed and Acceleration Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.10.1 Steady-State Traffic Speed Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.10.2 Acceleration Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.11 Specific Maneuvers at the Guidance Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.11.1 Overtaking and Passing in the Traffic Stream . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.11.1.1 Overtaking and Passing Vehicles (4-Lane or 1-Way) . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.11.1.2 Overtaking and Passing Vehicles (Opposing Traffic) . . . . . . . . . . . . . . . . . . . . . . . 3-24
3.12 Gap Acceptance and Merging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-24
3.12.1 Gap Acceptance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-24
3.12.2 Merging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-24
3.13 Stopping Sight Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-25
3.14 Intersection Sight Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-26
3.14.1 Case I: No Traffic Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-26
3.14.2 Case II: Yield Control for Secondary Roadway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-26
3.14.3 Case III: Stop Control on Secondary Roadway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-26
3.15 Other Driver Performance Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27
3.15.1 Speed Limit Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27
3.15.2 Distractors On/Near Roadway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27
3.15.3 Real-Time Driver Information Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
4. CAR FOLLOWING MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-1

4.1 Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2
4.2 Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6
4.2.1 Local Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6
4.2.2 Asymptotic Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9
4.2.1.1 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10
4.2.1.2 Next-Nearest Vehicle Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13
4.3 Steady-State Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14
4.4 Experiments And Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-20
4.4.1 Car Following Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-22
4.4.1.1 Analysis of Car Following Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-23
4.4.2 Macroscopic Observations: Single Lane Traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-32
4.5 Automated Car Following . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-38
4.6 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-38
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-39
5. CONTINUUM FLOW MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-1

5.1 Conservation and Traffic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-1
5.2 The Kinematic Wave Model of LWR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-6
5.2.1 The LWR Model and Characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-6
5.2.2 The Riemann Problem and Entropy Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-7
5.2.3 Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8
ii
5.2.4 Extensions to the LWR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-9
5.2.5 Limitations of the LWR Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-11
5.3 High Order Continuum Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-13
5.3.1 Propagation of Traffic Sound Waves in Higher-Order Models . . . . . . . . . . . . . . . . . . . . . . . . . 5-15
5.3.2 Propagation of Shock and Expansion Waves .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-16
5.3.3 Traveling Waves, Instability and Roll Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-20
5.3.4 Summary and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-23
5 .4 Diffusive, Viscous and Stochastic Traffic Flow Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-24
5.4.1 Diffusive and Viscous Traffic Flow Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-24
5.4.2 Acceleration Noise and a Stochastic Flow Model .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25
5.5 Numerical Approximations of Continuum Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25
5.5.1 Finite Difference Methods for SOlving Inviscid Models .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27
5.5.2 Finite Element Methods for Solving Viscous Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-30
5.5.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-33
5.5.3.1 Calibration of Model Parameters with Field Measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . 5-33
5.5.3.2 Multilane Traffic Flow Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-35
5.5.3.3 Traffic Flow on a Ring Road With a Bottleneck. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-35
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-45
6. MACROSCOPIC FLOW MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-1

6.1 Travel Time Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-1
6.1.1 General Traffic Characteristics as a Function of the Distance from the CBD . . . . . . . . . . . . . . . 6-2
6.1.2 Average Speed as a Function of Distance from the CBD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-3
6.2 General Network Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-6
6.2.1 Network Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-6
6.2.2 Speed and Flow Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-8
6.2.3 General Network Models Incorporating Network Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 6-11
6.2.4 Continuum Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-16
6.3 Two-Fluid Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-16
6.3.1 Two-Fluid Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-17
6.3.2 Two-Fluid Parameters: Influence of Driver Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-20
6.3.3 Two-Fluid Parameters: Influence of Network Features (Field Studies) . . . . . . . . . . . . . . . . . . 6-20
6.3.4 Two-Fluid Parameters: Estimation by Computer Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 6-22
6.3.5 Two-Fluid Parameters: Influence of Network Features (Simulation Studies) . . . . . . . . . . . . . 6-22
6.3.6 Two-Fluid Model: A Practical Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-23
6.4 Two-Fluid Model and Traffic Network Flow Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-23
6.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-25
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-29
7. TRAFFIC IMPACT MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1

7.1 Traffic and Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1
7.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1
7.1.2 Flow and Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1
7.1.3 Logical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-2
7.1.4 Empirical Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.1.4.1 Kinds Of Study And Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.1.4.2 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.1.4.3 Parameter Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
7.1.5 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-7
iii
7.2 Fuel Consumption Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-8
7.2.1 Factors Influencing Vehicular Fuel Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-8
7.2.2 Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-8
7.2.3 Urban Fuel Consumption Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
7.2.4 Highway Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-11
7.2.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-12
7.3 Air Quality Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-13
7.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-13
7.3.2 Air Quality Impacts of Transportation Control Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-13
7.3.3 Tailpipe Control Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-14
7.3.4 Highway Air Quality Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-15
7.3.4.1 UMTA Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-15
7.3.4.2 CALINE-4 Dispersion Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-15
7.3.4.3 Mobile Source Emission Factor Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-16
7.3.4.4 MICRO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-18
7.3.4.5 The TRRL Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-19
7.3.5 Other Mobile Source Air Quality Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-20
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-20
8. UNSIGNALIZED INTERSECTION THEORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1
8.1.1 The Attributes of a Gap Acceptance Analysis Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1
8.1.2 Interaction of Streams at Unsignalized Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1
8.1.3 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1
8.2 Gap Acceptance Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-2
8.2.1 Usefulness of Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-2
8.2.2 Estimation of the Critical Gap Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
8.2.3 Distribution of Gap Sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.3 Headway Distributions Used in Gap Acceptance Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.3.1 Exponential Headways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.3.2 Displaced Exponential Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7
8.3.3 Dichotomized Headway Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7
8.3.4 Fitting the Different Headway Models to Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-8
8.4 Interaction of Two Streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-11
8.4.1 Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-11
8.4.2 Quality of Traffic Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-16
8.4.3 Queue Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-19
8.4.4 Stop Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-22
8.4.5 Time Dependent Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-23
8.4.6 Reserve Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-26
8.4.7 Stochastic Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-27
8.5 Interaction of Two or More Streams in the Priority Road . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-28
8.5.1 The Benefit of Using a Multi-Lane Stream Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-28
8.6 Interaction of More than Two Streams of Different Ranking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-31
8.6.1 Hierarchy of Traffic Streams at a Two Way Stop Controlled Intersection . . . . . . . . . . . . . . . . 8-31
8.6.2 Capacity for Streams of Rank 3 and Rank 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-32
8.7 Shared Lane Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-35
8.7.1 Shared Lanes on the Minor Street . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-35
8.7.2 Shared Lanes on the Major Street . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-35
8.8 Two-Stage Gap Acceptance and Priority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-36
8.9 All-Way Stop Controlled Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-37
8.9.1 Richardson’s Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-37
iv
8.10 Empirical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-39
8.10.1 Kyte's Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-39
8.11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-41
9. TRAFFIC FLOW AT SIGNALIZED INTERSECTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-1

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-1
9.2 Basic Concepts of Delay Models at Isolated Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-2
9.3 Steady-State Delay Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
9.3.1 Exact Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
9.3.2 Approximate Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-5
9.4 Time-Dependent Delay Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
9.5 Effect of Upstream Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-15
9.5.1 Platooning Effect On Signal Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-15
9.5.2 Filtering Effect on Signal Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-17
9.6 Theory of Actuated and Adaptive Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-19
9.6.1 Theoretically-Based Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-19
9.6.2 Approximate Delay Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-23
9.6.3 Adaptive Signal Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-27
9.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-27
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-28
10. TRAFFIC SIMULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-1

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-1
10.2 An Illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-1
10.3 Car-Following . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-2
10.4 Random Number Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-2
10.5 Classification of Simulation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-3
10.6 Building Simulation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-5
10.7 Illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-5
10.8 Statistical Analysis of Simulation Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-17
10.8.1 Statistical Analysis for a Single System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-17
10.8.1.1 Fixed Sample-Size Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-20
10.8.1.2 Sequential Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-21
10.8.2 Alternative System Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-22
10.8.3 Variance Reduction Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-22
10.8.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-23
10.9 Descriptions of Some Available Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-23
10.10 Looking to the Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-24
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-25
11. KINETIC THEORIES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-1

11.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-1
11.2 Status of the Prigogine-Herman Kinetic Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11-2
11.2.1 The Prigogine-Herman Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-2
11.2.2 Criticisms of the Prigogine-Herman Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-3
11.2.3 Accomplishments of the Prigogine-Herman Model. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-4
11.3 Other Kinetic Models
11.4 Continuum Models from Kinetic Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-6
11.5 Direct Solution of Kinetic Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-7
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..11-9
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-1
v
LIST OF FIGURES
2. TRAFFIC STREAM CHARACTERISTICS

Figure 2.1
Time-space Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
Figure 2.2
Trajectories in Time-space Region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5
Figure 2.3
Trajectories of Vehicle Fronts and Rears. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-7
Figure 2.4
Three-dimensional representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9
Figure 2.5
Effect of measurement location on nature of data (modified from Hall, Hurdle, Banks 1992,and May 1990. . 2-17
Figure 2.6
Generalized shape of speed-flow curve proposed by Hall, Hurdle and Banks (1992). . . . . . . . . . . . . . . . . . . 2-17
Figure 2.7
Generalized shape of speed-flow curve proposed by Hall, Hurdle and Banks (1992).. . . . . . . . . . . . . . . . . . . 2-18
Figure 2.8
Results from fitting polygon speed-flow curve to German data (Heidemann and Hotop). . . . . . . . . . . . . . . . . 2-18
Figure 2.9
Data for 4-lane German Autobahns (2 lanes per direction), as reported by Stappert and Theis(1990). . . . . 2-20
Figure 2.10
Greenshields' Speed-Flow Curve and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-20
Figure 2.11
Greenshields' Speed-Density Graph and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-23
Figure 2.12
Speed-Concentration Data from Merritt Parkway and Fitted Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-23
Figure 2.13
Three Parts of Edie's Hypothesis for the Speed-Density Function, Fitted to Chicago Data . . . . . . . . . . . . . . . 2-25
Figure 2.14
Greenshields' Speed-Flow Function Fitted to Chicago Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-28
Figure 2.15
Four Days of Flow-Occupancy Data from Near Toronto . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-28
Figure 2.16
The Three-Dimensional Surface for Traffic Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-30
Figure 2.17
One Perspective on Three-dimensional Relationship (Gilchrist and Hall) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-30
Figure 2.18
Second Perspective on Three-Dimensional Relationship (Gilchrist and Hall). . . . . . . . . . . . . . . . . . . . . . . . . . 2-32
Figure 2.19
Catastrophe Theory Surface Showing Sketch of a Possible Freeway Function. . . . . . . . . . . . . . . . . . . . . . . . 2-32
vi
3. HUMAN FACTORS
Figure 3.1
Generalized Block Diagram of the Car-Driver-Roadway System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-2
Figure 3.2
Lognormal Distribution of Perception-Reaction Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-4
Figure 3.3
A Model of Traffic Control Device Information Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10
Figure 3.4
Looming as a Function of Distance from Object. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14
Figure 3.5
Pursuit Tracking Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19
Figure 3.6
Typical Deceleration Profile for a Driver without Antiskid Braking System on a Dry Surface. . . . . . . . . . . . . . 3-22
Figure 3.7
Typical Deceleration Profile for a Driver without Antiskid Braking System on a Wet Surface. . . . . . . . . . . . . 3-22
4. CAR FOLLOWING MODELS

Figure 4.1
Schematic Diagram of Relative Speed Stimulus and a Weighing Function Versus Time . . . . . . . . . . . . . . . . . 4-4
Figure 4.1a
Block Diagram of Car-Following . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5
Figure 4.1b
Block Diagram of the Linear Car-Following Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5
Figure 4.2
Detailed Motion of Two Cars Showing the Effect of a Fluctuation in the Acceleration of the Lead Car . . . . . . 4-8
Figure 4.3
Changes in Car Spacings from an Original Constant Spacing Between Two Cars . . . . . . . . . . . . . . . . . . . . . . 4-9
Figure 4.4
Regions of Asymptotic Stability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-11
Figure 4.5
Inter-Vehicle Spacings of a Platoon of Vehicles Versus Time for the Linear Car Following. . . . . . . . . . . . . . . 4-11
Figure 4.6
Asymptotic Instability of a Platoon of Nine Cars. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12
Figure 4.7
Envelope of Minimum Inter-Vehicle Spacing Versus Vehicle Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13
Figure 4.8
Inter-Vehicle Spacings of an Eleven Vehicle Platoon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14
Figure 4.9
Speed (miles/hour) Versus Vehicle Concentration (vehicles/mile). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17
Figure 4.10
Normalized Flow Versus Normalized Concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17
Figure 4.11
Speed Versus Vehicle Concentration(Equation 4.39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-18
Figure 4.12
Normalized Flow Versus Normalized Vehicle Concentration (Equation 4.40) . . . . . . . . . . . . . . . . . . . . . . . . . 4-18
Figure 4.13
Normalized Flow Versus Normalized Concentration (Equations 4.51 and 4.52) . . . . . . . . . . . . . . . . . . . . . . 4-21
vii
Figure 4.14
Normalized Flow versus Normalized Concentration Corresponding to the
Steady-State Solution of Equations 4.51 and 4.52 for m=1 and Various Values of # . . . . . . . . . . . . . . . . . . . . 4-21
Figure 4.15
Sensitivity Coefficient Versus the Reciprocal of the Average Vehicle Spacing. . . . . . . . . . . . . . . . . . . . . . . . . 4-24
Figure 4.16
Gain Factor, , Versus the Time Lag, T, for All of the Test Runs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-24
Figure 4.17
Gain Factor, , Versus the Reciprocal of the Average Spacing for Holland Tunnel Tests. . . . . . . . . . . . . . . . 4-25
Figure 4.18
Gain Factor, ,Versus the Reciprocal of the Average Spacing for Lincoln Tunnel Tests . . . . . . . . . . . . . . . . 4-26
Figure 4.19
Sensitivity Coefficient, a0,0 ,Versus the Time Lag, T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-28
Figure 4.20
Sensitivity Coefficient Versus the Reciprocal of the Average Spacing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-29
Figure 4.21
Sensitivity Coefficient Versus the Ratio of the Average Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-29
Figure 4.22
Relative Speed Versus Spacing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-31
Figure 4.23
Relative Speed Thresholds Versus Inter-Vehicle Spacing for Various Values of the Observation Time. . . . . 4-32
Figure 4.24
Speed Versus Vehicle Concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-34
Figure 4.25
Flow Versus Vehicle Concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-34
Figure 4.26
Speed Versus Vehicle Concentration (Comparison of Three Models) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-35
Figure 4.27
Flow Versus Concentration for the Lincoln and Holland Tunnels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-36
Figure 4.28
Average Speed Versus Concentration for the Ten-Bus Platoon Steady-State Test Runs . . . . . . . . . . . . . . . . 4-37
5. CONTINUUM FLOW MODELS
Figure 5.1
Geometric Representation of Shocks, Sound Waves and Traffic Speeds in the k-q phase plane . . . . . . . . . . 5-4
Figure 5.2
Field Representation of Shocks and Conservation of Flow. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . 5-5
Figure 5.3
A Shock Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8
Figure 5.4
A Rarefaction Solution.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8
Figure 5.5
Phase Transition Diagram in the Solution of Riemann Problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-20
Figure 5.6
Roll Waves in the Moving Coordinate X . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-22
Figure 5.7
Traveling Waves and Shocks in the PW Modelic Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-22
Figure 5.8
Time-space Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-26
viii
Figure 5.9
The Kerner-Konhauser Model of Speed-Density and Flow-Density Relations. . . . . . . . . . . . . . . . . . . . . . . . . 5-36
Figure 5.10
Initial Condition (114) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-36
Figure 5.11
Solutinos of the Homogeneous LWR Model With Initial Condition in Figure 10 . . . . . . . . . . . . . . . . . . . . . . . . 5-37
Figure 5.12
Initial Condition (116) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-38
Figure 5.13
Solutions of the Inhomogeneous LWR Model With Initial Condition (116). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-39
Figure 5.14
Solutions of the PW Model With Initial Condition (117). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-41
Figure 5.15
Solutions of the PW Model With Initial Condition (118). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-42
Figure 5.16
Comparison of the LWR Model and the PW Model on a Homogeneous Ring Road . . . . . . . . . . . . . . . . . . . 5-43
Figure 5.17
Comparison of the LWR Model and the PW Model on an Inhomogeneous Ring Road. . . . . . . . . . . . . . . . . . 5-44
ix
6. MACROSCOPIC FLOW MODELS
Figure 6.1
Total Vehicle Distance Traveled Per Unit Area on Major Roads as a Function
of the Distance from the Town Center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
Figure 6.2
Grouped Data for Nottingham Showing Fitted (a) Power Curve,
(b) Negative Exponential Curve, and (c) Lyman-Everall Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-4
Figure 6.3
Complete Data Plot for Nottingham; Power Curve Fitted to the Grouped Data . . . . . . . . . . . . . . . . . . . . . . . . . 6-4
Figure 6.4
Data from Individual Radial Routes in Nottingham, Best Fit Curve for Each Route is Shown . . . . . . . . . . . . . . 6-5
Figure 6.5
Theoretical Capacity of Urban Street Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-7
Figure 6.6
Vehicles Entering the CBDs of Towns Compared with the Corresponding
Theoretical Capacities of the Road Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-7
Figure 6.7
Speeds and Flows in Central London, 1952-1966, Peak and Off-Peak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-8
Figure 6.8
Speeds and Scaled Flows, 1952-1966 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-9
Figure 6.9
Estimated Speed-Flow Relations in Central London (Main Road Network) . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-9
Figure 6.10
Speed-Flow Relations in Inner and Outer Zones of Central Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-10
Figure 6.11
Effect of Roadway Width on Relation Between Average (Journey) Speed and Flow in Typical Case . . . . . . 6-12
Figure 6.12
Effect of Number of Intersections Per Mile on Relation Between
Average (Journey) Speed and Flow in Typical Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-12
Figure 6.13
Effect of Capacity of Intersections on Relation Between
Average (Journey) Speed and Flow in Typical Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-13
Figure 6.14
Relationship Between Average (Journey) Speed and Number of Vehicles on Town Center Network . . . . . 6-13
Figure 6.15
Relationship Between Average (Journey) Speed of Vehicles and Total Vehicle Mileage on Network . . . . . 6-14
Figure 6.16
The -Relationship for the Arterial Networks of London and Pittsburgh, in Absolute Values . . . . . . . . . . . . . 6-14
Figure 6.17
The -Relationship for the Arterial Networks of London and Pittsburgh, in Relative Values . . . . . . . . . . . . . 6-15
Figure 6.18
The -Map for London, in Relative Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-16
Figure 6.19
Trip Time vs. Stop Time for the Non-Freeway Street Network of the Austin CBD . . . . . . . . . . . . . . . . . . . . . 6-18
Figure 6.20
Trip Time vs. Stop Time Two-Fluid Model Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-19
Figure 6.21
Trip Time vs. Stop Time Two-Fluid Model Trends Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-19
Figure 6.22
Two-Fluid Trends for Aggressive, Normal, and Conservative Drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-21
x
Figure 6.23
Simulation Results in a Closed CBD-Type Street Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-24
Figure 6.24
Comparison of Model System 1 with Observed Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-26
Figure 6.25
Figure 6.26
7. TRAFFIC IMPACT MODELS

Figure 7.1
Safety Performance Function and Accident Rate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-2
Figure 7.2
Shapes of Selected Model Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-5
Figure 7.3
Two Forms of the Model in Equation 7.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
Figure 7.4
Fuel Consumption Data for a Ford Fairmont (6-Cyl.)
Data Points represent both City and Highway Conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
Figure 7.5
Fuel Consumption Versus Trip Time per Unit Distance for a Number of Passenger Car Models. . . . . . . . . . 7-10
Figure 7.6
Fuel Consumption Data and the Elemental Model Fit for Two Types of Passenger Cars . . . . . . . . . . . . . . . 7-10
Figure 7.7
Constant-Speed Fuel Consumption per Unit Distance for the Melbourne University Test Car . . . . . . . . . . . . 7-12
8. UNSIGNALIZED INTERSECTION THEORY

Figure 8.1
Data Used to Evaluate Critical Gaps and Move-Up Times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
Figure 8.2
Regression Line Types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-4
Figure 8.3
Typical Values for the Proportion of Free Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
Figure 8.4
Exponential and Displaced Exponential Curves (Low flows example). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
Figure 8.5
Arterial Road Data and a Cowan (1975) Dichotomized Headway Distribution (Higher flows example). . . . . 8-10
Figure 8.6
Arterial Road Data and a Hyper-Erlang Dichotomized Headway Distribution (Higher Flow Example) . . . . . . 8-10
Figure 8.7
Illustration of the Basic Queuing System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-12
Figure 8.8
ComparisonRelation Between Capacity (q-m) and Priority Street Volume (q-p) . . . . . . . . . . . . . . . . . . . . . . . 8-14
Figure 8.9
Comparison of Capacities for Different Types of Headway Distributions in the Main Street Traffic Flow . . . . 8-14
Figure 8.10
The Effect of Changing in Equation 8.31 and Tanner's Equation 8.36. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-15
Figure 8.11
Probability of an Empty Queue: Comparison of Equations 8.50 and 8.52. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-18
xi
Figure 8.12
Comparison of Some Delay Formulae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-20
Figure 8.13
Average Steady State Delay per Vehicle Calculated Using Different Headway Distributions. . . . . . . . . . . . . . 8-20
Figure 8.14
Average Steady State Delay per Vehicle by
Geometric Platoon Size Distribution and Different Mean Platoon Sizes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-21
Figure 8.15
95-Percentile Queue Length Based on Equation 8.59 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-22
Figure 8.16
Approximate Threshold of the Length of Time Intervals For the Distinction
Between Steady-State Conditions and Time Dependent Situations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-25
Figure 8.17
The Co-ordinate Transform Technique. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-25
Figure 8.18
A Family of Curves Produced from the Co-Ordinate Transform Technique. . . . . . . . . . . . . . . . . . . . . . . . . . . 8-27
Figure 8.19
Average Delay, D, in Relation to Reserve Capacity R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-29
Figure 8.20
Modified 'Single Lane' Distribution of Headways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-30
Figure 8.21
Percentage Error in Estimating Adams' Delay Against the
Major Stream Flow for a Modified Single Lane Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-31
Figure 8.22
Traffic Streams And Their Level Of Ranking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-32
Figure 8.23
Reduction Factor to Account for the Statistical Dependence Between Streams of Ranks 2 and 3. . . . . . . . . 8-33
Figure 8.24
Minor Street Through Traffic (Movement 8) Crossing the Major Street in Two Phases. . . . . . . . . . . . . . . . . . 8-36
Figure 8.25
Average Delay For Vehicles on the Northbound Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-40
9. TRAFFIC FLOW AT SIGNALIZED INTERSECTIONS

Figure 9.1
Deterministic Component of Delay Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-2
Figure 9.2
Queuing Process During One Signal Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
Figure 9.3
Percentage Relative Errors for Approximate Delay Models by Flow Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9
Figure 9.4
Relative Errors for Approximate Delay Models by Green to Cycle Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9
Figure 9.5
The Coordinate Transformation Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-11
Figure 9.6
Comparison of Delay Models Evaluated by Brilon and Wu (1990) with Moderate Peaking (z=0.50). . . . . . . . 9-14
Figure 9.7
Comparison of Delay Models Evaluated by Brilon and Wu (1990) with High Peaking (z=0.70). . . . . . . . . . . 9-14
Figure 9.8
Observations of Platoon Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-16
Figure 9.9
HCM Progression Adjustment Factor vs Platoon Ratio Derived from TRANSYT-7F . . . . . . . . . . . . . . . . . . . 9-18
xii
Figure 9.10
Analysis of Random Delay with Respect to the Differential Capacity Factor (f) and
Var/Mean Ratio of Arrivals (I)- Steady State Queuing Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-19
Figure 9.11
Queue Development Over Time Under Fully-Actuated Intersection Control . . . . . . . . . . . . . . . . . . . . . . . . . . 9-21
Figure 9.12
Example of a Fully-Actuated Two-Phase Timing Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-25
10. TRAFFIC SIMULATION
Figure 10.1
Several Statistical Distributions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-7
Figure 10.2
Vehicle Positions During Lane-Change Maneuver. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-8
Figure 10.3
Structure Chart of Simulation Modules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-9
Figure 10.4
Comparison of Trajectories of Vehicles from Simulation Versus Field Data for Platoon 123. . . . . . . . . . . . 10-16
Figure 10.5
Graphical Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-18
Figure 10.6
Animation Snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-19
11. KINETIC THEORIES
Figure 11.1
Dependence of the mean speed upon density normalized to jam density. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-5
Figure 11.2
Evolution of the flow, according to a diffusively corrected Lighthill-Whitham model. . . . . . . . . . . . . . . . . . . . . . 11-8
xiii
List of Tables
3. HUMAN FACTORS
Table 3.1
Hooper-McGee Chaining Model of Perception-Response Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-4
Table 3.2
Brake PRT - Log Normal Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6
Table 3.3
Summary of PRT to Emergence of Barrier or Obstacle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6
Table 3.4
Percentile Estimates of PRT to an Unexpected Object . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7
Table 3.5
Movement Time Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
Table 3.6
Visual Acuity and Letter Sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
Table 3.7
Within Subject Variation for Sign Legibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
Table 3.8
Object Detection Visual Angles (Daytime) (Minutes of Arc) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
Table 3.9
Maneuver Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19
Table 3.10
Percentile Estimates of Steady State Unexpected Deceleration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
Table 3.11
Percentile Estimates of Steady State Expected Deceleration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
Table 3.12
Critical Gap Values for Unsignalized Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-25
Table 3.13
PRTs at Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-27
4. CAR FOLLOWING MODELS

Table 4.1
Results from Car-Following Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-25
Table 4.2
Comparison of the Maximum Correlations obtained for the Linear and Reciprocal Spacing Models
for the Fourteen Lincoln Tunnel Test Runs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-27
Table 4.3
Maximum Correlation Comparison for Nine Models, a , m the Fourteen Lincoln Tunnel Test Runs. . . . . . . . . 4-28
#
Table 4.4
Results from Car Following Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-30
Table 4.5
Macroscopic Flow Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-33
Table 4.6
Parameter Comparison (Holland Tunnel Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-35
5. CONTINUUM MODELS
Table 5.1
Oscillation Time and Magnitudes of Stop-and-go Traffic From German Measurement. . . . . . . . . . . . . . . . . . . 5-12
xiv
7. TRAFFIC IMPACT MODELS
Table 7.1
Federal Emission Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-14
Table 7.2
Standard Input Values for the CALINE4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-17
Table 7.3
Graphical Screening Test Results for Existing Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-19

Table 8.1
“A” Values for Equation 8.23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-8
Table 8.2
Evaluation of Conflicting Rank Volume qp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-34

Table 9.1
Maximum Relative Discrepancy between the Approximate Expressions and Ohno's Algorithm . . . . . . . . . . . 9-8
Table 9.2
Cycle Length Used For Delay Estimation for Fixed-Time and Actuated Signals Using Webster’s Formula . . 9-23
Table 9.3
Calibration Results of the Steady-State Overflow Delay Parameter ( k) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-26
10. TRAFFIC SIMULATION

Table 10.1
Classification of the TRAF Family of Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-4
Table 10.2
Executive Routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-9
Table 10.3
Routine MOTIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-10
Table 10.4
Routine CANLN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-11
Table 10. 5
Routine CHKLC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-12
Table 10.6
Routine SCORE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-13
Table 10.7
Routine LCHNG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-14
Table 10.8
Simulation Output Statistics: Measures of Effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-25
11. KINETIC THEORIES
Table 11.1
Status of various kinetic models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-6
xv

FOREWORD
This publication is an update and expansion of Additional acknowledgment is made to Alberto Santiago,
Transportation Research Board Special Report 165, Chief of State Programs at National Highway Institute of
"Traffic Flow Theory," published in 1975. This updating the FHWA (formerly with Intelligent Systems and
was undertaken on recommendation of the Transportation Technology Division), without whose initiative and
Research Board's Committee A3A11 on Traffic Flow support, this report simply would not have been possible.
Theory and Characteristics. The Federal Highway Thanks also go to Brenda Clark for the initial formatting
Administration (FHWA) funded a project to develop this of the report, Kathy Breeden for updating the graphics
report via an Interagency Agreement with the Department and text and coordinating the effort with the authors, Phil
of Energy's Oak Ridge National Laboratory (ORNL). Wolff for the creation and management of the report’s
The project was carried out by ORNL under supervision web-site, and to Elaine Thompson for her project
of an Advisory Committee that, in addition to the three management assistance.
co-editors, included the following prominent individuals:
Finally, we acknowledge the following individuals who
Richard Cunard, TRB Liaison Representative read and reviewed part or all of the manuscript and
Dr. Henry Lieu, Federal Highway Administration contributed valuable suggestions: Rahmi Akcelik, Rahim
Dr. Hani Mahmassani, University of Texas at Austin Benekohal, David Boyce, Micheal Brackstone, Werner
Brilon, Christine Buisson, Ennio Cascetta, Michael
While the general philosophy and organization of the Cassidy, Avishai Ceder, Arun Chatterjee, Ken Courage,
previous two reports have been retained, the text has Ray Derr, Mike Florian, Fred Hall, Benjamin Heydecker,
been completely rewritten and two new chapters have Ben Hurdle, Shinya Kikuchi, Helmut “Bill” Knee, Haris
been added. The primary reasons for doing such a major Koutsopoulos, Jean-Baptiste Lesort, John Leonard II,
revision were to bring the material up-to-date; to include Fred Mannering, William McShane, Kyriacos Mouskos,
new developments in traffic flow theory (e.g., network Panos Prevedourous, Vladimir Protopopescu, Bin Ran,
models); to ensure consistency among chapters and Tom Rockwell, Mitsuru Saito, and Natacha Thomas.
topics; and to emphasize the applications or practical
aspects of the theory. There are completely new chapters We believe that this present publication meets its
on human factors (Chapter 3) and network traffic models objective of synthesizing and reporting, in a single
(Chapter 5). document, the present state of knowledge or lack thereof
in traffic flow theory. It is sincerely hoped that this report
To ensure the highest degree of reliability, accuracy, and will be useful to the graduate students, researchers and
quality in the content of this report, the collaboration of a practitioners, and others in the transportation profession.
large number of experts was enlisted, and this report
presents their cooperative efforts. We believe that a Editors: Dr. Nathan Gartner
serious effort has been made by the contributing authors University of Massachusetts - Lowell
in this report to present theory and information that will
have lasting value. Our appreciation is extended to the Dr. Carroll J. Messer
many authors for their commendable efforts in writing Texas A&M University
this update, willingly sharing their valuable time,
knowledge, and their cooperative efforts throughout the Dr. Ajay K. Rathi
project. Oak Ridge National Laboratory
Project Leader.
We would also like to acknowledge the time spent by the
members of the Advisory Committee in providing
guidance and direction on the style of the report and
for their reviews of the many drafts of the report.
Foreword - 1

INTRODUCTION
BY NATHAN H. GARTNER1
CARROLL MESSER2
AJAY K. RATHI3
1
Professor, Department of Civil Engineering, University of Massachusetts at Lowell, 1 University Avenue, Lowell, MA 01854.
2
Professor, Department of Civil Engineering, Texas A&M University, TTI Civil Engineering Building, Suite 304C, College Station,
TX 77843-3135.
3
Senior R&D Program Manager and Group Leader, ITS Research, Center for Transportation Analysis, Oak Ridge National Laboratory,
P.O. Box 2008, Oak Ridge, TN 37831-6207.
1. INTRODUCTION
1.
INTRODUCTION
It is hardly necessary to emphasize the importance of By 1959 traffic flow theory had developed to the point where it
transportation in our lives. In the United States, we spend about appeared desirable to hold an international symposium. The
20 percent of Gross National Product (GNP) on transportation, First International Symposium on The Theory of Traffic Flow
of which about 85 percent is spent on highway transportation was held at the General Motors Research Laboratories in
(passenger and freight). We own and operate 150 million Warren, Michigan in December 1959 (Herman 1961). This was
automobiles and an additional 50 million trucks, bringing car the first of what has become a series of triennial symposia on
ownership to 56 per hundred population (highest in the world). The Theory of Traffic flow and Transportation. The most recent
These vehicles are driven an average of 10,000 miles per year in this series, the 12th symposium was held in Berkeley,
for passenger cars and 50,000 miles per year for trucks on a California in 1993 (Daganzo 1993). A glance through the
highway system that comprises more than 4 million miles. The proceedings of these symposia will provide the reader with a
indices in other countries may be somewhat different, but the good indication of the tremendous developments in the
importance of the transportation system, and especially the understanding and the treatment of traffic flow processes in the
highway component of it, is just the same or even greater. While past 40 years. Since that time numerous other symposia and
car ownership in some countries may be lower, the available specialty conferences are being held on a regular basis dealing
highway network is also smaller leading to similar or more with a variety of traffic related topics. The field of traffic flow
severe congestion problems. theory and transportation has become too diffuse to be covered
by any single type of meeting. Yet, the fundamentals of traffic
Traffic flow theories seek to describe in a precise mathematical flow theory, while better understood and more easily
way the interactions between the vehicles and their operators characterized through advanced computation technology, are just
(the mobile components) and the infrastructure (the immobile as important today as they were in the early days. They form the
component). The latter consists of the highway system and all its foundation for all the theories, techniques and procedures that
operational elements: control devices, signage, markings, etc. are being applied in the design, operation, and development of
As such, these theories are an indispensable construct for all advanced transportation systems.
models and tools that are being used in the design and operation
of streets and highways. The scientific study of traffic flow had It is the objective of this monograph to provide an updated
its beginnings in the 1930’s with the application of probability survey of the most important models and theories that
theory to the description of road traffic (Adams 1936) and the characterize the flow of highway traffic in its many facets. This
pioneering studies conducted by Bruce D. Greenshields at the monograph follows in the tracks of two previous works that were
Yale Bureau of Highway Traffic; the study of models relating sponsored by the Committee on Theory of Traffic Flow of the
volume and speed (Greenshields 1935) and the investigation of Transportation Research Board (TRB) and its predecessor the
performance of traffic at intersections (Greenshields 1947). Highway Research Board (HRB). The first monograph, which
After WWII, with the tremendous increase in use of automobiles was published as HRB Special Report 79 in 1964, consisted of
and the expansion of the highway system, there was also a surge selected chapters in the then fledgling Traffic Science each of
in the study of traffic characteristics and the development of which was written by a different author (Gerlough and Capelle
traffic flow theories. The 1950’s saw theoretical developments 1964). The contents included:
based on a variety of approaches, such as car-following, traffic
wave theory (hydrodynamic analogy) and queuing theory. Some Chapter 1. Part I: Hydrodynamic Approaches, by L. A. Pipes.
of the seminal works of that period include the works by Part II: On Kinematic Waves; A Theory of Traffic Flow on Long
Reuschel (1950a; 1950b; 1950c), Wardrop (1952), Pipes Crowded Roads, by M. J. Lighthill and G. B. Whitham.
(1953), Lighthill and Whitham (1955), Newell (1955), Webster
(1957), Edie and Foote (1958), Chandler et al. (1958) and other Chapter 2. Car Following and Acceleration Noise, by E. W.
papers by Herman et al. (see Herman 1992). Montroll and R. B. Potts.
1-1
1. INTRODUCTION
Chapter 3. Queuing Theory Approaches, by D. E. Cleveland their entirety and include the latest research and information in
and D. G. Capelle. their respective areas. Chapter 2 presents the various models
that have been developed to characterize the relationships among
Chapter 4. Simulation of Traffic Flow, by D. L. Gerlough. the traffic stream variables: speed, flow, and concentration.
Most of the relationships are concerned with uninterrupted traffic
Chapter 5. Some Experiments and Applications, by R. S. Foote. flow, primarily on freeways or expressways. The chapter
stresses the link between theory and measurement capability,
A complete rewriting of the monograph was done by Gerlough since to as large extent development of the first depends on the
and Huber (1975) and was published as TRB Special Report latter.
165 in 1975. It consisted of nine chapters, as follows:
Chapter 3, on Human Factors, discusses salient performance
Chapter 1. Introduction. aspects of the human element in the context of the person-
machine system, i.e. the motor vehicle. The chapter describes
Chapter 2. Measurement of Flow, Speed, and Concentration. first discrete components of performance, including: perception-
reaction time, control movement time, responses to traffic
Chapter 3. Statistical Distributions of Traffic Characteristics. control devices, to the movement of other vehicles, to hazards in
the roadway, and how different segments of the population differ
Chapter 4. Traffic Stream Models. in performance. Next, the kind of control performance that
underlies steering, braking, and speed control -- the primary
Chapter 5. Driver Information Processing Characteristics. control functions -- is described. Applications of open-loop and
closed-loop vehicle control to specific maneuvers such as lane
Chapter 6. Car Following and Acceleration Noise. keeping, car following, overtaking, gap acceptance, lane
closures, and sight distances are also described. To round out
Chapter 7. Hydrodynamic and Kinematic Models of Traffic. the chapter, a few other performance aspects of the driver-
vehicle system are covered, such as speed limit changes,
Chapter 8. Queuing Models (including Delays at distractions on the highway, and responses to real-time driver
Intersections). information. The most obvious application of human factors is
in the development of Car Following Models (Chapter 4). Car
Chapter 9. Simulation of Traffic Flow. following models examine the manner in which individual
vehicles (and their drivers) follow one another. In general, they
This volume is now out of print and in 1987 the TRB Committee are developed from a stimulus-response relationship, where the
on Traffic Flow Theory and Characteristics recommended that response of successive drivers in the traffic stream is to
a new monograph be prepared as a joint effort of committee accelerate or decelerate in proportion to the magnitude of the
members and other authors. While many of the basic theories stimulus at time t after a time lag T. Car following models form
may not have changed much, it was felt that there were a bridge between the microscopic behavior of individual vehicles
significant developments to merit writing of a new version of the and the macroscopic characteristics of a single-lane traffic
monograph. The committee prepared a new outline which stream with its corresponding flow and stability properties.
formed the basis for this monograph. After the outline was
agreed upon, the Federal Highway Administration supported this Chapter 5 deals with Continuous Flow Models. Because traffic
effort through an interagency agreement with the Oak Ridge involves flows, concentrations, and speeds, there is a natural
National Laboratory. An Editorial Committee was appointed, tendency to attempt to describe traffic in terms of fluid behavior.
consisting of N. H. Gartner, C. J. Messer, and A. K. Rathi, which Car following models recognize that traffic is made up of
was charged with the editorial duties of the preparation of the discrete particles and it is the interactions between these
manuscripts for the different chapters. particles that have been developed for fluids (i.e., continuum
models) is concerned more with the over all statistical behavior
The first five chapters follow similarly titled chapters in the of the traffic stream rather than with the interactions between the
previous monograph; however, they all have been rewritten in particles. In the fluid flow analogy, the traffic stream is treated
1-2
1. INTRODUCTION
as a one dimensional compressible fluid. This leads to two basic intersections among two or more streams and provides
assumptions: (i) traffic flow is conserved, which leads to the calculations of capacities and quality of traffic operations based
conservation or continuity equation, and (ii) there is a one-to-one on queuing modeling.
relationship between speed and density or between flow and
density. The simple continuum model consists of the Traffic Flow at Signalized Intersections is discussed in Chapter
conservation equation and the equation of state (speed-density or 9. The statistical theory of traffic flow is presented, in order to
flow-density relationship). If these equations are solved together provide estimates of delays and queues at isolated intersections,
with the basic traffic flow equation (flow equals density times including the effect of upstream traffic signals. This leads to the
speed) we can obtain the speed, flow and density at any time and discussion of traffic bunching, dispersion and coordination at
point of the roadway. By knowing these basic traffic flow traffic signals. The fluid (shock-wave) approach to traffic
variables, we know the state of the traffic system and can derive signal analysis is not covered in this chapter; it is treated to
measures of effectiveness, such as delays, stops, travel time, total some extent in Chapter 5. Both pretimed and actuated signal
travel and other measures that allow the analysts to evaluate how control theory are presented in some detail. Further, delay
well the traffic system is performing. In this chapter, both models that are founded on steady-state queue theory as well as
simple and high order models are presented along with analytical those using the so-called coordinate transform method are
and numerical methods for their implementation. covered. Adaptive signal control is discussed only in a
qualitative manner since this topic pertains primarily to the
Chapter 6, on Macroscopic Flow Models, discards the development of optimal signal control strategies, which is
microscopic view of traffic in terms of individual vehicles or outside the scope of this chapter.
individual system components (such as links or intersections)
and adopts instead a macroscopic view of traffic in a network. The last chapter, Chapter 10, is on Traffic Simulation.
A variety of models are presented together with empirical Simulation modeling is an increasingly popular and effective tool
evidence of their applicability. Variables that are being for analyzing a wide variety of dynamical problems which are
considered, for example, include the traffic intensity (the distance not amenable to study by other means. These problems are
travelled per unit area), the road density (the length or area of usually associated with complex processes which can not readily
roads per unit area of city), and the weighted space mean speed. be described in analytical terms. To provide an adequate test
The development of such models extends traffic flow theory into bed, the simulation model must reflect with fidelity the actual
the network level and can provide traffic engineers with the traffic flow process. This chapter describes the traffic models
means to evaluate system-wide control strategies in urban areas. that are embedded in simulation packages and the procedures
Furthermore, the quality of service provided to motorists can be that are being used for conducting simulation experiments.
monitored to assess the city's ability to manage growth. Network
performance models could also be used to compare traffic Consideration was also given to the addition of a new chapter on
conditions among different cities in order to determine the Traffic Assignment Models. Traffic assignment is the process
allocation of resources for transportation system improvements. of predicting how a given set of origin-destination (OD) trip
Chapter 7 addresses Traffic Impact Models, specifically, the demands will manifest themselves onto a transportation network
following models are being discussed: Traffic and Safety, Fuel of links and nodes in terms of flows and queues. It has major
Consumption Models, and Air Quality Models. applications in both transportation planning models and in
dynamic traffic management models which are the bedrock of
Chapter 8 is on Unsignalized Intersection Theory. Unsignalized Intelligent Transportation Systems (ITS). Generally, the
intersections give no positive indication or control to the driver. assignment process consists of a macroscopic simulation of the
The driver alone must decide when it is safe to enter the behavior of travelers in a network of transportation facilities. At
intersection, typically, he looks for a safe opportunity or "gap" in the same time it reflects the interconnection between the
the conflicting traffic. This model of driver behavior is called microscopic models of traffic behavior that are discussed in this
"gap acceptance." At unsignalized intersections the driver must monograph and the overall distribution of traffic demands
also respect the priority of other drivers. This chapter discusses throughout the network. This is expressed by link cost functions
in detail the gap acceptance theory and the headway distributions that serve as a basis for any assignment or route choice process.
used in gap acceptance calculations. It also discusses the After much deliberation by the editorial and advisory committees
1-3
1. INTRODUCTION
it was decided that the subject cannot be presented adequately in and Iida (1997). This is a lively research area and new
a short chapter within this monograph. It would be better served publications abound.
by a dedicated monograph of its own, or by reference to the
extensive literature in this area. Early references include the Research and developments in transportation systems and,
seminal works of Wardrop (1952), and Beckmann, McGuire and concomitantly, in the theories that accompany them proceed at
Winsten (1956). Later publications include books by Potts and a furious pace. Undoubtedly, by the time this monograph is
Oliver (1972), Florian (1976), Newell (1980), and Sheffi printed, distributed, and read, numerous new developments will
(1985). Recent publications, which reflect modern approaches have occurred. Nevertheless, the fundamental theories will not
to equilibrium assignment and to dynamic traffic assignment, have changed and we trust that this work will provide a useful
include books by Patriksson (1994), Ran and Boyce (1994), source of information for both newcomers to the field and
Gartner and Improta (1995), Florian and Hearn (1995), and Bell experienced workers.
References
Adams, W. F. (1936). Road Traffic Considered as a Random Greenshields, B. D. (1935). A Study in Highway Capacity.
Series, J. Inst. Civil Engineers, 4, pp. 121-130, U.K. Highway Research Board, Proceedings, Vol. 14, p. 458.
Beckmann, M., C.B. McGuire and C.B. Winsten (1956). Greenshields, B. D., D. Schapiro, and E. L. Erickson, (1947).
Studies in the Economics of Transportation. Yale Traffic Performance at Urban Intersections. Bureau of
University Press, New Haven. Highway Traffic, Technical Report No. 1. Yale University
Bell, M.G.H. and Y. Iida (1997). Transportation Network Press, New Haven, CT.
Analysis. John Wiley & Sons. Herman, R., Editor (1961). Theory of Traffic Flow. Elsevier
Chandler, R. E., R. Herman, and E. W. Montroll, (1958). Science Publishers.
Traffic Dynamics: Studies in Car Following, Opns. Herman, R., (1992). Technology, Human Interaction, and
Res. 6, pp. 165-183. Complexity: Reflections on Vehicular Traffic Science.
Daganzo, C. F., Editor (1993). Transportation and Traffic Operations Research, 40(2), pp. 199-212.
Theory. Proceedings, 12th Intl. Symposium. Elsevier Lighthill, M. J. and G. B. Whitham, (1955). On Kinematic
Science Publishers. Waves: II. A Theory of Traffic Flow on Long Crowded
Edie, L. C. and R. S. Foote, (1958). Traffic Flow in Tunnels, Roads. Proceedings of the Royal Society: A229, pp. 317-
Proc. Highway Research Board, 37, pp. 334-344. 347, London.
Florian, M.A., Editor (1976). Traffic Equilibrium Methods. Newell, G. F. (1955). Mathematical Models for Freely
Lecture Notes in Economics and Mathematical Systems, Flowing Highway Traffic. Operations Research 3,
Springer-Verlag. pp. 176-186.
Florian, M. and D. Hearn (1995). Network Equilibrium Newell, G. F. (1980). Traffic Flow on Transportation
Models and Algorithms. Chapter 6 in Network Routing Networks. The MIT Press, Cambridge, Massachusetts.
(M.O. Ball et al., Editors), Handbooks in OR & MS, Vol. Patriksson, M. (1994). The Traffic Assignment Problem;
8, Elsevier Science. Models and Methods. VSP BV, Utrecht, The Netherlands.
Gartner, N.H. and G. Improta, Editors (1995). Urban Traffic Pipes, L. A. (1953). An Operational Analysis of Traffic
Networks; Dynamic Flow Modeling and Control. Dynamics. J. Appl. Phys., 24(3), pp. 274-281.
Springer-Verlag. Potts, R.B. and R.M. Oliver (1972). Flows in Transportation
Gerlough, D. L. and D. G. Capelle, Editors (1964). An Networks. Academic Press.
Introduction to Traffic Flow Theory. Special Report 79. Ran, B. and D. E. Boyce (1994). Dynamic Urban
Highway Research Board, Washington, D.C. Transportation Network Models; Theory and Implications
Gerlough, D. L. and M. J. Huber, (1975). Traffic Flow Theory for Intelligent Vehicle-Highway Systems. Lecture Notes in
- A Monograph. Special Report 165, Transportation Economics and Mathematical Systems, Springer-Verlag.
Research Board.
1-4
1. INTRODUCTION
Reuschel, A. (1950a). Fahrzeugbewegungen in der Kolonne. Wardrop, J. G. (1952). Some Theoretical Aspects of Road
Oesterreichisches Ingenieur-Aarchiv 4, No. 3/4, Traffic Research. Proceedings of the Institution of Civil
pp. 193-215. Engineers, Part II, 1(2), pp. 325-362, U.K.
Reuschel, A. (1950b and 1950c). Fahrzeugbewegungen in der Webster, F. V. (1958). Traffic Signal Settings. Road Research
Kolonne bei gleichformig beschleunigtem oder verzogertem Technical Paper No. 39. Road Research Laboratory,
Leitfahrzeug. Zeitschrift des Oesterreichischen Ingenieur- London, U.K.
und Architekten- Vereines 95, No. 7/8 59-62, No. 9/10 pp.
73-77.
Sheffi, Y. (1985). Urban Transportation Networks;
Equilibrium Analysis with Mathematical Programming
Methods. Prentice-Hall.
1-5
TRAFFIC STREAM CHARACTERISTICS
BY FRED L. HALL4
4
Professor, McMaster University, Department of Civil Engineering and Department of Geography, 1280 Main Street West,
Hamilton, Ontario, Canada L8S 4L7.
CHAPTER 2 - Frequently used Symbols
k density of a traffic stream in a specified length of road
L length of vehicles of uniform length
ck constant of proportionality between occupancy and

density, under certain simplifying assumptions
ki the (average) density of vehicles in substream I
qi the average rate of flow of vehicles in substream I
Å average speed of a set of vehicles
A A(x,t) the cumulative vehicle arrival function over

space and time
kj jam density, i.e. the density when traffic is so heavy that

it is at a complete standstill
uf free-flow speed, i.e. the speed when there are no

constraints placed on a driver by other vehicles on the
road
Fourth draft of revised chapter 2

2001 October 14
Chapter 2: Traffic Stream Characteristics
Author’s note: This material has benefited greatly from the assistance of Michael Cassidy, of
the University of California at Berkeley. He declined the offer to be listed as a co-author of the
chapter, although that would certainly have been warranted. With his permission, and that of
the publisher, several segments of this material have been reproduced directly from his chapter,
“Traffic Flow and Capacity”, which appears as Chapter 6 in Handbook of transportation
science, edited by Randolph W. Hall, and published by Kluwer Academic Publishers in 1999.
That material appears in italics below, in section 2.1. The numbering of Figures and equations
has been altered from his numbers to correspond to numbering within this chapter.
In this chapter, properties that describe highway traffic are introduced, such as flow, density, and
average vehicle speed; issues surrounding their measurements are discussed; and various models
that have been proposed for describing relationships among these properties are presented. Most
of the work dealing with these relationships has been concerned with uninterrupted traffic flow,
primarily on freeways or expressways, but the general relationships will apply to all kinds of
traffic flow.
This chapter starts with a section on definitions of key variables and terms. Because of the
importance of measurement capability to theory development, the second section deals with
measurement issues, including historical developments in measurement procedures, and criteria
for selecting good measurement characteristics. The third section discusses a number of the
bivariate models that have been proposed in the past to relate key variables. That is followed by
a short section on attempts to deal simultaneously with the three key variables. The final,
summary section provides links to a number of the other chapters in this monograph.
2.1 Definitions and terms
This section provides definitions of some properties commonly used to characterize traffic
streams, together with some generalized definitions that preserve useful relations among these
properties. Before turning to the definitions, however, a useful graphical tool is introduced,
namely trajectories plotted on a time-space diagram. This is followed by definitions of traffic
stream properties, time-mean and space-mean properties, and the generalized definitions.
Following that is a discussion of the relationship between density, which is often used in the
generalized definitions, and occupancy, which is measured by many freeway systems. The final
topic covered is three-dimensional representations of traffic streams. The discussion in this
section follows very closely or repeats some of that in the chapter by Cassidy (1999), which in
turn credits notes composed by Newell (unpublished) and a textbook written by Daganzo (1997).
2-1
2.1.1 The Time-Space Diagram.
The Time-Space Diagram. Objects are commonly constrained to move along a one-
dimensional guideway, be it, for example, a highway lane, walkway, conveyor belt, charted course
or flight path. Thus, the relevant aspects of their motion can often be described in cartesian
coordinates of time, t, and space, x. Figure 2-1 illustrates the trajectories of some objects traversing
a facility of length L during time interval T; these objects may be vehicles, pedestrians or cargo.
Each trajectory is assigned an integer label in the ascending order that the object would be seen by
a stationary observer. If one object overtakes another, their trajectories may exchange labels, as
shown for the fourth and fifth trajectories in the figure. Thus, the "th trajectory describes the
location of a reference point (e.g. the front end) of object " as a function of time t, x" (t).
Figure 2-1. Time-space diagram.
The characteristic geometries of trajectories on a time-space diagram describe the motion of

objects in detail. These diagrams thus offer the most complete way of displaying the observations
that may have actually been measured along a facility. As a practical matter, however, one is not
likely to collect all the data needed to construct trajectories. Rather, time-space diagrams derive
their (considerable) value by providing a means to highlight the key features of a traffic stream using
only coarsely approximated data or hypothetical data from “thought experiments.”
2.1.2 Definitions of Some Traffic Stream Properties. It is evident from Figure 2-1 that the
slope of the "th trajectory is object "’s instantaneous velocity, v" (t), i.e.,
v (t) { dx (t)/dt , (2.1)
and that the curvature is its acceleration. Further, there exist observable properties of a traffic
stream that relate to the times that objects pass a fixed location, such as location x1, for example.
2-2
These properties are described with trajectories that cross a horizontal line drawn through the time-
space diagram at x1.
Referring to Figure 2-1, the headway of some ith object at x1, hi(x1), is the difference between the
arrival times of i and i-1 at x1, i.e.,
hi(x1 ) { ti(x1 ) - ti-1(x1 ) (2.2)
Flow at x1 is m, the number of objects passing x1, divided by the observation interval T,
q(T, x1 ) { m/T. (2.3)
For observation intervals containing large m,

m
¦ hi ( x1 ) | T
i 1 (2.4)
and thus,
1 1
q(T , x1 ) | ,
1 m h ( x1 ) (2.5)
¦ hi ( x1 )
mi 1
i.e., flow is the reciprocal of the average headway.
Analogously, some properties relate to the locations of objects at a fixed time, as observed, for
example, from an aerial photograph. These properties may be described with trajectories that cross
a vertical line in the t-x plane. For example, the spacing of object j at some time t1, sj(t1 ), is the
distance separating j from the next downstream object; i.e.,
sj(t1 ) { xj-1(t1 ) - xj(t1 ). (2.6)
Density at instant t1 is n, the number of objects on a facility at that time, divided by L, the facility’s
physical length; i.e.,
k(L, t1 ) { n /L. (2.7)
If the L contains large n,

n
¦ s j (t1 ) | L (2.8)
j 1
and
1 1
k ( L, t1 ) | , (2.9)
1 n
s (t1 )
¦ s j (t1 )
nj1
giving a relation between density and the average spacing parallel to that of flow and the average
headway.
2-3
2.1.3 Time-Mean and Space-Mean Properties. For an object’s attribute D, where D might be
its velocity, physical length, number of occupants, etc., one can define an average of the m
objects passing some fixed location x1 over observation interval T,
1 m
D(T , x1 ) ¦ D i ( x1 ),
mi 1 (2.10)
i.e., a time-mean of attribute D. If D is headway, for example, D(T, x1) is the average headway or the
reciprocal of the flow.
Conversely, the space-mean of attribute D at some time t1, D(L, t1), is obtained from the observations
taken at that time over a segment of length L, i.e.,
1 n
D( L, t1 ) ¦ D j (t1 ).
nj1 (2.11)
If, for example, D is spacing, D(L, t1) is the average spacing or the reciprocal of the density.
For any attribute D, there is no obvious relation between its time and space means. The reader may
confirm this (using the example of D as velocity) by envisioning a rectangular time-space region L u
T traversed by vehicles of two classes, fast and slow, which do not interact. For each class, the
trajectories are parallel, equidistant and of constant slope; such conditions are said to be stationary.
The fraction of fast vehicles distributed over L as seen on an aerial photograph taken at some instant
within T will be smaller than the fraction of fast vehicles crossing some fixed point along L during
the interval T. This is because the fast vehicles spend less time in the region than do the slow ones.
Analogously, one might envision a closed loop track and note that a fast vehicle passes a stationary
observer more often than does a slow one.
2.1.4 Generalized Definitions of Traffic Stream Properties. To describe a traffic stream, one
usually seeks to measure properties that are not sensitive to the variations in the individual objects
(e.g. the vehicles or their operators) without averaging-out features of interest. This is the trade-off
inherent in choosing between short and long measurement intervals, as previously noted. It was
partly to address this trade-off that Edie (1965, 1974) proposed some generalized definitions of flow
and density that averaged these properties in the manner described below.
To begin this discussion, the thin, horizontal rectangle in Figure 2-2 corresponds to a fixed
observation point. As per its conventional definition provided earlier, the flow at this point is m / T,
where m = 4 in the figure. Since this point in space is a region of temporal duration T and elemental
m dx
spatial dimension dx, the flow can be expressed equivalently as . The denominator is the
T dx
euclidean area of the thin horizontal rectangle, expressed in units of distance u time. The numerator
2-4
Figure 2-2. Trajectories in time-space region.
is the total distance traveled by all objects in this thin region, since objects cannot enter or exit the
region via its elementally small left and right sides.
That flow, then, is the ratio of the distance traveled in a region to the region’s area is valid for any
time-space region, since all regions are composed of elementary rectangles. Taking, for example,
region A in Figure 2-2, Edie’s generalized definition of the flow in A, q(A), is d(A) / |A|, where d(A)
is the total distance traveled in A and |A| is used to denote the region’s area.
As the analogue to this, the thin, vertical rectangle in Figure 2-2 corresponds to an instant in time.
As per its conventional definition, density is n / L (where n = 2 in this figure) and this can be
n dt
expressed equivalently as . It follows that Edie’s generalized definition of density in a region A,
L dt
k(A), is t(A) / |A|, where t(A) is the total time spent in A.
It should be clear that these generalized definitions merely average the flows collected over all
points, and the densities collected at each instant, within the region of interest. Dividing this flow by
this density gives d(A) / t(A), which can be taken as the average velocity of objects in A, v(A). The
reader will note that, with Edie’s definitions, the average velocity is the ratio of flow to density.
Traffic measurement devices, such as loop detectors installed beneath the road surface, can be used
to measure flows, densities and average vehicle velocities in ways that are consistent with these
generalized definitions. Discussion of this is offered in Cassidy and Coifman (1997).
As a final note regarding v(A), when A is taken as a thin horizontal rectangle of spatial dimension
dx, the time spent in the region by object i is dx / vi, where vi is i’s velocity. Thus, for this thin,
2-5
m 1
horizontal region A, t(A) = dx ¦ . Given that for the same region, d(A) = m·dx, the generalized
v
i 1 i
mean velocity becomes

d ( A) 1 (2.12)
v( A) ,
t ( A) 1 m 1
¦
m i 1 vi
i.e., the reciprocal of the mean of the reciprocal velocities, or the harmonic mean velocity. The 1 / vi
1
is often referred to as the pace of i, pi, and thus
ª1 m º
v( A) « m i¦1 pi » .
¬ ¼ (2.13)
Eq. 2.15 applies for regions with L > dx provided that all i span the L and that each pi (or vi) is i’s
average over the L.
It follows that when conditions in a region A are stationary, the harmonic mean of the velocities
measured at a fixed point in A is the v(A). By the same token, the v(A) is the space-mean velocity
measured at any instant in A (provided, again, that conditions are stationary).
2.1.5 The Relation Between Density and Occupancy. Occupancy is conventionally defined as
the percentage of time that vehicles spend atop a loop detector. It is a commonly-used property for
describing highway traffic streams; it is used later in this chapter, for example, for diagnosing
freeway traffic conditions. In particular, occupancy is a proxy for density. The following discussion
demonstrates that the former is merely a dimensionless version of the latter.
One can readily demonstrate this relation by adopting a generalized definition of occupancy
analogous to the definitions proposed by Edie. Such a definition is made evident by illustrating each
trajectory with two parallel lines tracing the vehicle’s front and rear (as seen by a detector) and this
is exemplified in Figure 2-3. The (generalized) occupancy in the region A, U(A), can be taken as the
fraction of the region’s area covered by the shaded strips in the figure. From this, it follows that the
U(A) and the k(A) are related by an average of the vehicle lengths. This average vehicle length is, by
definition, the area of the shaded strips within A divided by the t(A); i.e., it is the ratio of the U(A) to
the k(A),
U( A) area of the shaded strips | A |

average vehicle length . (2.14)
k ( A) | A| t ( A)
2-6
Figure 2-3. Trajectories of vehicle fronts and rears.

Notably, an average of the vehicle lengths also relates the k(A) to U, where the latter is the
occupancy as conventionally defined (i.e., the percentage of time vehicles spend atop the
detector). Toward illustrating this relation, the L in Figure 6-4 is assumed to be the length of
road “visible” to the loop detector, the so-called detection zone. The T is some interval of time;
e.g. the interval over which the detector collects measurements. The time each ith vehicle spends
atop the detector is denoted as Wi. Thus, if m vehicles pass the detector during time T, the
m
¦ Wi
U i 1
.
T
As shown in Figure 6-4, L i is the summed length of the detection zone and the length of vehicle i.
Therefore,
m m m
1
¦W i
i 1
¦ L i
i 1 vi
¦L
i 1
i pi (2.15)
if the front end of each i has a constant vi over the distance L i. Since
m 1 m
¦ Wi ¦ Wi
m i1 1 m (2.16)
i 1
q ( A) ¦ Wi ,
T 1 m i1
T
m
2-7
it follows that
1 m
U q ( A) ¦ L i pi ,
m i1
1 ª 1 m º
U q ( A) « v ( A) ¦ L i pi »,
v ( A) ¬ m i1 ¼
ª m º
« ¦ L i pi »
U k ( A) « i 1
m
»,
« pi »
«¬ ¦ i 1
»¼ (2.17)
where the term in brackets is the average vehicle length relating U to the k(A); it is the so-
called average effective vehicle length weighted by the paces. If pace and vehicle length are
uncorrelated, the term in brackets in (2.17) can be approximated by the unweighted average
of the vehicle lengths in the interval T.
When measurements are taken by two closely spaced detectors, a so-called speed trap, the pi
are computed from each vehicle’s arrival times at the two detectors. The Li are thus
computed by assuming that the pi are constant over the length of the speed trap. When only a
single loop detector is available, vehicle velocities are often estimated by using an assumed
average value of the (effective) vehicle lengths.
2.1.6 Three-Dimensional Representation of Vehicle Streams. It is useful to display flows

and densities using a three-dimensional representation described by Makagami et al. (1971).
For this representation, an axis for the cumulative number of objects, N, is added to the t-x
coordinate system so that the resulting surface N(t, x) is like a staircase with each trajectory
being the edge of a step. As shown in Figure 2-4, curves of cumulative count versus time are
obtained by taking cross-sections of this surface at some fixed locations and viewing the
exposed regions in the t-N plane. Analogously, cross-sections at fixed times viewed in the N-x
plane reveal curves of cumulative count versus space.
Figure 2-4 shows cumulative curves at two locations and for two instants in time. The former
display the trip times of objects and the time-varying accumulations between the two
locations, as labeled on the figure. These cumulative curves can be transformed into a
queueing diagram (as described in Chapter 5) by translating the curve at upstream x1
forward by the free-flow (i.e., the undelayed) trip time from x1 to x2. Also displayed in Figure
2-4, the curves of cumulative count versus space show the number of objects crossing a fixed
location during the interval t2 - t1 and the distances traveled by individual objects during this
same interval.
If one is dealing with many objects so that measuring the exact integer numbers is not
important, it is advantageous to construct the cumulative curves with piece-wise linear
approximations; e.g. the curves may be smoothed using linear interpolations that pass
through the crests of the steps. The time-dependent flows past some location are the slopes of
the smoothed curve of t versus N
2-8
constructed at that location (Moskowitz, 1954; Edie and Foote, 1960; Newell, 1971, 1982).
Analogously, the location-dependent densities at some instant are the negative slopes of a smoothed
curve of N versus x; densities are the negative slopes because objects are numbered in the reverse
direction to their motion. This three-dimensional model has been applied by Newell (1993) in work
on kinematic waves in traffic (see Chapter 5). In addition, Part I of his paper contains some
historical notes on the use of this approach for modelling.
Figure 2-4. Three-dimensional representation.
2.2 Measurement issues
This section addresses issues related to the measurement of the key variables defined in the
previous section. Four topics are covered. First, there is a discussion of measurement
procedures that have traditionally been used. Second, potential measurement errors arising from
the mismatch between the definitions and the measurement methods are discussed. Third,
measurement difficulties that can potentially arise from the particular location used for collecting
measurements are considered. The final topic is the nature of the time intervals from which to
collect data, drawing on the cumulative curves just presented.
2.2.1 Measurement procedures Measurement technology for obtaining traffic data has
2-9
changed over the 60-year span of interest in traffic flow, but most of the basic procedures remain
largely the same. Five measurement procedures are discussed in this section:
x measurement at a point;
x measurement over a short section (by which is meant less than about 10 meters (m);
x measurement over a length of road [usually at least 0.5 kilometers (km)];
x the use of an observer moving in the traffic stream; and
x wide-area samples obtained simultaneously from a number of vehicles, as part of Intelligent
Transportation Systems (ITS).
The first two were discussed with reference to the time-space diagram in Fig 2.2, and the
remaining three are also readily interpreted on that diagram. Details on each of these methods
can be found in the ITE's Manual of Transportation Engineering Studies (Robertson, 1994). The
wide-area samples from ITS are similar to having a number of moving observers at various
points and times within the system. New developments such as this will undoubtedly change the
way some traffic measurements are obtained in the future, but they have not been in operation
long enough to have a major effect on the material to be covered in this chapter.
Measurement at a point by hand tallies or pneumatic tubes, was the first procedure used for
traffic data collection. This method is easily capable of providing volume counts and therefore
flow rates directly, and can provide headways if arrival times are recorded. The technology for
making measurements at a point on freeways changed over 30 years ago from using pneumatic
tubes placed across the roadway to using point detectors (May et al. 1963; Athol 1965). The
most commonly used point detectors are based on inductive loop technology, but other methods
in use include microwave, radar, photocells, ultrasonics, and closed circuit television cameras.
Except for the case of a stopped vehicle, speeds at a 'point' can be obtained only by radar or
microwave detectors: dx/dt obviously requires some dx, however small. (Radar and microwave
frequencies of operation mean that a vehicle needs to move only about one centimeter during the
speed measurement.) In the absence of such instruments for a moving vehicle, a second
observation location is necessary to obtain speeds, which moves the discussion to that of
measurements over a short section.
Measurements over a short section Early studies used a second pneumatic tube, placed very
close to the first, to obtain speeds. More recent systems have used paired presence detectors,
such as inductive loops spaced perhaps five to six meters apart. With video camera technology,
two detector 'lines' placed close together provide the same capability for measuring speeds. All
of these presence detectors continue to provide direct measurement of volume and of time
headways, as well as of speed when pairs of them are used.
Most point detectors currently used, such as inductive loops or microwave beams, take up space
on the road, and are therefore a short section measurement. These detectors also measure
occupancy, which was not available from earlier technology. This variable is available because
the loop gives a continuous reading (at 50 or 60 Hz usually), which pneumatic tubes or manual
2-10
counts could not do. Because occupancy depends on the size of the detection zone of the
instrument, the measured occupancy may differ from site to site for identical traffic, depending
on the nature and construction of the detector.
Measurements along a length of road come either from aerial photography, or from cameras
mounted on tall buildings or poles. On the basis of a single frame from such sources, only
density can be measured. The single frame gives no sense of time, so neither volumes nor speed
can be measured. Once several frames are available, as from a video-camera or from time-lapse
photography over short time intervals, speeds and volumes can also be measured, as per the
generalized definitions provided above.
Despite considerable improvements in technology, and the presence of closed circuit television
on many freeways, there is very little use of measurements taken over a long section at the
present time. The one advantage such measurements might provide would be to yield true
journey times over a lengthy section of road, but that would require better computer vision
algorithms (to match vehicles at both ends of the section) than are currently possible. There
have been some efforts toward the objective of collecting journey time data on the basis of the
details of the 'signature' of particular vehicles or platoons of vehicles across a series of loops
over an extended distance (Kühne and Immes 1993), but few practical implementations as yet.
The moving observer method was used in some early studies, but is not used as the primary data
collection technique now because of the prevalence of the other technologies. There are two
approaches to the moving observer method. The first is a simple floating car procedure in which
speeds and travel times are recorded as a function of time and location along the road. While the
intention in this method is that the floating car behaves as an average vehicle within the traffic
stream, the method cannot give precise average speed data. It is, however, effective for
obtaining qualitative information about freeway operations without the need for elaborate
equipment or procedures. One form of this approach uses a second person in the car to record
speeds and travel times. A second form uses a modified recording speedometer of the type
regularly used in long-distance trucks or buses. One drawback of this approach is that it means
there are usually significantly fewer speed observations than volume observations. An example
of this kind of problem appears in Morton and Jackson (1992).
The other approach was developed by Wardrop and Charlesworth (1954) for urban traffic
measurements and is meant to obtain both speed and volume measurements simultaneously.
Although the method is not practical for major urban freeways, it is included here because it may
be of some value for rural expressway data collection, where there are no automatic systems.
While it is not appropriate as the primary mode of data collection on a busy freeway, there are
some useful points that come out of the literature that should be noted by those seeking to obtain
average speeds through the moving car method.
The method developed by Wardrop and Charlesworth is based on a survey vehicle that travels in
both directions on the road. The formulae allow one to estimate both speeds and flows for one
direction of travel. The two formulae are
2-11
(x + y)
q= (2.18)
( ta + tw )
y
t tw (2.19)
q
where
q is the estimated flow on the road in the direction of interest,
x is the number of vehicles traveling in the direction of interest, which are met by the
survey vehicle while traveling in the opposite direction,
y is the net number of vehicles that overtake the survey vehicle while traveling in the
direction of interest (i.e. those passing minus those overtaken),
ta is the travel time taken for the trip against the stream,
tw is the travel time for the trip with the stream, and
t is the estimate of mean travel time in the direction of interest.
Wright (1973) revisited the theory behind this method. His paper also serves as a review of the
papers dealing with the method in the two decades between the original work and his own. He
finds that, in general, the method gives biased results, although the degree of bias is not
significant in practice, and can be overcome. Wright's proposal is that the driver should fix the
journey time in advance, and keep to it. Stops along the way would not matter, so long as the
total time taken is as determined prior to travel. Wright's other point is that turning traffic
(exiting or entering) can upset the calculations done using this method. This fact means that the
route to be used for this method needs to avoid major exits or entrances. It should be noted also
that a large number of observations are required for reliable estimation of speeds and flows;
without that, the method has very limited precision.
2.2.2 Error caused by the mismatch between definitions and usual measurements
The overview in the previous section described methods that have historically been used
to collect observations on the key traffic variables. As was mentioned within that section, the
methods do not always accord with the definitions of these variables, as they were presented in
Section 2.1. One of the most common methods for collecting these data currently is based on
inductive loops embedded in the roadway. When speed data are also to be collected, a pair of
closely spaced loops (often called a speed trap) is needed, a known distance apart. Equivalent
systems are based on microwave beams that cover a part of a roadway surface. Cassidy and
Coifman (1997) point out the criteria that need to be met for the data from these systems to meet
the definitional requirements.
In practical terms, these criteria can be reduced to ensuring that any vehicle entering the
speed trap also clears it within the time interval for which the data are obtained – or that the
number of vehicles in the time interval is quite large. Neither of these criteria is met by the
ordinary implementation of inductive loop speed trap detectors. Whether or not the last vehicle
2-12
entering a speed trap will clear it within the interval is a simple random variable. There is no
guarantee that this criterion will regularly be met. If the number of vehicles were very large, this
would not be an issue, but because of the need for timely information, many systems poll the
loop detector controllers at least every minute, with some systems going to 30 second or even 20
second data collection. For intervals of that size, the volume counts on a single freeway lane
will be at most 40 (or 25 or 20) vehicles, which is not large enough to overcome the error
introduced by missing one of the vehicle’s speeds. While the data from these detector systems
are certainly good enough for operational decision-making, the data may give misleading results
if used directly with these equations because of the mismatch between the collection and the
definitions.
2.2.3 Importance of location to the nature of the data
Almost all of the bivariate models to be discussed represent efforts to explain the
behaviour of traffic variables over the full range of operation. In turning the models from
abstract representations into numerical models with specific parameter values, an important
practical question arises: can one expect that the data collected will cover the full range that the
model is intended to cover? If the answer is no, then the difficult question follows of how to do
curve fitting (or parameter estimation) when there may be essential data missing.
This issue can be explained more easily with an example. At the risk of oversimplifying
a relationship prior to a more detailed discussion of it, consider the simple representation of the
speed-flow curve as shown in Figure 2.5, for three distinct sections of roadway. The underlying
curve is assumed to be the same at all three locations. Between locations A and B, a major
entrance ramp adds considerable traffic to the road. If location B reaches capacity due to this
entrance ramp volume, there will be a queue of traffic on the mainstream at location A. These
vehicles can be considered to be waiting their turn to be served by the bottleneck section
immediately downstream of the entrance ramp. The data superimposed on graph A reflect the
situation whereby traffic at A had not reached capacity before the added ramp flow caused the
backup. There is a good range of uncongested data (on the top part of the curve), and congested
data concentrated in one area of the lower part of the curve. The flows for that portion reflect the
capacity flow at B less the entering ramp flows.
At location B, the full range of uncongested flows is observed, right out to capacity, but
the location never becomes congested, in the sense of experiencing stop-and-go traffic. It does,
however, experience congestion in the sense that speeds are below those observed in the absence
of the upstream congestion. Drivers arrive at the front end of the queue moving very slowly, and
accelerate away from that point, increasing speed as they move through the bottleneck section.
This segment of the speed-flow curve has been referred to as queue discharge flow, QDF (Hall et
al. 1992). The particular speed observed at B will depend on how far it is from the front end of
the queue (Persaud and Hurdle 1988). Consequently, the only data that will be observed at B are
on the top portion of the curve, and at some particular speed in the QDF segment.
If the exit ramp between B and C removes a significant portion of the traffic that was
observed at B, flows at C will not reach the levels they did at B. If there is no downstream
situation similar to that between A and B, then C will not experience congested operations, and
2-13
the data observable there will be as shown in Figure 2.5.
None of these locations taken alone can provide the data to identify the full speed-flow
curve. Location C can help to identify the uncongested portion, but cannot deal with capacity, or
with congestion. Location B can provide information on the uncongested portion and on
capacity operation, but cannot contribute to the discussion of congested operations. Location A
can provide some information on both uncongested and congested operations, but cannot tell
anything about capacity operations. This would all seem obvious enough. A similar discussion
appears in Drake et al (1967). It is also explained by May (1990). Other aspects of the effect of
location on data patterns are discussed by Hsu and Banks (1993). Yet a number of important
efforts to fit data to theory have ignored this key point (for example Ceder and May 1976; Easa
and May 1980). They have recognized that location-A data are needed to fit the congested
portion of the curve, but have not recognized that at such a location data are missing that are
needed to identify capacity. Consequently, discussion of bivariate models will look at the nature
of the data used in each study, and at where the data were collected (with respect to bottlenecks)
in order to evaluate the theoretical ideas. It is possible that the apparent need for several
different models, or for different parameters for the same model at different locations, or even
for discontinuous models instead of continuous ones, arose because of the nature (location) of
the data each was using.
2.2.4 Selecting intervals from which to extract data

In addition to the location for the data, there are also several issues relating to the time
intervals for collecting data. The first is the issue of obtaining representative data. By
examining trends on the cumulative curves, one can observe how flows and densities change
with time and space. By selecting flows and densities as they appear on the cumulative curves,
their values may be taken over intervals that exhibit near-constant slopes. In this way, the values
assigned to these properties are not affected by some arbitrarily selected measurement
interval(s). Choosing intervals arbitrarily is undesirable because data extracted over short
measurement intervals are highly susceptible to the effects of statistical fluctuations while the
use of longer intervals may average-out the features of interest. Further discussion and
demonstration of this in the context of freeway traffic is offered in (Cassidy, 1998).
There is also an issue of how many observations are needed to obtain good estimates of key
variables such as the capacity. A bottleneck’s capacity, qmax , is the maximum flow it can sustain for
a very long time (in the absence of any influences from restrictions further downstream). It can be
expressed mathematically as
§N ^ ·
q max { lim ¨¨ max ¸¸,
T of
© T ¹ (2.20)
where Nmax denotes that the vehicles counted during very long time T discharged through the
bottleneck at a maximum rate. The engineer assigns a capacity to a bottleneck by obtaining a value
for the estimator qmax (since one cannot actually observe a maximum flow for a time period
approaching infinity). It is desirable that the expected value of this estimator equal the capacity,
E(qmax) qmax. For this reason, one would collect samples (i.e., counts) immediately downstream of
an active bottleneck so as to measure vehicles discharging at a maximum rate. The amount qmax can
2-14
deviate from qmax is controlled by the sample size, N. A formula for determining N to estimate a
bottleneck’s capacity to a specified precision is derived below.
To begin this derivation, the estimator may be taken as
M
q^max ¦ nm /( M W),
m 1 (2.21)
where nm is the count collected in the mth interval and each of these M intervals has a duration of W.
If the {nm} can be taken as independent, identically distributed
^ random variables (e.g. the counts
were collected from consecutive intervals with W sufficiently large), then the variance of qmax can be
expressed as^
1 ª variance(n) º 1 ª variance(n) º
variance(q^max )
W 2 «¬ M »
¼ T «¬ W »
¼ (2.22)
since qmax is a linear function of the independent nm and the (finite) observation period T is the
denominator in (2.21).
The bracketed term variance(n) / W is a constant. Thus, by multiplying the top and bottom of this
quotient by E(n), the expected value of the counts, and by noting that E(n) / W = qmax , one obtains
^ J ^
variance(q max ) , (2.23)
T
where J is the index of dispersion; i.e., the ratio variance(n)/E(n).
The variance(qmax ) is the square of the standard error. Thus, by isolating T in (2.23) and then
multiplying both sides of the resulting expression by qmax , one arrives at
J
qmax T ,
H2 (2.24)
where qmax T N, the number of observations (i.e., the count) needed to estimate capacity to a
specified percent error H. Note, for example, that H = 0.05 to obtain an estimate within 5 percent of
qmax. The valueôf J may be estimated by collecting a presample and, notably, N increases rapidly as
H diminishes.
The expression N = J/H2 may be used to determine an adequate sample size when vehicles, or any
objects, discharging through an active bottleneck exhibit a nearly stationary flow; i.e., when the
cumulative count curve exhibits a nearly constant slope. If necessary, the N samples may be
obtained by concatenating observations from multiple days. Naturally, one would take samples
during time periods thought to be representative of the conditions of interest. For example, one
should probably not use vehicle counts taken in inclement weather to estimate the capacity for
fair weather conditions.
2.3 Bivariate models
This section provides an overview of work to establish relationships among pairs of the
variables described in the opening section. Some of these efforts begin with mathematical
models; others are primarily empirical, with little or no attempt to generalize. Both are
important components for understanding the relationships. For reasons discussed above, two
aspects of these efforts are emphasized here: the measurement methods used to obtain the data;
and the location at which the measurements were obtained.
2-15
2.3.1 Speed-flow models
The speed-flow relationship is the bivariate relationship on which there has been the
greatest amount of work recently, so is the first one discussed. This section starts from current
understanding, which provides some useful insights for interpreting earlier work, and then moves
to a chronological review of some of the major models.
The bulk of then recent empirical work on the relationship between speed and flow (as
well as the other relationships) was summarized in a paper by Hall, Hurdle, and Banks (1992).
In it, they proposed the model for traffic flow shown in Figure 2.6. This figure is the basis for
the background speed-flow curve in Figure 2.5.
It is useful to summarize some of the antecedents of the understanding depicted in Figure

2.6. The initial impetus came from a paper by Persaud and Hurdle (1988), in which they
demonstrated (Figure 2.7) that the vertical line for queue discharge flow in Figure 2.6 was a
reasonable result of measurements taken at various distances downstream from a queue. This
study was an outgrowth of an earlier one by Hurdle and Datta (1983) in which they raised a
number of questions about the shape of the speed-flow curve near capacity.
Additional empirical work dealing with the speed-flow relationship was conducted by
Banks (1989, 1990), Hall and Hall (1990), Chin and May (1991), Wemple, Morris and May
(1991), Agyemang-Duah and Hall (1991) and Ringert and Urbanik (1993). All of these studies
supported the idea that speeds remain nearly constant even at quite high flow rates. Another of
the important issues they dealt with is one that had been around for over thirty years
(Wattleworth 1963): is there a reduction in flow rates within the bottleneck at the time that the
queue forms upstream? Figure 2.6 shows such a drop on the basis of two studies. Banks (1991a,
1991b) reports roughly a 3% drop from pre-queue flows, on the basis of nine days of data at one
site in California. Agyemang-Duah and Hall (1991) found about a 5% decrease, on the basis of
52 days of data at one site in Ontario. This decrease in flow is often not observable, however, as
in many locations high flow rates do not last long enough prior to the onset of congestion to
yield the stable flow values that would show the drop.
Two empirical studies in Germany support the upper part of the curve in Figure 2.6 quite
well. Heidemann and Hotop (1990) found a piecewise-linear 'polygon' for the upper part of the
curve (Figure 2.8). Unfortunately, they did not have data beyond 1700 veh/hr/lane, so could not
address what happens at capacity. Stappert and Theis (1990) conducted a major empirical study
of speed-flow relationships on various kinds of roads. However, they were interested only in
estimating parameters for a specific functional form,
V = (A - eBQ) e-c - K edQ (2.25)
where V is speed, Q is traffic volume, c and d are constant "Krummungs factors" taking values
between 0.2 and 0.003, and A, B, and K are parameters of the model. This function tended to
2-16
2-17
2-18
give the kind of result shown in Figure 2.9, despite the fact that the curve does not accord well
with the data near capacity. In Figure 2.9, each point represents a full hour of data, and the
graph represents five months of hourly data. Note that flows in excess of 2200 veh/h/lane were
sustained on several occasions, over the full hour.
The problem for traffic flow theory is that these curves are empirically derived. There is
not really any theory that would explain these particular shapes, except perhaps for Edie et
al.(1980), who propose qualitative flow regimes that relate well to these curves. The task that lies
ahead for traffic flow theorists is to develop a consistent set of equations that can replicate this
reality. The models that have been proposed to date, and will be discussed in subsequent
sections, do not necessarily lead to the kinds of speed-flow curves that data suggest are needed.
It is instructive to review the history of depictions of speed-flow curves in light of this

current understanding. Probably the seminal work on this topic was the paper by Greenshields in
1935, in which he derived the following parabolic speed-flow curve on the basis of a linear
speed-density relationship together with the equation, flow = speed * density:
u2
q = kj (u - u ) (2.26)
f
Figure 2.10 contains that curve, and the data it is based on, redrawn. The numbers against the
data points represent the "number of 100-vehicle groups observed", on Labor Day 1934, in one
direction on a two-lane two-way road (p. 464). In counting the vehicles on the road, every 10th
vehicle started a new group (of 100), so there is 90% overlap between two adjacent groups (p.
451): the groups are not independent observations. Equally important, the data have been
grouped in flow ranges of 200 veh/h and the averages of these groups taken prior to plotting.
The one congested point, representing 51 (overlapping) groups of 100 observations, came from a
different roadway entirely, with different cross-section and pavement, which were collected on a
different day.
These details are mentioned here because of the importance to traffic flow theory of
Greenshields' work. The parabolic shape he derived was accepted as the proper shape of the
curve for decades. In the 1965 Highway Capacity Manual, for example, the shape shown in
Figure 2.10 appears exactly, despite the fact that data displayed elsewhere in the 1965 HCM
showed that contemporary empirical results did not match the figure. In the 1985 HCM, the
same parabolic shape was retained, although broadened considerably. In short, Greenshields'
model dominated the field for over 50 years, despite the fact that by current standards of research
the method of analysis of the data, with overlapping groups and averaging prior to curve-fitting,
would not be acceptable.
There is another criticism that can be addressed to Greenshields' work as well, although it
is one of which a number of current researchers seem unaware. Duncan (1976; 1979) has shown
that calculating density from speed and flow, fitting a line to the speed-density data, and then
converting that line into a speed-flow function, gives a biased result relative to direct estimation
2-19
2-20
of the speed-flow function. This is a consequence of three things discussed earlier: the non-
linear transformations involved in both directions, the stochastic nature of the observations, and
the inability to match the time and space measurement frames exactly.
It is interesting to contrast the emphasis on speed-flow models in recent years with that
20 years ago, as represented in TRB SR 165 (Gerlough and Huber 1975). In that volume, the
discussion of speed-flow models took up less than a page of text, and none of the five
accompanying diagrams dealt with freeways. (Three dealt with the artificial situation of a test
track.) In contrast, five pages and eleven figures were devoted to the speed-density relationship.
Speed-flow models are important for freeway management strategies, and will be of
fundamental importance for intelligent vehicle/highway systems (IVHS) implementation of
alternate routing; hence there is considerably more work on this topic than on the remaining two
bivariate topics. Twenty and more years ago, the other topics were of more interest. As
Gerlough and Huber stated the matter (p. 61), "once a speed-concentration model has been
determined, a speed-flow model can be determined from it." That is in fact the way most earlier
speed-flow work was treated (including that of Greenshields). Hence it is sensible to turn to
discussion of speed-concentration models, and to deal with any other speed-flow models as a
consequence of speed-concentration work, which is the way they were developed.
2.3.2 Speed-concentration models
Greenshields' (1935) linear model of speed and density was mentioned in the previous
section. It is:
u = uf (1-k/kj) (2.27)
where uf is the free-flow speed, and kj is the jam density. The measured data were speeds and
flows; density was calculated using equation 2.27. The most interesting aspect of this particular
model is that its empirical basis consisted of half a dozen points in one cluster near free-flow
speed, and a single observation under congested conditions (Figure 2.11). The linear
relationship comes from connecting the cluster with the single point: "since the curve is a
straight line it is only necessary to determine accurately two points to fix its direction" (p. 468).
What is surprising is not that such simple analytical methods were used in 1935, but that their
results (the linear speed-density model) have continued to be so widely accepted for so long.
While there have been studies that claimed to have confirmed this model, such as that in Figure
2.12a (Huber 1957), they tended to have similarly sparse portions of the full range of data,
usually omitting both the lowest flows, and flow in the range near capacity. There have also
been a number of studies that found contradictory evidence, most importantly that by Drake, et
al. (1965), which will be discussed in more detail subsequently.
A second early model was that put forward by Greenberg (1959), showing a logarithmic
relationship:
2-21
u = um ln(k/kj) (2.28)
The paper showed the fit of the model to two data sets, both of which visually looked very
reasonable. However, the first data set was derived from speed and headway data on individual
vehicles, which "was then separated into speed classes and the average headway was calculated
for each speed class" (p. 83). In other words, the vehicles that appear in one data point (speed
class) may not even have been travelling together! While a density can always be calculated as
the reciprocal of average headway, when that average is taken over vehicles that may well not
have been travelling together, it is not clear what that density is meant to represent. The second
data set used by Greenberg was Huber's. This is the same data that appears in Figure 2.12a;
Greenberg's graph is shown in Figure 2.12b. Visually, the fit is quite good, but Huber reported
an R of 0.97, which does not leave much room for improvement.
A third model from the same period is that by Edie (1961). His model was an attempt to
deal with a discontinuity that had consistently been found in data near the critical density, which
“suggested the existence of some kind of change of state” (p. 72). He proposed two linear
models for the two states of flow. The first related density to the logarithm of velocity above the
“optimum velocity”, i.e. “non-congested flow”. The second related velocity to the logarithm of
spacing (i.e. the inverse of density) for congested flow. The model was fitted to the same
Lincoln Tunnel data as used by Greenberg.
These three forms of the speed-density curve, plus four others, were investigated in an
empirical test by Drake et al. in 1967. The test used data from the middle lane of the Eisenhower
Expressway in Chicago, one-half mile (800 m) upstream from a bottleneck whose capacity was
only slightly less than the capacity of the study site. This location was chosen specifically in
order to obtain data over as much of the range of operations as possible. A series of 1224 1-
minute observations were initially collected. The measured data consisted of volume, time-mean
speed, and occupancy. Density was calculated from volume and time-mean speed. A sample
was then taken from among the 1224 data points in order to create a data set that was uniformly
distributed along the density axis, as is assumed by regression analysis of speed on density.
The intention in conducting the study was to compare the seven speed-density hypotheses
statistically, and thereby to select the best one. In addition to Greenshields' linear form,
Greenberg's exponential curve, and Edie’s two-part logarithmic model, the other four
investigated were a two-part and three-part piecewise linear models, Underwood's (1961)
transposed exponential curve, and a bell-shaped curve. Despite the intention to use "a rigorous
structure of falsifiable tests" (p. 75) in this comparison, and the careful way the work was done,
the statistical analyses proved inconclusive: "almost all conclusions were based on intuition
alone since the statistical tests provided little decision power after all" (p. 76). To assert that
intuition alone was the basis is no doubt a bit of an exaggeration. Twenty-one graphs help
considerably in differentiating among the seven hypotheses and their consequences for both
speed-volume and volume-density graphs.
2-22
2-23
Figure 2.13 provides an example of the three types of graphs used, in this case the ones
based on the Edie model. Their comments about this model (p. 75) were: "The Edie formulation
gave the best estimates of the fundamental parameters. While its R2 was the second lowest, its
standard error was the lowest of all hypotheses." One interesting point with respect to Figure
2.13 is that the Edie model was the only one of the seven to replicate capacity operations closely
on the volume-density and speed-volume plots. The other models tended to underestimate the
maximum flows, often by a considerable margin, as is illustrated in Figure 2.14, which shows the
speed-volume curve resulting from Greenshields' hypothesis of a linear speed-density
relationship. (It is interesting to note that the data in these two figures are quite consistent with
the currently accepted speed-flow shape identified earlier in Figures 2.5 and 2.6.) The overall
conclusion one might draw from the Drake et al. study is that none of the seven models they
tested provide a particularly good fit to or explanation of the data, although it should be noted
that they did not state their conclusion this way, but rather dealt with each model separately.
There are four additional issues that arise from the Drake et al. study that are worth
noting here. The first is the methodological one identified by Duncan (1976; 1979), and
discussed earlier with regard to Greenshields' work. Duncan showed that calculating density
from speed and flow data, fitting a speed-density function to that data, and then transforming the
speed-density function into a speed-flow function results in a curve that does not fit the original
speed-flow data particularly well. This is the method used by Drake et al, and certainly most of
their resulting speed-flow functions did not fit the original speed-flow data very well. Duncan's
1979 paper expanded on the difficulties to show that minor changes in the speed-density function
led to major changes in the speed-flow function, suggesting the need for further caution in using
this method of double transformations to get a speed-flow curve.
The second is that of the data collection location, as discussed above. The data were
collected upstream of a bottleneck, which produced the kind of discontinuity that Edie had
earlier identified. The Drake et al. approach was to try to fit the data as they had been obtained,
without considering whether there was a portion of the data that was missing. They had
intentionally tried to obtain data from as wide a range as possible, but as discussed above it may
not be possible to get data from all three parts of the curve at one location.
The third issue is that identified by Cassidy, relating to the time period for collection of
the data. The Drake et al. data came from standard loop detectors, working on fixed time
intervals. As a consequence, there will be some measurement error, which may well affect the
estimation of the bivariate relationships.
The fourth issue is the relationship between car-following models (see Chapter 5) and the
models tested by Drake et al. They noted that four of the models they tested "have been shown
to be directly related to specific car-following rules", and cited articles by Gazis and co-authors
(1959; 1961). The interesting question to raise in the context of the overall appraisal of the
Drake et al. results is whether those results raise doubts about the validity of the car-following
models for freeways. The car-following models gave rise to four of the speed-density models
tested by Drake et al. The results of the testing suggest that the speed-density models are not
2-24
2-25
particularly good. Modus tollens in logic says that if the consequences of a set of premises are
shown to be false, then one (at least) of the premises is not valid. It is possible, then, that the
car-following models are not valid for freeways. (This is not surprising, as they were not
developed for this context. Nor, it seems, are they used as the basis for contemporary
microscopic freeway simulation models.) On the other hand, any of the three issues just
identified may be the source of the failure of the models, rather than their development from car-
following models.
2.3.3 Flow-concentration models
Although Gerlough and Huber did not give the topic of flow-concentration models such
extensive treatment as they gave the speed-concentration models, they nonetheless thought this
topic to be very important, as evidenced by their introductory paragraph for the section dealing
with these models (p. 55):
Early studies of highway capacity followed two principal approaches. Some

investigators examined speed-flow relationships at low concentrations; others discussed
headway phenomena at high concentrations. Lighthill and Whitham [1955] have
proposed use of the flow-concentration curve as a means of unifying these two
approaches. Because of this unifying feature, and because of the great usefulness of the
flow-concentration curve in traffic control situations (such as metering a freeway),
Haight [1960; 1963] has termed the flow-concentration curve "the basic diagram of
traffic".
Nevertheless, most flow-concentration models have been derived from assumptions about the
shape of the speed-concentration curve. This section deals primarily with work that has focused
on the flow-concentration relationship directly. Under that heading is included work that uses
either density or occupancy as the measure of concentration.
Edie was perhaps the first to point out that empirical flow-concentration data frequently
have discontinuities in the vicinity of what would be maximum flow, and to suggest that
therefore discontinuous curves might be needed for this relationship. (An example of his type of
curve appears in Figure 2.13 above.) This suggestion led to a series of investigations by May
and his students (Ceder 1975; 1976; Ceder and May 1976; Easa and May 1980) to specify more
tightly the nature and parameters of these "two-regime" models (and to link those parameters to
the parameters of car-following models). The difficulty with their resulting models is that the
models often do not fit the data well at capacity (with results similar to those shown in Figure
2.14 for Greenshields' single-regime model). In addition, there seems little consistency in
parameters from one location to another. Even more troubling, when multiple days from the
same site were calibrated, the different days required quite different parameters.
Koshi, Iwasaki and Ohkura (1983) gave an empirically-based discussion of the flow-
density relationship, in which they suggested that a reverse lambda shape was the best
description of the data (p. 406): "the two regions of flow form not a single downward concave
2-26
curve... but a shape like a mirror image of the Greek letter lamda [sic] (Ȝ)". These authors also
investigated the implications of this phenomenon for car-following models, as well as for wave
propagation. Data with a similar shape to theirs appears in Figure 2.13; Edie’s equations fit
those data with a shape similar to the lambda shape Koshi et al. suggested.
Although most of the flow-concentration work that relies on occupancy rather than
density dates from the past decade, Athol suggested its use nearly 30 years earlier (in 1965). His
work presages a number of the points that have come out subsequently and are discussed in more
detail below: the use of volume and occupancy together to identify the onset of congestion; the
transitions between uncongested and congested operations at volumes lower than capacity; and
the use of time-traced plots (i.e. those in which lines connected the data points that occurred
consecutively over time) to better understand the operations.
After Athol's early efforts, there seems to have been a dearth of efforts to utilize the
occupancy data that was available, until the mid-1980s. One paper from that time (Hall et al.
1986) that utilized occupancy drew on the same approach Athol had used, namely the
presentation of time-traced plots. Figure 2.15 shows results for four different days from the
same location, 4 km upstream of a primary bottleneck. The data are for the left-most lane only
(the high-speed, or passing lane), and are for 5-minute intervals. The first point in the time-
connected traces is the one that occurred in the 5-minute period after the data-recording system
was turned on in the morning. In part D of the Figure, it is clear that operations had already
broken down prior to data being recorded. Part C is perhaps the most intriguing: operations
move into higher occupancies (congestion) at flows clearly below maximum flows. Although
Parts A and B may be taken to confirm the implicit assumption many traffic engineers have that
operations pass through capacity prior to breakdown, Part C gives a clear indication that this
does not always happen. Even more important, all four parts of Figure 2.15 show that operations
do not go through capacity in returning from congested to uncongested conditions. Operations
can 'jump' from one branch of the curve to the other, without staying on the curve. (This same
result, not surprisingly, occurs for speed-flow data.)
Each of the four parts of Figure 2.15 show at least one data point between the two
'branches' of the usual curve during the return to uncongested conditions. Because these were 5-
minute data, the authors recognized that these points might be the result of averaging of data
from the two separate branches. Subsequently, however, additional work utilizing 30-second
intervals confirmed the presence of these same types of data (Persaud and Hall 1989). Hence
there appears to be strong evidence that traffic operations on a freeway can move from one
branch of the curve to the other without going all the way around the capacity point. This is an
aspect of traffic behaviour that none of the mathematical models discussed above either explain
or lead one to expect. Nonetheless, the phenomenon has been at least implicitly recognized
since Lighthill and Whitham's (1955) discussion of shock waves in traffic, which assumes
instantaneous jumps from one branch to the other on a speed-flow or flow-occupancy curve. As
well, queueing models (e.g. Newell 1982) imply that immediately upsteam from the back end of
a queue there must be points where the speed is changing rapidly from the uncongested branch
of the speed-flow curve to that of the congested branch. It would be beneficial if flow-
2-27
2-28
concentration (and speed-flow) models explicitly took this possibility into account.
One of the conclusions of the paper by Hall et al. (1986) from which Figure 2.15 is drawn
is that an inverted 'V' shape is a plausible representation of the flow-occupancy relationship.
Although that conclusion was based on limited data from near Toronto, Hall and Gunter (1986)
supported it with data from a larger number of stations. Banks (1989) tested their proposition
using data from the San Diego area, and confirmed the suggestion of the inverted 'V'. He also
offered a mathematical statement of this proposition and a behavioural interpretation of it (p.
58):
The inverted-V model implies that drivers maintain a roughly constant average
time gap between their front bumper and the back bumper of the vehicle in front
of them, provided their speed is less than some critical value. Once their speed
reaches this critical value (which is as fast as they want to go), they cease to be
sensitive to vehicle spacing....
2.4 Three-dimensional models
There has not been a lot of work that attempts to treat all three traffic flow variables
simultaneously. Gerlough and Huber presented one figure (reproduced as Figure 2.16) that
represented all three variables, but said little about this, other than (1) "The model must be on the
three-dimensional surface u = q/k," and (2) "It is usually more convenient to show the model of
[Figure 2.16] as one or more of the three separate relationships in two dimensions..." (p. 49). As
was noted earlier, however, empirical observations rarely accord exactly with the relationship
q=u k, especially when the observations are taken during congested conditions. Hence focusing
on the two-dimensional relationships will not often provide even implicitly a valid three-
dimensional relationship.
In this context, a paper by Gilchrist and Hall (1989) is interesting because it

presents three-dimensional representations of empirical observations, without attempting
to fit them to a theoretical representations. The study is limited, in that data from only
one location, upstream of a bottleneck, was presented. To enable better visualization of
the data, a time-connected trace was used. The projections onto the standard two-
dimensional surfaces of the data look much as one might expect. The surprises came in
looking at oblique views of the three-dimensional representation, as in Figures 2.17 and
2.18. From one perspective (Figure 2.17), the traditional sideways U-shape that we have
been led to expect is quite apparent, and projections of that are easily visualized onto, for
example, the speed-flow surface (the face labeled with a 3, on the left side of the ‘box’).
From a different perspective (Figure 2.18) that shape is hardly apparent at all, although
indications of an inverted 'V' can be seen, which would project onto a flow-occupancy
plot on face 1 of the box.. Black and white were alternated for five speed ranges. In both
figures, the dark lines at the left represent the data with speeds above 80 km/hr. The light
lines closest to the left cover the range 71 to 80 km/hr; the next, very small, area of dark
lines contains the range 61 to 70 km/hr; the remaining light lines represent the range 51
to 60 km/hr; and the dark lines to the right of the diagram represent speeds below 50
km/hr.
2-29
2-30
One recent approach to modelling the three traffic operations variables directly has been
based on the mathematics of catastrophe theory. (The name comes from the fact that while most
of the variables being modelled change in a continuous fashion, at least one of the variables can
make sudden discontinuous changes, referred to as catastrophes by Thom (1975), who originally
developed the mathematics for seven such models, ranging from two dimensions to eight.) The
first effort to apply these models to traffic data was that by Dendrinos (1978), in which he
suggested that the two-dimensional catastrophe model could represent the speed-flow curve. A
more fruitful model was proposed by Navin (1986), who suggested that the three-dimensional
'cusp' catastrophe model was appropriate for the three traffic variables. The feature of the cusp
catastrophe surface that makes it of interest in the traffic flow context is that while two of the
variables (the control variables) exhibit smooth continuous change, the third one (the state
variable) can undergo a sudden 'catastrophic' jump in its value. Navin suggested that speed was
the variable that underwent this catastrophic change, while flow and occupancy were the control
variables.
While Navin's presentation was primarily an intuitive one, without recourse to data, Hall
and co-authors picked up on the idea and attempted to flesh it out both mathematically and
empirically. The initial effort appears in Hall (1987), in which the basic idea was presented, and
some data applied. Figure 2.19 is a representation, showing the partially-folded (and torn)
surface that is the Maxwell version of the cusp catastrophe surface, approximately located traffic
data on that surface, and axes external to the surface representing the general correspondence
with traffic variables. Further elaboration of this model is provided by Persaud and Hall (1989).
Acha-Daza and Hall (1993) compared the ability of this model with the ability of some of the
earlier models discussed above, to estimate speeds from flows obtained using inductive loop
detectors and 30-second data, and found the catastrophe theory model to be slightly better than
they were. Despite its empirical success, the problem with the model is that it appears to be
inconsistent with the basic definitions and relationships with which this chapter opened.
2.5 Summary and links to other chapters
The current status of mathematical models for speed-flow-concentration relationships is

in a state of flux. The models that dominated the discourse for nearly 30 years are incompatible
with the data currently being obtained, and with currently accepted depictions of speed-flow
curves, but no replacement models have yet been developed. The analyses of Cassidy and
Newell have shown that the data used to develop most of the earlier models were flawed, as has
been described above. An additional difficulty was noted by Duncan (1976; 1979):
transforming variables, fitting equations, and then transforming the equations back to the original
variables can lead to biased results, and is very sensitive to small changes in the initial curve-
fitting. Progress has been made in understanding the relationships among the key traffic
variables, but there is considerable scope for better models still.
It is important to note that the variables and analyses that have been discussed in this
chapter are closely related to topics in several of the succeeding chapters. The human factors
discussed in Chapter 3, for example, help to explain some of the variability in the data that have
2-31
Figure 2.19
Catastrophe Theory Surface Showing Sketch of a Possible Freeway Function, and
Projections and Transformations from That (Hall 1987).
2-32
been discussed here. As mentioned earlier, some of the bivariate models discussed above have
been derived from car-following models, which are covered in Chapter 4. In addition, brief
mention was made here of Lighthill and Witham’s work, which has spawned a large literature
related to the behavior of shocks and waves in traffic flow, covered in Chapter 5. All of these
topics are inter-related, but have been addressed separately for ease of understanding.
Acknowledgements
The assistance of Michael Cassidy in developing this latest revision of this chapter was
very helpful, as noted in the preface to the chapter. Valuable comments on an early draft of this
material were received from Jim Banks, Frank Montgomery and Van Hurdle. The assistance of
Richard Cunard and the TRB in bringing this project to completion is also appreciated.
References
Acha-Daza, J.A. and Hall, F.L. 1993. A graphical comparison of the predictions for speed given
by catastrophe theory and some classic models, Transportation Research Record 1398, 119-124.
Agyemang-Duah, K. and Hall, F.L. 1991. Some issues regarding the numerical value of freeway
capacity, in Highway Capacity and Level of Service, Proceedings of the International
Symposium on Highway Capacity, Karlsruhe. U. (Brannolte, ed.) Rotterdam, Balkema, 1-15.
Athol, P. 1965. Interdependence of certain operational characteristics within a moving traffic

stream. Highway Research Record 72, 58-87.
Banks, J.H. 1989. Freeway speed-flow-concentration relationships: more evidence and

interpretations, Transportation Research Record 1225, 53-60.
Banks, J.H. 1990. Flow processes at a freeway bottleneck. Transportation Research Record
1287, 20-28.
Banks, J.H. 1991a. Two-capacity phenomenon at freeway bottlenecks: A basis for ramp
metering?, Transportation Research Record 1320, 83-90.
Banks, J.H. 1991b. The two-capacity phenomenon: some theoretical issues, Transportation
Research Record 1320, pp. 234-241.
Cassidy, MJ 1998 Bivariate relations in nearly stationary highway traffic. Transp Res B, 32B,
49-59.
Cassidy, MJ 1999 Traffic flow and capacity, in Handbook of transportation science, RW Hall
(ed). Norwell, MA: Kluwer Academic Publishers.
2-33
Cassidy , MJ and Coifman, B 1997 Relation among average speed, flow and density and the
analogous relation between density and occupancy. Transportation Research Record 1591, 1-6.
Ceder, A. 1975. Investigation of two-regime traffic flow models at the micro- and macro-scopic
levels. Ph.D. dissertation, University of California, Berkeley, California.
Ceder, A. 1976. A deterministic traffic flow model for the two-regime approach. Transportation
Research Record 567, 16-32.
Ceder, A. and May, A.D. 1976. Further evaluation of single- and two-regime traffic flow models.
Transportation Research Record 567, 1-15.
Chin, H.C. and May, A.D. 1991. Examination of the speed-flow relationship at the Caldecott
Tunnel, Transportation Research Record 1320.
Daganzo, CF 1997 Fundamentals of transportation and traffic operations. New York:

Elsevier.
Dendrinos, D.S. 1978. Operating speeds and volume to capacity ratios: the observed
relationship and the fold catastrophe, Transportation Research Record 12, 191-194.
Drake, J.S., Schofer, J.L., and May, A.D. 1967. A Statistical Analysis of Speed Density
Hypotheses, Highway Research Record 154, 53-87.
Duncan, N.C. 1976. A note on speed/flow/concentration relations, Traffic Engineering and

Control, 34-35.
Duncan, N.C. 1979. A further look at speed/flow/concentration, Traffic Engineering and

Control, 482-483.
Easa, S.M. and May, A.D. 1980. Generalized procedures for estimating single- and two-regime
traffic flow models, Transportation Research Record 772, 24-37.
Edie, L.C. 1961. Car following and steady-state theory for non-congested traffic, Operations
Research 9, 66-76.
Edie, LC 1965 Discussion of traffic stream measurements and definitions. Proc Int Symp. On
the Theory of Traffic Flow J. Almond (ed) Paris: OECD, 139-154.
Edie, LC 1974 Flow theories, in Traffic Science DC Gazis (ed) New York: Wiley, 8-20.
Edie, LC and Foote, R 1960 Effect of shock waves on tunnel traffic flow. Proc Highway Res
Board 39, 492-505.
2-34
Edie, L.C., Herman, R., and Lam, T., 1980. Observed multilane speed distributions and the
kinetic theory of vehicular traffic, Transportation Science 14, 55-76.
Gazis, D.C., Herman, R., and Potts, R. 1959. Car-following theory of steady-state traffic flow.
Operations Research 7, 499-505.
Gazis, D.C., Herman, R., and Rothery, R.W. 1961. Nonlinear follow-the-leader models of traffic
flow. Operations Research 9, 545-567.
Gerlough, D.L. and Huber, M.J. 1975. Traffic Flow Theory: a monograph. Special Report 165,
Transportation Research Board (Washington D.C.: National Research Council)
Gilchrist, R. 1988. Three-dimensional relationships in traffic flow theory variables.

Unpublished Master's degree report, Dept. of Civil Engineering, McMaster University,
Hamilton, Ontario, Canada.
Gilchrist, R.S. and Hall, F.L. 1989. Three-dimensional relationships among traffic flow theory
variables, Transportation Research Record 1225, 99-108.
Greenberg, H. 1959. An analysis of traffic flow, Operations Research, Vol. 7, 78-85.
Greenshields, B.D. 1935. A study of traffic capacity, HRB Proceedings 14, 448-477.
Haight, F.A. 1960. The volume, density relation in the theory of road traffic, Operations
Research 8, 572-573.
Haight, F.A. 1963. Mathematical Theories of Traffic Flow. (New York: Academic Press)
Hall, F.L. 1987. An interpretation of speed-flow-concentration relationships using catastrophe

theory. Transportation Research A, 21A, 191-201.
Hall, F.L., Allen, B.L. and Gunter, M.A. 1986. Empirical analysis of freeway flow-density
relationships, Transportation Research A, 20A, 197-210.
Hall, F.L. and Gunter, M.A. 1986. Further analysis of the flow-concentration relationship,
Hall, F.L. and Hall, L.M. 1990. Capacity and speed flow analysis of the QEW in Ontario"
Hall, F.L, Hurdle, V.F. and Banks, J.H. 1992. Synthesis of recent work on the nature of speed-
flow and flow-occupancy (or density) relationships on freeways. In Transportation Research
Record 1365, TRB, National Research Council, Washington, D.C. 12-18.
Heidemann, D. and R. Hotop, R. 1990. Verteilung der Pkw-Geschwindigkeiten im Netz der
2-35
Bundesautobahnen- Modellmodifikation und Aktualisierung. Straße und Autobahn - Heft 2, 106-

113.
HCM 1965. Special Report 87: Highway Capacity Manual. Transportation Research Board.
HCM 1985. Special Report 209: Highway Capacity Manual. Transportation Research Board.
Hsu, P. and Banks, J.H. 1993. Effects of location on congested-regime flow-concentration

relationships for freeways, Transportation Research Record 1398, 17-23.
Huber, M.J. 1957. Effect of temporary bridge on parkway performance. Highway Research
Board Bulletin 167 63-74.
Hurdle, V.F. and Datta, P.K. 1983. Speeds and flows on an urban freeway: some measurements
and a hypothesis. Transportation Research Record 905, 127-137.
Koshi, M., Iwasaki, M. and Okhura, I. 1983. Some findings and an overview on vehicular flow
characteristics. Proceedings, 8th International Symposium on Transportation and Traffic Flow
Theory (Edited by Hurdle, V.F., Hauer, E. and Steuart, G.F.) University of Toronto Press,
Toronto, Canada, 403-426.
Kühne, R. and Immes, S. 1993. Freeway control systems for using section-related traffic variable
detection. Proceedings of the Pacific Rim TransTech Conference, Seattle. New York: Am. Soc.
Civil Eng., 56-62.
Lighthill, M.J., and Whitham, G.B. 1955. On kinematic waves: II. A theory of traffic flow on
long crowded roads. Proceedings of the Royal Society A229 No. 1178, 317-145.
Makagami, Y., Newell, G.F. and Rothery, R. 1971. Three-dimensional representation of traffic
flow, Transportation Science, Vol. 5, No. 3, pp. 302-313.
May, A.D. 1990. Traffic Flow Fundamentals. Prentice-Hall.
May, A.D., Athol, P., Parker, W. and Rudden, J.B. 1963. Development and evaluation of
Congress Street Expressway pilot detection system, Highway Research Record 21, 48-70.
Morton, T.W. and Jackson, C.P. 1992. Speed/flow geometry relationships for rural dual-
carriageways and motorways. TRRL Contractor Report 279. 55 pages. Transport and Road
Research Laboratory, Crowthorne, Berkshire.
Moskowitz, K 1954 Waiting for a gap in a traffic stream Proc Highway Res Board 33, 385-395.
Navin, F. 1986. Traffic congestion catastrophes. Transportation Planning Technology 11, 19-25.
2-36
Newell G.F. 1971 Applications of Queueing Theory. London: Chapman Hall.
Newell, G.F. 1982. Applications of Queueing Theory, second edition. London: Chapman and
Hall.
Newell, G.F. 1993. A simplified theory of kinematic waves in highway traffic, Transportation
Research B, 27B: Part I, General theory, 281-287; Part II, Queueing at freeway bottlenecks,
289-303; Part III, Multi-destination flows, 305-313.
Newell, GF unpublished Notes on transportation operations Berkeley, CA: Univesity of

California.
Persaud, B.N. and Hall, F.L. 1989. Catastrophe theory and patterns in 30-second freeway data --
implications for incident detection, Transportation Research A 23A, 103-113.
Persaud, B.N. and Hurdle, V.F. 1988. Some new data that challenge some old ideas about speed-
flow relationships. Transportation Research Record 1194, 191-198.
Ringert, J. and Urbanik, II, T. 1993. Study of freeway bottlenecks in Texas, Transportation
Robertson, D. 1994. Manual of Transportation Engineering Studies, fifth edition. Arlington,

VA: Institute of Transportation Engineers.
Stappert, K.H. and Theis, T.J. 1990. Aktualisierung der Verkehrsstärke-

/Verkehrsgeschwindigkeitsbeziehungen des BVWP-Netzmodells. Research Report
VU18009V89 Heusch/Boesefeldt, Aachen.
Thom, R. 1975. Structural Stability and Morphogenesis. English translation by D.H. Fowler of
Stabilite Structurelle et Morphogenese. Benjamin: Reading, Massachusetts.
Underwood, R.T. 1961. Speed, volume, and density relationships: Quality and theory of traffic
flow. Yale Bureau of Highway Traffic, pp. 141-188, as cited in Drake et al. 1967.
Wardrop, J.G. and Charlesworth, G. 1954. A method of estimating speed and flow of traffic from
a moving vehicle. Proceedings of the Institution of Civil Engineers, Part II, Vol. 3, 158-171.
Wattleworth, J.A. 1963. Some aspects of macroscopic freeway traffic flow theory, Traffic
Engineering 34: 2, 15-20.
Wemple, E.A., Morris, A.M. and May, A.D. 1991. Freeway capacity and level of service
concepts, in Highway Capacity and Level of Service, Proc. of the International Symposium on
Highway Capacity, Karlsruhe (U. Brannolte, Ed.) Rotterdam: Balkema, 439-455.
Wright, C. 1973. A theoretical analysis of the moving observer method, Transportation

Research 7, 293-311.
2-37

HUMAN FACTORS
BY RODGER J. KOPPA5
5
Associate Professor, Texas A&M University, College Station, TX 77843.
= parameter of log normal distribution ~ standard deviation

= parameter of log normal distribution ~ median
) = standard deviation
0 = value of standard normal variate
aGV = maximum acceleration on grade
aLV = maximum acceleration on level
A = movement amplitude
Cr = roadway curvature
C(t) = vehicle heading
CV = coefficient of variation
d = braking distance
D = distance from eye to target
E(t) = symptom error function
f = coefficient of friction
Fs = stability factor
g = acceleration of gravity
g(s) = control displacement
G = gradient
H = information (bits)
K = gain (dB)
l = wheel base
L = diameter of target (letter or symbol)
LN = natural log
M = mean
MT = movement time
N = equiprobable alternatives
PRT = perception-response time
R(t) = desired input forcing function
RT = reaction time (sec)
s = Laplace operator
SR = steering ratio (gain)
SSD = stopping sight distance
t = time
TL = lead term constant
T #
= lag term constant
TN = neuro-muscular time constant
u = speed
V = initial speed
W = width of control device
Z = standard normal score
3.
HUMAN FACTORS
3.1 Introduction
In this chapter, salient performance aspects of the human in the 3.1.1 The Driving Task
context of a person-machine control system, the motor vehicle,
will be summarized. The driver-vehicle system configuration is Lunenfeld and Alexander (1990) consider the driving task to be
ubiquitous. Practically all readers of this chapter are also a hierarchical process, with three levels: (1) Control,
participants in such a system; yet many questions, as will be (2) Guidance, and (3) Navigation. The control level of
seen, remain to be answered in modeling the behavior of the performance comprises all those activities that involve second-
human component alone. Recent publications (IVHS 1992; to-second exchange of information and control inputs between
TRB 1993) in support of Intelligent Transportation Systems the driver and the vehicle. This level of performance is at the
(ITS) have identified study of "Plain Old Driving" (POD) as a control interface. Most control activities, it is pointed out, are
fundamental research topic in ITS. For the purposes of a performed "automatically," with little conscious effort. In short,
transportation engineer interested in developing a molecular the control level of performance is skill based, in the approach
model of traffic flow in which the human in the vehicle or an to human performance and errors set forth by Jens Rasmussen as
individual human-vehicle comprises a unit of analysis, some presented in Human Error (Reason 1990).
important performance characteristics can be identified to aid in
the formulation, even if a comprehensive transfer function for the Once a person has learned the rudiments of control of the
driver has not yet been formulated. vehicle, the next level of human performance in the driver-
vehicle control hierarchy is the rules-based (Reason 1990)
This chapter will proceed to describe first the discrete guidance level as Rasmussen would say. The driver's main
components of performance, largely centered around activities "involve the maintenance of a safe speed and proper
neuromuscular and cognitive time lags that are fundamental path relative to roadway and traffic elements ." (Lunenfeld and
parameters in human performance. These topics include Alexander 1990) Guidance level inputs to the system are
perception-reaction time, control movement time, responses to dynamic speed and path responses to roadway geometrics,
the presentation of traffic control devices, responses to the hazards, traffic, and the physical environment. Information
movements of other vehicles, handling of hazards in the presented to the driver-vehicle system is from traffic control
roadway, and finally how different segments of the driving devices, delineation, traffic and other features of the
population may differ in performance. environment, continually changing as the vehicle moves along
the highway.
Next, the kind of control performance that underlies steering,
braking, and speed control (the primary control functions) will These two levels of vehicle control, control and guidance, are of
be described. Much research has focused on the development of paramount concern to modeling a corridor or facility. The third
adequate models of the tracking behavior fundamental to (and highest) level in which the driver acts as a supervisor apart,
steering, much less so for braking or for speed control. is navigation. Route planning and guidance while enroute, for
example, correlating directions from a map with guide signage
After fundamentals of open-loop and closed-loop vehicle control in a corridor, characterize the navigation level of performance.
are covered, applications of these principles to specific Rasmussen would call this level knowledge-based behavior.
maneuvers of interest to traffic flow modelers will be discussed. Knowledge based behavior will become increasingly more
Lane keeping, car following, overtaking, gap acceptance, lane important to traffic flow theorists as Intelligent Transportation
closures, stopping and intersection sight distances will also be Systems (ITS) mature. Little is currently known about how
discussed. To round out the chapter, a few other performance enroute diversion and route changes brought about by ITS
aspects of the driver-vehicle system will be covered, such as technology affect traffic flow, but much research is underway.
speed limit changes and distractions on the highway. This chapter will discuss driver performance in the conventional
highway system context, recognizing that emerging ITS
technology in the next ten years may radically change many
driver's roles as players in advanced transportation systems.
3-1
3. HUMAN FACTORS
At the control and guidance levels of operation, the driver of a driver-vehicle system from other vehicles, the roadway, and the
motor vehicle has gradually moved from a significant prime driver him/herself (acting at the navigation level of
mover, a supplier of forces to change the path of the vehicle, to performance).
an information processor in which strength is of little or no
consequence. The advent of power assists and automatic The fundamental display for the driver is the visual field as seen
transmissions in the 1940's, and cruise controls in the 1950's through the windshield, and the dynamics of changes to that field
moved the driver more to the status of a manager in the system. generated by the motion of the vehicle. The driver attends to
There are commercially available adaptive controls for severely selected parts of this input, as the field is interpreted as the visual
disabled drivers (Koppa 1990) which reduce the actual world. The driver as system manager as well as active system
movements and strength required of drivers to nearly the component "hovers" over the control level of performance.
vanishing point. The fundamental control tasks, however, Factors such as his or her experience, state of mind, and
remain the same. stressors (e.g., being on a crowded facility when
30 minutes late for a meeting) all impinge on the supervisory or
These tasks are well captured in a block diagram first developed monitoring level of performance, and directly or indirectly affect
many years ago by Weir (1976). This diagram, reproduced in the control level of performance. Rules and knowledge govern
Figure 3.1, forms the basis for the discussion of driver driver decision making and the second by second psychomotor
performance, both discrete and continuous. Inputs enter the activity of the driver. The actual control
Figure 3.1
Generalized Block Diagram of the Car-Driver-Roadway System.
3-2
3. HUMAN FACTORS
movements made by the driver couple with the vehicle control As will be discussed, a considerable amount of information is
at the interface of throttle, brake, and steering. The vehicle, in available for some of the lower blocks in this diagram, the ones
turn, as a dynamic physical process in its own right, is subject to associated with braking reactions, steering inputs, andvehic le
inputs from the road and the environment. The resolution of control dynamics. Far less is really known about the higher-
control dynamics and vehicle disturbance dynamics is the vehicle order functions that any driver knows are going on while he or
path. she drives.
3.2 Discrete Driver Performance

3.2.1 Perception-Response Time
Nothing in the physical universe happens instantaneously. Underlying the Hick-Hyman Law is the two-component concept:
Compared to some physical or chemical processes, the simplest part of the total time depends upon choice variables, and part is
human reaction to incoming information is very slow indeed. common to all reactions (the intercept). Other components can
Ever since the Dutch physiologist Donders started to speculate be postulated to intervene in the choice variable component,
in the mid 19th century about central processes involved in other than just the information content. Most of these models
choice and recognition reaction times, there have been numerous have then been chaining individual components that are
models of this process. The early 1950's saw Information presumably orthogonal or uncorrelated with one another.
Theory take a dominant role in experimental psychology. The Hooper and McGee (1983) postulate a very typical and plausible
linear equation model with such components for braking response time,
illustrated in Table 3.1.
RT = a + bH (3.1)
Each of these elements is derived from empirical data, and is in
Where: the 85th percentile estimate for that aspect of time lag. Because
it is doubtful that any driver would produce 85th percentile
RT = Reaction time, seconds values for each of the individual elements, 1.50 seconds
H = Estimate of transmitted information probably represents an extreme upper limit for a driver's
H = log2N , if N equiprobable alternatives perception-reaction time. This is an estimate for the simplest
a = Minimum reaction time for that modality kind of reaction time, with little or no decision making. The
b = Empirically derived slope, around 0.13 driver reacts to the input by lifting his or her foot from the
seconds (sec) for many performance situations accelerator and placing it on the brake pedal. But a number of
writers, for example Neuman (1989), have proposed perception-
that has come to be known as the Hick-Hyman "Law" expresses reaction times (PRT) for different types of roadways, ranging
a relationship between the number of alternatives that must be from 1.5 seconds for low-volume roadways to 3.0 seconds for
sorted out to decide on a response and the total reaction time, urban freeways. There are more things happening, and more
that is, that lag in time between detection of an input (stimulus) decisions to be made per unit block of time on a busy urban
and the start of initiation of a control or other response. If the facility than on a rural county road. Each of those added factors
time for the response itself is also included, then the total lag is increase the PRT. McGee (1989) has similarly proposed
termed "response time." Often, the terms "reaction time" and different values of PRT as a function of design speed. These
"response time" are used interchangeably, but one (reaction) is estimates, like those in Table 3.1, typically include the time for
always a part of the other (response). the driver to move his or her foot from the accelerator to the
brake pedal for brake application.
3-3
3. HUMAN FACTORS
Table 3.1
Hooper-McGee Chaining Model of Perception-Response Time
Time Cumulative Time

Component (sec) (sec)
1) Perception
Latency 0.31 0.31
Eye Movement 0.09 0.4
Fixation 0.2 1
Recognition 0.5 1.5
2) Initiating Brake 1.24 2.74

Application
Any statistical treatment of empirically obtained PRT's should skew, because there cannot be such a thing as a negative reaction
take into account a fundamental if not always vitally important time, if the time starts with onset of the signal with no
fact: the times cannot be distributed according to the normal or anticipation by the driver. Taoka (1989) has suggested an
gaussian probability course. Figure 3.2 illustrates the actual adjustment to be applied to PRT data to correct for the non-
shape of the distribution. The distribution has a marked positive normality, when sample sizes are "large" --50 or greater.
Figure 3.2
Lognormal Distribution of Perception-Reaction Time.
3-4
3. HUMAN FACTORS
The log-normal probability density function is widely used in

quality control engineering and other applications in which
LN(t)
values of the observed variable, t, are constrained to values equal
Z (3.6)
!
to or greater than zero, but may take on extreme positive values,
exactly the situation that obtains in considering PRT. In such
situations, the natural logarithm of such data may be assumed to
approach the normal or gaussian distribution. Probabilities
associated with the log-normal distribution can thus be Therefore, the value of LN(t) for such percentile levels as 0.50
determined by the use of standard-score tables. Ang and Tang (the median), the 85th, 95th, and 99th can be obtained by
(1975) express the log-normal probability density function f(t) substituting in Equation 3.6 the appropriate Z score of 0.00,
as follows: 1.04, 1.65, and 2.33 for Z and then solving for t. Converting
data to log-normal approximations of percentile values should be
1 LN(t) 2 considered when the number of observations is reasonably large,
f(t)
exp (3.2) over 50 or more, to obtain a better fit. Smaller data sets will
2 !t !
benefit more from a tolerance interval approach to approximate
percentiles (Odeh 1980).
where the two parameters that define the shape of the A very recent literature review by Lerner and his associates
distribution are and !. It can be shown that these two (1995) includes a summary of brake PRT (including brake
parameters are related to the mean and the standard deviation of onset) from a wide variety of studies. Two types of response
a sample of data such as PRT as follows: situation were summarized: (1) The driver does not know when
or even if the stimulus for braking will occur, i.e., he or she is
surprised, something like a real-world occurrence on the
)2
!2
LN 1 (3.3) highway; and (2) the driver is aware that the signal to brake will
μ2 occur, and the only question is when. The Lerner et al. (1995)
composite data were converted by this writer to a log-normal
transformation to produce the accompanying Table 3.2.
The parameter is related to the median of the distribution being
described by the simple relationship of the natural logarithm of Sixteen studies of braking PRT form the basis for Table 3.2.
the median. It can also be shown that the value of the standard Note that the 95th percentile value for a "surprise" PRT (2.45
normal variate (equal to probability) is related to these seconds) is very close to the AASHTO estimate of 2.5 seconds
parameters as shown in the following equation: which is used for all highway situations in estimating both
stopping sight distance and other kinds of sight distance (Lerner
et al. 1995).
μ

LN (3.4)
1)2/μ2 In a very widely quoted study by Johansson and Rumar (1971),
drivers were waylaid and asked to brake very briefly if they
heard a horn at the side of the highway in the next 10 kilometers.
Mean PRT for 322 drivers in this situation was 0.75 seconds
with an SD of 0.28 seconds. Applying the Taoka conversion to
LN(t)
0
0.5, 0.85, etc. (3.5) the log normal distribution yields:
!
50th percentile PRT = 0.84 sec
and the standard score associated with that value is given by: 95th percentile PRT = 1.27 sec
3-5
3. HUMAN FACTORS
Table 3.2
Brake PRT - Log Normal Transformation
"Surprise" "Expected"
Mean 1.31 (sec) 0.54

Standard Dev 0.61 0.1
0.17 (no unit) -0.63 (no unit)
! 0.44 (no unit) 0.18 (no unit)
50th percentile 1.18 0.53

In very recent work by Fambro et al. (1994) volunteer drivers in Additional runs were made with other drivers in their own cars
two age groups (Older: 55 and up; and Young: 18 to 25) were equipped with the same instrumentation. Nine of the 12 drivers
suddenly presented with a barrier that sprang up from a slot in made stopping maneuvers in response to the emergence of the
the pavement in their path, with no previous instruction. They barrier. The results are given in Table 3.3 as Case 2. In an
were driving a test vehicle on a closed course. Not all 26 drivers attempt (Case 3) to approximate real-world driving conditions,
hit the brakes in response to this breakaway barrier. The PRT's Fambro et al. (1994) equipped 12 driver's own vehicles with
of the 22 who did are summarized in Table 3.3 (Case 1). None instrumentation. They were asked to drive a two-lane undivided
of the age differences were statistically significant. secondary road ostensibly to evaluate the drivability of the road.
Table 3.3
Summary of PRT to Emergence of Barrier or Obstacle
Case 1. Closed Course, Test Vehicle

12 Older: Mean = 0.82 sec; SD = 0.16 sec
10 Young: Mean = 0.82 sec; SD = 0.20 sec
Case 2. Closed Course, Own Vehicle
Case 3. Open Road, Own Vehicle
3-6
3. HUMAN FACTORS
A braking incident was staged at some point during this test event to an expected event ranges from 1.35 to 1.80 sec,
drive. A barrel suddenly rolled out of the back of a pickup consistent with Johansson and Rumar (1971). Note, however,
parked at the side of the road as he or she drove by. The barrel that one out of 12 of the drivers in the open road barrel study
was snubbed to prevent it from actually intersecting the driver's (Case 3) did not appear to notice the hazard at all. Thirty
path, but the driver did not know this. The PRT's obtained by percent of the drivers confronted by the artificial barrier under
this ruse are summarized in Table 3.4. One driver failed to closed-course conditions also did not respond appropriately.
notice the barrel, or at least made no attempt to stop or avoid it. How generalizable these percentages are to the driver population
remains an open question that requires more research. For
Since the sample sizes in these last two studies were small, it analysis purposes, the values in Table 3.4 can be used to
was considered prudent to apply statistical tolerance intervals to approximate the driver PRT envelope for an unexpected event.
these data in order to estimate proportions of the driving PRT's for expected events, e.g., braking in a queue in heavy
population that might exhibit such performance, rather than traffic, would range from 1.06 to 1.41 second, according to the
using the Taoka conversion. One-sided tolerance tables ratios given above (99th percentile).
published by Odeh (1980) were used to estimate the percentage
of drivers who would respond in a given time or shorter, based These estimates may not adequately characterize PRT under
on these findings. These estimates are given in Table 3.4 (95 conditions of complete surprise, i.e., when expectancies are
percent confidence level), with PRT for older and younger greatly violated (Lunenfeld and Alexander 1990). Detection
drivers combined. times may be greatly increased if, for example, an unlighted
vehicle is suddenly encountered in a traffic lane in the dark, to
The same researchers also conducted studies of driver response say nothing of a cow or a refrigerator.
to expected obstacles. The ratio of PRT to a totally unexpected
Table 3.4
Percentile Estimates of PRT to an Unexpected Object
Case 1 Case 2 Case 3

Test Vehicle Own Vehicle Own Vehicle
Percentile Closed Course Closed Course Open Road
50th 0.82 sec 1.09 sec 1.11 sec
75th 1.02 sec 1.54 sec 1.40 sec
90th 1.15 sec 1.81 sec 1.57 sec
95th 1.23 sec 1.98 sec 1.68 sec
99th 1.39 sec 2.31 sec 1.90 sec
Adapted from Fambro et al. (1994).
3-7
3. HUMAN FACTORS
3.3 Control Movement Time

Once the lag associated with perception and then reaction has is the "Index of Difficulty" of the movement, in binary units, thus
ensued and the driver just begins to move his or her foot (or linking this simple relationship with the Hick-Hyman equation
hand, depending upon the control input to be effected), the discussed previously in Section 3.3.1.
amount of time required to make that movement may be of
interest. Such control inputs are overt motions of an appendage Other researchers, as summarized by Berman (1994), soon
of the human body, with attendant inertia and muscle fiber found that certain control movements could not be easily
latencies that come into play once the efferent nervous impulses modeled by Fitts' Law. Accurate tapping responses less than
arrives from the central nervous system. 180 msec were not included. Movements which are short and
quick also appear to be preplanned, or "programmed," and are
3.3.1 Braking Inputs open-loop. Such movements, usually not involving visual
feedback, came to be modeled by a variant of Fitts' Law:
As discussed in Section 3.3.1 above, a driver's braking response
is composed of two parts, prior to the actual braking of the
MT
a b A (3.9)
vehicle: the perception-reaction time (PRT) and immediately
following, movement time (MT ).
Movement time for any sort of response was first modeled by in which the width of the target control (W ) plays no part.
Fitts in 1954. The simple relationship among the range or
amplitude of movement, size of the control at which the control Almost all such research was devoted to hand or arm responses.
movement terminates, and basic information about the minimum In 1975, Drury was one of the first researchers to test the
"twitch" possible for a control movement has long been known applicability of Fitts' Law and its variants to foot and leg
as "Fitts' Law." movements. He found a remarkably high association for fitting
foot tapping performance to Fitts' Law. Apparently, all
appendages of the human body can be modeled by Fitts' Law or
2A
MT
a b Log2 (3.7) one of its variants, with an appropriate adjustment of a and b, the
W empirically derived parameters. Parameters a and b are
sensitive to age, condition of the driver, and circumstances such
as degree of workload, perceived hazard or time stress, and pre-
where, programming by the driver.
a = minimum response time lag, no movement
b = slope, empirically determined, different for In a study of pedal separation and vertical spacing between the
each limb planes of the accelerator and brake pedals, Brackett and Koppa
A = amplitude of movement, i.e., the distance (1988) found separations of 10 to 15 centimeters (cm), with little
from starting point to end point or no difference in vertical spacing, produced
W = width of control device (in direction of control movement in the range of 0.15 to 0.17 sec. Raising the
movement) brake pedal more than 5 cm above the accelerator lengthened
this time significantly. If pedal separation ( = A in Fitts' Law)
The term was varied, holding pedal size constant, the mean MT was 0.22
sec, with a standard deviation of 0.20 sec.
2A
Log2 (3.8)
W In 1991, Hoffman put together much of the extant literature and
conducted studies of his own. He found that the Index of
Difficulty was sufficiently low (<1.5) for all pedal placements
found on passenger motor vehicles that visual control was
3-8
3. HUMAN FACTORS
unnecessary for accurate movement, i.e., movements were other relationship between these two times. A recent analysis by
ballistic in nature. MT was found to be greatly influenced by the writer yielded a Pearson Product-Moment Correlation
vertical separation of the pedals, but comparatively little by Coefficient (r) value of 0.17 between these two quantities in a
changes in A, presumably because the movements were ballistic braking maneuver to a completely unexpected object in the path
or open-loop and thus not correctable during the course of the of the vehicle (based on 21 subjects). Total PRT's as presented
movement. MT was lowest at 0.20 sec with no vertical in Section 3.3.1 should be used for discrete braking control
separation, and rose to 0.26 sec if the vertical separation (brake movement time estimates; for other situations, the modeler could
pedal higher than accelerator) was as much as 7 cm. A very use the tolerance levels in Table 3.5 for MT, chaining them after
recent study by Berman (1994) tends to confirm these general an estimate of perception (including decision) and reaction time
MT evaluations, but adds some additional support for a ballistic for the situation under study (95 percent confidence level). See
model in which amplitude A does make a difference. Section 3.14 for a discussion on how to combine these estimates.
Her MT findings for a displacement of (original) 16.5 cm and

(extended) 24.0 cm, or change of 7.5 cm can be summarized as 3.3.2 Steering Response Times
follows:
Original pedal Extended pedal Summala (1981) covertly studied driver responses to the sudden
Mean 0.20 sec 0.29 sec opening of a car door in their path of travel. By "covert" is
Standard Deviation 0.05 sec 0.07 sec meant the drivers had no idea they were being observed or were
95 percent participating in the study. This researcher found that neither the
tolerance level 0.32 sec 0.45 sec latency nor the amount of deviation from the pre-event pathway
99 percent was dependent upon the car's prior position with respect to the
tolerance level 0.36 sec 0.51 sec opening car door. Drivers responded with a ballistic "jerk" of
the steering wheel. The mean response latency for these Finnish
The relationship between perception-reaction time and MT has drivers was 1.5 sec, and reached the half-way point of maximum
been shown to be very weak to nonexistent. That is, a long displacement from the original path in about 2.5 sec. The
reaction time does not necessarily predict a long MT, or any
Table 3.5
Movement Time Estimates
Source N Mean 75th 90th 95th 99th

(Std) Sec Sec Sec Sec
Brackett (Brackett and Koppa 1988) 24 0.22 (0.20) 0.44 0.59 0.68 0.86
Hoffman (1991) 18 0.26 (0.20) 0.50 0.66 0.84 1.06
Berman (1994) 24 0.20 (0.05) 0.26 0.29 0.32 0.36
3-9
3. HUMAN FACTORS
3.4 Response Distances and Times to Traffic Control Devices

The driving task is overwhelmingly visual in nature; external very rich with data related to target detection in complex visual
information coming through the windshield constitutes nearly all environments, and a TCD's target value also depends upon the
the information processed. A major input to the driver which driver's predilection to look for and use such devices.
influences his or her path and thus is important to traffic flow
theorists is traffic information imparted by traffic control devices
(TCD). The major issues concerned with TCD are all related to 3.4.1 Traffic Signal Change
distances at which they may be (1) detected as objects in the
visual field; (2) recognized as traffic control devices: signs, From the standpoint of traffic flow theory and modeling, a major
signals, delineators, and barricades; (3) legible or identifiable so concern is at the stage of legibility or identification and a
that they may be comprehended and acted upon. Figure 3.3 combination of "read" and "understand" in the diagram in Figure
depicts a conceptual model for TCD information processing, and 3.3. One of the most basic concerns is driver response or lag to
the many variables which affect it. The research literature is
Figure 3.3
A Model of Traffic Control Device Information Processing.
3 - 10
3. HUMAN FACTORS
changing traffic signals. Chang et al. (1985) found through Mean PRT to signal change = 1.30 sec
covert observations at signalized intersections that drivers 85th percentile PRT = 1.50 sec
response lag to signal change (time of change to onset of brake 95th percentile PRT = 2.50 sec
lamps) averaged 1.3 sec, with the 85th percentile PRT estimated 99th percentile PRT = 2.80 sec
at 1.9 sec and the 95th percentile at 2.5 sec. This PRT to signal
change is somewhat inelastic with respect to distance from the If the driver is stopped at a signal, and a straight-ahead maneuver
traffic signal at which the signal state changed. The mean PRT is planned, PRT would be consistent with those values given in
(at 64 kilometers per hour (km/h)) varied by only 0.20 sec within Section 3.3.1. If complex maneuvers occur after signal change
a distance of 15 meters (m) and by only 0.40 sec within 46 m. (e.g., left turn yield to oncoming traffic), the Hick-Hyman Law
(Section 3.2.1) could be used with the y intercept being the basic
Wortman and Matthias (1983) found similar results to Chang et PRT to onset of the traffic signal change. Considerations related
al. (1985) with a mean PRT of 1.30 sec, and a 85th percentile to intersection sight distances and gap acceptance make such
PRT of 1.5 sec. Using tolerance estimates based on their predictions rather difficult to make without empirical validation.
sample size, (95 percent confidence level) the 95th percentile These considerations will be discussed in Section 3.15.
PRT was 2.34 sec, and the 99th percentile PRT was 2.77 sec.
They found very little relationship between the distance from the
intersection and either PRT or approach speed (r2 = 0.08). So 3.4.2 Sign Visibility and Legibility
the two study findings are in generally good agreement, and the
following estimates may be used for driver response to signal The psychophysical limits to legibility (alpha-numeric) and
change: identification (symbolic) sign legends are the resolving power of
Table 3.6
Visual Acuity and Letter Sizes
Snellen Acuity Visual angle of letter or symbol Legibility Index
SI (English) 'of arc radians m/cm

6/3 (20/10) 2.5 0.00073 13.7
6/6 (20/20) 5 0.00145 6.9
6/9 (20/30) 7.5 0.00218 4.6
6/12 (20/40) 10 0.00291 3.4
6/15 (20/50) 12.5 0.00364 2.7
6/18 (20/60) 15 0.00436 2.3
3 - 11
3. HUMAN FACTORS
the visual perception system, the effects of the optical train confirmed these earlier findings, and also notes that extreme
leading to presentation of an image on the retina of the eye, variability exists from trial to trial for the same observer on a
neural processing of that image, and further processing by the given sign's recognition distance. Presumably, word signs would
brain. Table 3.6 summarizes visual acuity in terms of visual manifest as much or even more variability. Complex, fine detail
angles and legibility indices. signs such as Bicycle Crossing (MUTCD W11-1) were
observed to have coefficients of variation between subjects of 43
The exact formula for calculating visual angle is percent. Coefficient of Variation (CV) is simply:
¬
2 arctan L
2D
(3.10)
CV = 100 (Std Deviation/Mean) (3.11)
In contrast, very simple symbol signs such as T-Junction

(MUTCD W2-4) had a CV of 28 percent. Within subject
where, L = diameter of the target (letter or symbol) variation (from trial to trial) on the same symbol sign is
D = distance from eye to target in the same units summarized in Table 3.7.
All things being equal, two objects that subtend the same visual Before any reliable predictions can be made about legibility or
angle will elicit the same response from a human observer, recognition distances of a given sign, Greene (1994) found that
regardless of their actual sizes and distances. In Table 3.6 the six or more trials under controlled conditions must be made,
Snellen eye chart visual acuity ratings are related to the size of either in the laboratory or under field conditions. Greene (1994)
objects in terms of visual arc, radians (equivalent for small sizes found percent differences between high-fidelity laboratory and
to the tangent of the visual arc) and legibility indices. Standard field recognition distances to range from 3 to 21 percent,
transportation engineering resources such as the Traffic Control depending upon sign complexity. These differences consistent
Devices Handbook (FHWA 1983) are based upon these with most researchers, were all in the direction of laboratory
fundamental facts about visual performance, but it should be distances being greater than actual distances; the laboratory
clearly recognized that it is very misleading to extrapolate tends to overestimate field legibility distances. Variability in
directly from letter or symbol legibility/recognition sizes to sign legibility distances, however, is as great in the laboratory as it is
perceptual distances, especially for word signs. There are other under field trials.
expectancy cues available to the driver, word length, layout, etc.
that can lead to performance better than straight visual angle With respect to visual angle required for recognition, Greene
computations would suggest. Jacobs, Johnston, and Cole (1975) found, for example, that the Deer Crossing at the mean
also point out an elementary fact that 27 to 30 percent of the recognition distance had a mean visual angle of 0.00193 radian,
driving population cannot meet a 6/6 (20/20) criterion. Most or 6.6 minutes of arc. A more complex, fine detail sign such as
states in the U.S. have a 6/12 (20/40) static acuity criterion for Bicycle Crossing required a mean visual angle of 0.00345 radian
unrestricted licensure, and accept 6/18 (20/60) for restricted or 11.8 minutes of arc to become recognizable.
(daytime, usually) licensure. Such tests in driver license offices
are subject to error, and examiners tend to be very lenient. With these considerations in mind, here is the best
Night-time static visual acuity tends to be at least one Snellen recommendation that this writer can make. For the purposes of
line worse than daytime, and much worse for older drivers (to be predicting driver comprehension of signs and other devices that
discussed in Section 3.8). require interpretation of words or symbols use the data in Table
3.6 as "best case," with actual performance expected to be
Jacobs, et al. also point out that the sign size for 95th percentile somewhat to much worse (i.e., requiring closer distances for a
recognition or legibility is 1.7 times the size for 50th percentile given size of character or symbol). The best visual acuity that
performance. There is also a pervasive notion in the research can be expected of drivers under optimum contrast conditions so
that letter sign legibility distances are half symbol sign far as static acuity is concerned would be 6/15 (20/50) when the
recognition distances, when drivers are very familiar with the sizable numbers of older drivers is considered [13 percent in
symbol (Greene 1994). Greene (1994), in a very recent study, 1990 were 65 or older (O'Leary and Atkins 1993)].
3 - 12
3. HUMAN FACTORS
Table 3.7
Within Subject Variation for Sign Legibility
Young Drivers Older Drivers
Sign Min CV Max CV Min CV Max CV
WG-3 2 Way Traffic 3.9 21.9 8.9 26.7
W11-1 Bicycle Cross 6.7 37.0 5.5 39.4
W2-1 Crossroad 5.2 16.3 2.0 28.6
W11-3 Deer Cross 5.4 21.3 5.4 49.2
W8-5 Slippery 7.7 33.4 15.9 44.1
W2-5 T-Junction 5.6 24.6 4.9 28.7
3.4.3 Real-Time Displays and Signs "Research...has indicated that a minimum exposure time of one
second per short word (four to eight characters) (exclusive of
With the advent of Intelligent Transportation Systems (ITS), prepositions and other similar connectors) or two seconds per
traffic flow modelers must consider the effects of changeable unit of information, whichever is largest, should be used for
message signs on driver performance in traffic streams. unfamiliar drivers. On a sign having 12 to 16 characters per
Depending on the design of such signs, visual performance to line, this minimum exposure time will be two seconds per line."
them may not differ significantly from conventional signage. "Exposure time" can also be interpreted as "reading time" and so
Signs with active (lamp or fiber optic) elements may not yield used in estimating how long drivers will take to read and
the legibility distances associated with static signage, because comprehend a sign with a given message.
Federal Highway Administration, notably the definitive manual
by Dudek (1990). Suppose a sign reads:
Traffic Conditions
3.4.4 Reading Time Allowance Next 2 Miles
Disabled Vehicle on I-77
For signs that cannot be comprehended in one glance, i.e., word Use I-77 Bypass Next Exit
message signs, allowance must be made for reading the
information and then deciding what to do, before a driver in Drivers not familiar with such a sign ("worst case," but able to
traffic will begin to maneuver in response to the information. read the sign) could take at least 8 seconds and according to the
Reading speed is affected by a host of factors (Boff and Lincoln Dudek formula above up to 12 seconds to process this
1988) such as the type of text, number of words, sentence information and begin to respond. In Dudek's 1990 study, 85
structure, information order, whatever else the driver is doing, percent of drivers familiar with similar signs read this 13-word
the purpose of reading, and the method of presentation. The message (excluding prepositions) with 6 message units in 6.7
USAF resource (Boff and Lincoln 1988) has a great deal of seconds. The formulas in the literature properly tend to be
general information on various aspects of reading sign material. conservative.
For purposes of traffic flow modeling, however, a general rule of
thumb may suffice. This can be found in Dudek (1990):
3 - 13
3. HUMAN FACTORS
3.5 Response to Other Vehicle Dynamics

Vehicles in a traffic stream are discrete elements with motion ahead) is the symmetrical magnification of a form or texture in
characteristics loosely coupled with each other via the driver's the field of view. Visual angle transitions from a near-linear to
processing of information and making control inputs. Effects of a geometric change in magnitude as an object approaches at
changes in speed or acceleration of other elements as perceived constant velocity, as Figure 3.4 depicts for a motor vehicle
and acted on by the driver of any given element are of interest. approaching at a delta speed of 88 km/h. As the rate of change
Two situations appear relevant: (1) the vehicle ahead and (2) the of visual angle becomes geometric, the perceptual system
vehicle alongside (in the periphery). triggers a warning that an object is going to collide with the
observer, or, conversely, that the object is pulling away from the
observer. This phenomenon is called looming. If the rate of
3.5.1 The Vehicle Ahead change of visual angle is irregular, that is information to the
perceptual system that the object in motion is moving at a
Consideration of the vehicle ahead has its basis in thresholds for changing velocity (Schiff 1980). Sekuler and Blake (1990)
detection of radial motion (Schiff 1980). Radial motion is report evidence that actual looming detectors exist in the human
change in the apparent size of a target. The minimum condition visual system. The relative change in visual angle is roughly
for perceiving radial motion of an object (such as a vehicle equal to the reciprocal of "time-to-go" (time to impact), a special
Figure 3.4
Looming as a Function of Distance from Object.
3 - 14
3. HUMAN FACTORS
case of the well-known Weber fraction, S = I/I, the magnitude respond to changes in their headway, or the change in angular
of a stimulus is directly related to a change in physical energy but size of the vehicle ahead, and use that as a cue to determine the
inversely related to the initial level of energy in the stimulus. speed that they should adopt when following another vehicle."
Human visual perception of acceleration (as such) of an object

in motion is very gross and inaccurate; it is very difficult for a
driver to discriminate acceleration from constant velocity unless 3.5.2 The Vehicle Alongside
the object is observed for a relatively long period of time - 10 or
15 sec (Boff and Lincoln 1988). Motion detection in peripheral vision is generally less acute than
in foveal (straight-ahead) vision (Boff and Lincoln 1988), in that
The delta speed threshold for detection of oncoming collision or a greater relative velocity is necessary for a driver "looking out
pull-away has been studied in collision-avoidance research. of the corner of his eye" to detect that speed
Mortimer (1988) estimates that drivers can detect a change in change than if he or she is looking to the side at the subject
distance between the vehicle they are driving and the one in front vehicle in the next lane. On the other hand, peripheral vision is
when it has varied by approximately 12 percent. If a driver were very blurred and motion is a much more salient cue than a
following a car ahead at a distance of 30 m, at a change of 3.7 m stationary target is. A stationary object in the periphery (such as
the driver would become aware that distance is decreasing or a neighboring vehicle exactly keeping pace with the driver's
increasing, i.e., a change in relative velocity. Mortimer notes vehicle) tends to disappear for all intents and purposes unless it
that the major cue is rate of change in visual angle. This moves with respect to the viewer against a patterned background.
threshold was estimated in one study as 0.0035 radians/sec. Then that movement will be detected. Relative motion in the
periphery also tends to look slower than the same movement as
This would suggest that a change of distance of 12 percent in 5.6 seen using fovea vision. Radial motion (car alongside swerving
seconds or less would trigger a perception of approach or pulling toward or away from the driver) detection presumably would
away. Mortimer concludes that "...unless the relative velocity follow the same pattern as the vehicle ahead case, but no study
between two vehicles becomes quite high, the drivers will concerned with measuring this threshold directly was found.
3.6 Obstacle and Hazard Detection, Recognition, and Identification

Drivers on a highway can be confronted by a number of different roadway were unexpectedly encountered by drivers on a closed
situations which dictate either evasive maneuvers or stopping course. Six objects, a 1 x 4 board, a black toy dog, a white toy
maneuvers. Perception-response time (PRT) to such encounters dog, a tire tread, a tree limb with leaves, and a hay bale were
have already been discussed in Section 3.3.1. But before a placed in the driver's way. Both detection and recognition
maneuver can be initiated, the object or hazard must first be distances were recorded. Average visual angles of detection for
detected and then recognized as a hazard. The basic these various objects varied from the black dog at 1.8 minimum
considerations are not greatly different than those discussed under of arc to 4.9 min of arc for the tree limb. Table 3.8 summarizes
driver responses to traffic control devices (Section 3.5), but some the detection findings of this study.
specific findings on roadway obstacles and hazards will also be
discussed. At the 95 percent level of confidence, it can be said from these
findings that an object subtending a little less than 5 minutes of
arc will be detected by all but 1 percent of drivers under daylight
3.6.1 Obstacle and Hazard Detection conditions provided they are looking in the object's direction.
Since visual acuity declines by as much as two Snellen lines after
Picha (1992) conducted an object detection study in which nightfall, to be dtected such targets with similar contrast would
representative obstacles or objects that might be found on a
3 - 15
3. HUMAN FACTORS
Table 3.8
Object Detection Visual Angles (Daytime)
(Minutes of Arc)
Tolerance, 95th confidence

Object Mean STD 95th 99th
1" x 4" Board, 24" x 1"* 2.47 1.21 5.22 6.26
Black toy dog, 6" x 6" 1.81 0.37 2.61 2.91
White toy dog, 6" x 6" 2.13 0.87 4.10 4.84
Tire tread, 8" x 18" 2.15 0.38 2.95 3.26
Tree Branch, 18" x 12" 4.91 1.27 7.63 8.67
Hay bale, 48" x 18" 4.50 1.28 7.22 8.26
All Targets 3.10 0.57 4.30 4.76
*frontal viewing plan dimensions
have to subtend somewhere around 2.5 times the visual angle that 15 cm or less in height very seldom are causal factors in accidents
they would at detection under daylight conditions. (Kroemer et al. 1994).
The majority of objects encountered on the highway that

constitute hazard and thus trigger avoidance maneuvers are larger
3.6.2 Obstacle and Hazard Recognition than 60 cm in height. Where it may be of interest to establish a
and Identification visual angle for an object to be discriminated as a hazard or non-
hazard, such decisions require visual angles on the order of at
Once the driver has detected an object in his or her path, the next least the visual angles identified in Section 3.4.2 for letter or
job is to: (1) decide if the object, whatever it is, is a potential symbol recognition, i.e., about 15 minutes of arc to take in 99
hazard, this is the recognition stage, followed by (2) the percent of the driver population. It would be useful to reflect that
identification stage, even closer, at which a driver actually can tell the full moon subtends 30 minutes of arc, to give the reader an
what the object is. If an object (assume it is stationary) is small intuitive feel for what the minimum visual angle might be for
enough to pass under the vehicle and between the wheels, it object recognition. At a distance somewhat greater than this, the
doesn't matter very much what it is. So the first estimate is driver decides if an object is sizable enough to constitute a hazard,
primarily of size of the object. If the decision is made that the largely based upon roadway lane width size comparisons and the
object is too large to pass under the vehicle, then either evasive size of the object with respect to other familiar roadside objects
action or a braking maneuver must be decided upon. Objects (such as mailboxes, bridge rails). Such judgements improve if
the object is identified.
3 - 16
3. HUMAN FACTORS
3.7 Individual Differences in Driver Performance

In psychological circles, variability among people, especially that visual and cognitive changes affecting driver performance will
associated with variables such as gender, age, socio-economic be discussed in the following paragraphs.
levels, education, state of health, ethnicity, etc., goes by the name
"individual differences." Only a few such variables are of interest CHANGES IN VISUAL PERCEPTION
to traffic flow modeling. These are the variables which directly
affect the path and velocity the driven vehicle follows in a given Loss of Visual Acuity (static) - Fifteen to 25 percent of the
time in the operational environment. Other driver characteristics population 65 and older manifest visual acuities (Snellen) of less
which may be of interest to the reader may be found in the NHTSA than 20/50 corrected, owing to senile macular degeneration
Driver Performance Data book (1987, 1994). (Marmor 1982). Peripheral vision is relatively unaffected,
although a gradual narrowing of the visual field from 170 degrees
to 140 degrees or less is attributable to anatomical changes (eyes
3.7.1 Gender become more sunk in the head). Static visual acuity among
drivers is not highly associated with accident experience and is
Kroemer, Kroemer, and Kroemer-Ebert (1994) summarize probably not a very significant factor in discerning path guidance
relevant gender differences as minimal to none. Fine finger devices and markings.
dexterity and color perception are areas in which women perform
better than men, but men have an advantage in speed. Reaction Light Losses and Scattering in Optic Train - There is some
time tends to be slightly longer for women than for men the recent evidence (Ordy et al. 1982) that the scotopic (night) vision system
popular book and PBS series, Brain Sex (Moir and Jessel 1991) ages faster than the photopic (daylight) system does. In addition,
has some fascinating insights into why this might be so. This scatter and absorption by the stiff, yellowed, and possibly
difference is statistically but not practically significant. For the cateracted crystalline lens of the eye accounts for much less light
purpose of traffic flow analysis, performance differences between hitting the degraded retina. The pupil also becomes stiffer with
men and women may be ignored. age, and dilates less for a given amount of light impingement
(which considering that the mechanism of pupillary size is in part
driven by the amount of light falling on the retina suggests actual
physical atrophy of the pupil--senile myosis). There is also more
3.7.2 Age
matter in suspension in the vitreous humor of the aged eye than
exists in the younger eye. The upshot is that only 30 percent of
Research on the older driver has been increasing at an exponential
the light under daytime conditions that gets to the retina in a 20
rate, as was noted in the recent state-of-the-art summary by the
year old gets to the retina of a 60 year old. This becomes much
Transportation Research Board (TRB 1988). Although a number
worse at night (as little as 1/16), and is exacerbated by the
of aspects of human performance related to driving change with
scattering effect of the optic train. Points of bright light are
the passage of years, such as response time, channel capacity and
surrounded by halos that effectively obscure less bright objects
processing time needed for decision making, movement ranges
in their near proximity. Blackwell and Blackwell (1971)
and times, most of these are extremely variable, i.e., age is a poor
estimated that, because of these changes, a given level of contrast
predictor of performance. This was not so for visual perception.
of an object has to be increased by a factor of anywhere from 1.17
Although there are exceptions, for the most part visual
to 2.51 for a 70 year old person to see it, as compared to a 30 year
performance becomes progressively poorer with age, a process
old.
which accelerates somewhere in the fifth decade of life.
Glare Recovery - It is worth noting that a 55 year old person
Some of these changes are attributable to optical and
requires more than 8 times the period of time to recover from
physiological conditions in the aging eye, while others relate to
glare if dark adapted than a 16 year old does (Fox 1989). An
changes in neural processing of the image formed on the retina.
older driver who does not use the strategy to look to the right and
There are other cognitive changes which are also central to
shield his or her macular vision from oncoming headlamp glare
understanding performance differences as drivers age. Both
3 - 17
3. HUMAN FACTORS
is literally driving blind for many seconds after exposure. As lived in before becoming "older drivers," and they have driven
described above, scatter in the optic train makes discerning any under modern conditions and the urban environment since their
marking or traffic control device difficult to impossible. The slow teens. Most of them have had classes in driver education and
re-adaptation to mesopic levels of lighting is well-documented. defensive driving. They will likely continue driving on a routine
basis until almost the end of their natural lives, which will be
Figure/Ground Discrimination - Perceptual style changes with happening at an ever advancing age. The cognitive trends briefly
age, and many older drivers miss important cues, especially under discussed above are very variable in incidence and in their actual
higher workloads (Fox 1989). This means drivers may miss a effect on driving performance. The future older driver may well
significant guideline or marker under unfamiliar driving exhibit much less decline in many of these performance areas in
conditions, because they fail to discriminate the object from its which central processes are dominant.
background, either during the day or at night.
CHANGES IN COGNITIVE PERFORMANCE 3.7.3 Driver Impairment

Information Filtering Mechanisms - Older drivers reportedly Drugs - Alcohol abuse in isolation and combination with other
experience problems in ignoring irrelevant information and drugs, legal or otherwise, has a generally deleterious effect on
correctly identifying meaningful cues (McPherson et al. 1988). performance (Hulbert 1988; Smiley 1974). Performance
Drivers may not be able to discriminate actual delineation or differences are in greater variability for any given driver, and in
signage from roadside advertising or faraway lights, for example. generally lengthened reaction times and cognitive processing
Work zone traffic control devices and markings that are meant times. Paradoxically, some drug combinations can improve such
to override the pre-work TCD's may be missed. performance on certain individuals at certain times. The only
drug incidence which is sufficiently large to merit consideration
Forced Pacing under Highway Conditions - In tasks that in traffic flow theory is alcohol.
require fine control, steadiness, and rapid decisions, forced paced
tasks under stressful conditions may disrupt the performance of Although incidence of alcohol involvement in accidents has been
older drivers. They attempt to compensate for this by slowing researched for many years, and has been found to be substantial,
down. Older people drive better when they can control their own very little is known about incidence and levels of impairment in
pace (McPherson et al. 1988). To the traffic flow theorist, a the driving population, other than it must also be substantial.
sizable proportion of older drivers in a traffic stream may result Because these drivers are impaired, they are over-represented in
in vehicles that lag behind and obstruct the flow. accidents. Price (1988) cites estimates that 92 percent of the
adult population of the U.S. use alcohol, and perhaps 11 percent
Central vs. Peripheral Processes - Older driver safety problems of the total adult population (20-70 years of age) have alcohol
relate to tasks that are heavily dependent on central processing. abuse problems. Of the 11 percent who are problem drinkers,
These tasks involve responses to traffic or to roadway conditions seven percent are men, four percent women. The incidence of
(emphasis added) (McPherson et al. 1988). problem drinking drops with age, as might be expected. Effects
on performance as a function of blood alcohol concentration
The Elderly Driver of the Past or Even of Today is Not the (BAC) are well-summarized in Price, but are too voluminous to
Older Driver of the Future - The cohort of drivers who will be be reproduced here. Price also summarizes effects of other drugs
65 in the year 2000, which is less than five years from now, were such as cocaine, marijuana, etc. Excellent sources for more
born in the 1930's. Unlike the subjects of gerontology studies information on alcohol and driving can be found in a
done just a few years ago featuring people who came of driving Transportation Research Board Special Report (216).
age in the 1920's or even before, when far fewer people had cars
and traffic was sparse, the old of tomorrow started driving in the Medical Conditions - Disabled people who drive represent a
1940's and after. They are and will be more affluent, better small but growing portion of the population as technology
educated, in better health, resident in the same communities they advances in the field of adaptive equipment. Performance
3 - 18
3. HUMAN FACTORS
studies and insurance claim experience over the years with illnesses or conditions for which driving is contraindicated,
(Koppa et al. 1980) suggest tha t such driver's performance they are probably not enough of these to account for them in any
is indistinguishable from the general driving population. traffic flow models.
Although there are doubtless a number of people on the highways
3.8 Continuous Driver Performance

The previous sections of this chapter have sketched the relevant nothing can be more commonplace than steering a motor vehicle.
discrete performance characteristics of the driver in a traffic Figure 3.5 illustrates the conceptual model first proposed by
stream. Driving, however, is primarily a continuous dynamic Sheridan (1962). The human operator looks at both inputs, R(t)
process of managing the present heading and thus the future path the desired input forcing function (the road and where it seems
of the vehicle through the steering function. The first and second to be taking the driver), and E(t) the system error function, the
derivatives of location on the roadway in time, velocity and difference between where the road is going and what C(t) the
acceleration, are also continuous control processes through vehicle seems to be doing. The human operator can look ahead
modulated input using the accelerator (really the throttle) and the (lead), the prediction function, and also can correct for perceived
brake controls. errors in the path. If the driver were trying to drive by viewing
the road through a hole in the floor of the car, then the prediction
function would be lost, which is usually the state of affairs for
3.8.1 Steering Performance servomechanisms. The two human functions of prediction and
compensation are combined to make a control input to the vehicle
The driver is tightly coupled into the steering subsystem of the via the steering wheel which (for power steering) is also a servo
human-machine system we call the motor vehicle. It was only in its own right. The control output from this human-steering
during the years of World War II that researchers and engineers process combination is fed back (by the perception of the path
first began to model the human operator in a tracking situation of the vehicle) to close the loop. Mathematically, the setup in
by means of differential equations, i.e., a transfer function. The Figure 3.5 is expressed as follows, if the operator is reasonably
first paper on record to explore the human transfer function was linear:
by Tustin in 1944 (Garner 1967), and the subject was antiaircraft
gun control. The human operator in such tracking situations can Ke ts (1 TLs)
be described in the same terms as if he or she is a linear feedback g(s)
R (3.12)
(1 TLs)(1 TNs)
control system, even though the human operator is noisy, non-
linear, and sometimes not even closed-loop.
Sheridan (1962) reported some parameters for this Laplace

3.8.1.1 Human Transfer Function for Steering transform transfer function of the first order. K, the gain or
sensitivity term, varies (at least) between +35 db to -12 db. Gain,
Steering can be classified as a special case of the general pursuit how much response the human will make to a given input, is the
tracking model, in which the two inputs to the driver (which are parameter perhaps most easily varied, and tends to settle at some
somehow combined to produce the correction signal) are (1) the point comfortably short of instability (a phase margin of 60
desired path as perceived by the driver from cues provided by the degrees or more). The exponential term e-ts ranges from 0.12 to
roadway features, the streaming of the visual field, and higher 0.3 sec and is best interpreted as reaction time. This delay is the
order information; and (2) the perceived present course of the dominant limit to the human's ability to adapt to fast-changing
vehicle as inferred from relationship of the hood to roadway conditions. The T factors are all time constants, which however
features. The exact form of either of these two inputs are still may not stay constant at all. They must usually be empirically
subjects of investigation and some uncertainty, even though derived for a given control situation.
3 - 19
3. HUMAN FACTORS
Figure 3.5
Pursuit Tracking Configuration (after Sheridan 1962).
Sheridan reported some experimental results which show TL Table 3.9

(lead) varying between 0 and 2, Tl (lag) from 0.0005 to 25, and Maneuver Classification
TN (neuromuscular lag) from 0 to 0.67. R, the remnant term, is
usually introduced to make up for nonlinearities between input Driver Control Mode
and output. Its value is whatever it takes to make output track
input in a predictable manner. In various forms, and sometimes Compen- Pursuit Precognitive
with different and more parameters, Equation 3.10 expresses the Maneuver satory
basic approach to modeling the driver's steering behavior.
Novice drivers tend to behave primarily in the compensatory Highway Lane 1
Regulation
tracking mode, in which they primarily attend to the difference,
say, between the center of the hood and the edge line of the Precision Course 2 1
pavement, and attempt to keep that difference at some constant Control
visual angle. As they become more expert, they move more to
pursuit tracking as described above. There is also evidence that Turning; Ramp 2 1
Entry/Exit
there are "precognitive" open-loop steering commands to
particular situations such as swinging into an accustomed parking Lane Change 2 1
place in a vehicle the driver is familiar with. McRuer and Klein
(1975) classify maneuvers of interest to traffic flow modelers as Overtake/Pass 2 1
is shown in Table 3.9.
Evasive Lane 2 1
Change
In Table 3.9, the entries under driver control mode denote the
order in which the three kinds of tracking transition from one to
the other as the maneuver transpires. For example, for a turning
movement, the driver follows the dotted lines in an intersection 3.8.1.2 Performance Characteristics Based on Models
and aims for the appropriate lane in the crossroad in a pursuit
mode, but then makes adjustments for lane position during the The amplitude of the output from this transfer function has been
latter portion of the maneuver in a compensatory mode. In an found to rapidly approach zero as the frequency of the forcing
emergency, the driver "jerks" the wheel in a precognitive (open- function becomes greater than 0.5 Hz (Knight 1987). The driver
loop) response, and then straightens out the vehicle in the new makes smaller and smaller corrections as the highway or wind
lane using compensatory tracking. gusts or other inputs start coming in more frequently than one
complete cycle every two seconds.
3 - 20
3. HUMAN FACTORS
The time lag between input and output also increases with Then a stage of steady-state curve driving follows, with the driver
frequency. Lags approach 100 msec at an input of 0.5 Hz and now making compensatory steering corrections. The steering
increase almost twofold to 180 msec at frequencies of 2.4 Hz. wheel is then restored to straight-ahead in a period that covers the
The human tracking bandwidth is of the order of 1 to 2 Hz. endpoint of the curve. Road curvature (perceived) and vehicle
Drivers can go to a precognitive rhythm for steering input to speed predetermines what the initial steering input will be, in the
better this performance, if the input is very predictable, e.g., a following relationship:
"slalom" course. Basic lane maintenance under very restrictive
conditions (driver was instructed to keep the left wheels of a
SRl (1Fsu 2)Cr
vehicle on a painted line rather than just lane keep) was studied gs
(3.13)
very recently by Dulas (1994) as part of his investigation of 1000
changes in driving performance associated with in-vehicle
displays and input tasks. Speed was 57 km/h. Dulas found
average deviations of 15 cm, with a standard deviation of 3.2 cm. where,
Using a tolerance estimation based on the nearly 1000 Cr = roadway curvature
observations of deviation, the 95th percentile deviation would be SR = steering ratio
21 cm, the 99th would be 23 cm. Thus drivers can be expected Fs = stability factor
to weave back and forth in a lane in straightaway driving in an l = wheelbase
envelope of +/- 23 cm or 46 cm across. Steering accuracy with u = speed
degrade and oscillation will be considerably more in curves. since gs = steering wheel angle (radians)
such driving is mixed mode, with rather large errors at the
beginning of the maneuver, with compensatory corrections toward Godthelp found that the standard deviation of anticipatory steering
the end of the maneuver. Godthelp (1986) described this process inputs is about 9 percent of steering wheel angle gs. Since sharper
as follows. The driver starts the maneuver with a lead term before curves require more steering wheel input, inaccuracies will be
the curve actually begins. This precognitive control action proportionately greater, and will also induce more oscillation from
finishes shortly after the curve is entered. side to side in the curve during the compensatory phase of the
maneuver.
3.9 Braking Performance

The steering performance of the driver is integrated with either or more wheels locking and consequent loss of control at speeds
braking or accelerator positioning in primary control input. higher than 32 km/h, unless the vehicle is equipped with antiskid
Human performance aspects of braking as a continuous control brakes (ABS, or Antilock Brake System). Such a model of
input will be discussed in this section. After the perception- human braking performance is assumed in the time-hallowed
response time lag has elapsed, the actual process of applying the AASHTO braking distance formula (AASHTO 1990):
brakes to slow or stop the motor vehicle begins.
V2
d
(3.14)
257.9f
3.9.1 Open-Loop Braking Performance
where,
The simplest type of braking performance is "jamming on the d = braking distance - meters
brakes." The driver exerts as much force as he or she can muster, V = Initial speed - km/h
and thus approximates an instantaneous step input to the motor f = Coefficient of friction, tires to pavement
vehicle. Response of the vehicle to such an input is out of scope surface, approximately equal to deceleration
for this chapter, but it can be remarked that it can result in one in g units
3 - 21
3. HUMAN FACTORS
Figure 3.6 shows what such a braking control input really looks Figure 3.7 shows a braking maneuver on a tangent with the same
like in terms of the deceleration profile. This maneuver was on driver and vehicle, this time on a wet surface. Note the
a dry tangent section at 64 km/h, under "unplanned" conditions characteristic "lockup" footprint, with a steady-state deceleration
(the driver does not know when or if the signal to brake will be after lockup of 0.4 g.
given) with a full-size passenger vehicle not equipped with
ABS. Note the typical steep rise in deceleration to a peak of over From the standpoint of modeling driver input to the vehicle, the
0.9 g, then steady state at approximately 0.7 g for a brakes locked open-loop approximation is a step input to maximum braking
stop. The distance data is also on this plot: the braking distance effort, with the driver exhibiting a simple to complex PRT delay
on this run was 23 m feet. Note also that the suspension bounce prior to the step. A similar delay term would be introduced prior
produces a characteristic oscillation after the point at which the to release of the brake pedal, thus braking under stop-and-go
vehicle is completely stopped, just a little less than five seconds conditions would be a sawtooth.
into the run.
Figure 3.6
Typical Deceleration Profile for a Driver without
Antiskid Braking System on a Dry Surface.
3 - 22
3. HUMAN FACTORS
Figure 3.7
Typical Deceleration Profile for a Driver without
Antiskid Braking System on a Wet Surface.
3.9.2 Closed-Loop Braking Performance Table 3.10

Percentile Estimates of Steady State Unexpected
Recent research in which the writer has been involved provide Deceleration
some controlled braking performance data of direct application
to performance modeling (Fambro et al. 1994). "Steady state" Mean -0.55g
approximations or fits to these data show wide variations among
drivers, ranging from -0.46 g to -0.70 g. Standard Deviation 0.07
75th Percentile -0.43

Table 3.10 provides some steady-state derivations from empirical
data collected by Fambro et al. (1994). These were all responses 90th -0.37
to an unexpected obstacle or object encountered on a closed
course, in the driver's own (but instrumented) car. 95th -0.32
99th -0.24
Table 3.11 provides the same derivations from data collected on
drivers in their own vehicle in which the braking maneuver was
anticipated; the driver knows that he or she would be braking, but
during the run were unsure when the signal (a red light inside the pavement conditions. Pavement friction (short of ice) played
car) would come. very little part in driver's setting of these effort levels. About
0.05 to 0.10 g difference between wet pavement and dry
The ratio of unexpected to expected closed-loop braking effort pavement steady-state g was found.
was estimated by Fambro et al. to be about 1.22 under the same
3 - 23
3. HUMAN FACTORS
Table 3.11 to come to rest. Driver input to such a planned braking situation
Percentile Estimates of Steady-State Expected approximates a "ramp" (straight line increasing with time from
Deceleration zero) function with the slope determined by the distance to the
desired stop location or steady-state speed in the case of a
Mean -0.45g platoon being overtaken. The driver squeezes on pedal pressure
to the brakes until a desired deceleration is obtained. The
Standard Deviation 0.09 maximum "comfortable" braking deceleration is generally
accepted to be in the neighborhood of -0.30 g, or around 3
75th Percentile -0.36 m/sec2 (ITE 1992).
90th -0.31
The AASHTO Green Book (AASHTO 1990) provides a graphic
95th -0.27 for speed changes in vehicles, in response to approaching an
intersection. When a linear computation of decelerations from
99th -0.21
this graphic is made, these data suggest decelerations in the
neighborhood of -2 to -2.6 m/sec2 or -0.20 to -0.27 g. More
recent research by Chang et al. (1985) found values in response
3.9.3 Less-Than-Maximum to traffic signals approaching -0.39 g, and Wortman and
Matthias (1983) observed a range of -0.22 to -0.43 g, with a
Braking Performance mean level of -0.36 g. Hence controlled braking performance
that yields a g force of about -0.2 g would be a reasonable lower
The flow theorist may require an estimate of “comfortable”
level for a modeler, i.e., almost any driver could be expected to
braking performance, in which the driver makes a stop for
change the velocity of a passenger car by at least that amount,
intersections or traffic control devices which are discerned
but a more average or "typical" level would be around -0.35 g.
considerably in advance of the location at which the vehicle is
3.10 Speed and Acceleration Performance

The third component to the primary control input of the driver at the moment, etc. Drivers in heavy traffic use relative
to the vehicle is that of manual (as opposed to cruise control) perceived position with respect to other vehicles in the stream
control of vehicle velocity and changes in velocity by means of as a primary tracking cue (Triggs, 1988). A recent study
the accelerator or other device to control engine RPM. (Godthelp and Shumann 1994) found errors between speed
desired and maintained to vary from -0.3 to -0.8 m/sec in a lane
change maneuver; drivers tended to lose velocity when they
3.10.1 Steady-State Traffic Speed Control made such a maneuver. Under steady-stage conditions in a traffic
stream, the range of speed error might be estimated to be no
The driver's primary task under steady-state traffic conditions more than +/- 1.5 m/sec (Evans and Rothery 1973), basically
is to perform a tracking task with the speedometer as the display, modeled by a sinusoid. The growing prevalence of cruise
and the accelerator position as the control input. Driver controls undoubtedly will reduce the amplitude of this speed
response to the error between the present indicated speed and error pattern in a traffic stream by half or more.
the desired speed (the control signal) is to change the pedal
position in the direction opposite to the trend in the error
indication. How much of such an error must be present depends 3.10.2 Acceleration Control
upon a host of factors: workload, relationship of desired speed
to posted speed, location and design of the speedometer, and The performance characteristics of the vehicle driver are the
personal considerations affecting the performance of the driver limiting constraints on how fast the driver can accelerate the
3 - 24
3. HUMAN FACTORS
vehicle. The actual acceleration rates, particularly in a traffic 1992). If the driver removes his or her foot from the accelerator
stream as opposed to a standing start, are typically much lower pedal (or equivalent control input) drag and rolling resistance
than the performance capabilities of the vehicle, particularly a produce deceleration at about the same level as "unhurried"
passenger car. A nominal range for "comfortable" acceleration acceleration, approximately 1 m/sec2 at speeds of 100 km/h or
at speeds of 48 km/h and above is 0.6 m/sec2 to 0.7 m/sec2 higher. In contrast to operation of a passenger car or light truck,
(AASHTO 1990). Another source places the nominal heavy truck driving is much more limited by the performance
acceleration rate drivers tend to use under "unhurried" capabilities of the vehicle. The best source for such information
circumstances at approximately 65 percent of maximum is the Traffic Engineering Handbook (ITE 1992).
acceleration for the vehicle, somewhere around 1 m/sec2 (ITE
3.11 Specific Maneuvers at the Guidance Level

The discussion above has briefly outlined most of the more G = gradient (5/8)
fundamental aspects of driver performance relevant to modeling g = acceleration of gravity (9.8 m/sec2)
the individual driver-vehicle human-machine system in a traffic
stream. A few additional topics will now be offered to further The maximum acceleration capabilities of passenger vehicles
refine this picture of the driver as an active controller at the range from almost 3 m/sec2 from standing to less than 2 m/sec2
guidance level of operation in traffic. from 0 to highway speed.
Acceleration is still less when the maneuver begins at higher

3.11.1 Overtaking and Passing speeds, as low as 1 m/sec2 on some small subcompacts. In
in the Traffic Stream Equation 3.15, overtaking acceleration should be taken as 65
percent of maximum (ITE 1992). Large trucks or tractor-trailer
combinations have maximum acceleration capabilities on a level
3.11.1.1 Overtaking and Passing Vehicles
(4-Lane or 1-Way) roadway of no more than 0.4 m/sec2 at a standing start, and
decrease to 0.1 m/sec2 at speeds of 100 km/h. Truck drivers
Drivers overtake and pass at accelerations in the sub-maximal "floorboard" in passing maneuvers under these circumstances,
range in most situations. Acceleration to pass another vehicle and maximum vehicle performance is also typical driver input.
(passenger cars) is about 1 m/sec2 at highway speeds (ITE
1992). The same source provides an approximate equation for
acceleration on a grade: 3.11.1.2 Overtaking and Passing Vehicles
(Opposing Traffic)
Gg The current AASHTO Policy on Geometric Design (AASHTO

aGV ³ aLV (3.15) 1990) provides for an acceleration rate of 0.63 m/sec2 for an
100 initial 56 km/h, 0.64 m/sec2 for 70 km/h, and 0.66 m/sec2 for
speeds of 100 km/h. Based upon the above considerations, these
where, design guidelines appear very conservative, and the theorist may
wish to use the higher numbers in Section 3.11.1.1 in a
aGV = max acceleration rate on grade
sensitivity analysis.
aLV = max acceleration rate on level
3 - 25
3. HUMAN FACTORS
3.12 Gap Acceptance and Merging

3.12.1 Gap Acceptance
The driver entering or crossing a traffic stream must evaluate Table 3.12 provides very recent design data on these situations
the space between a potentially conflicting vehicle and himself from the Highway Capacity Manual (TRB 1985). The range of
or herself and make a decision whether to cross or enter or not. gap times under the various scenarios presented in Table 3.12
The time between the arrival of successive vehicles at a point is from a minimum of 4 sec to 8.5 sec. In a stream traveling at
is the time gap, and the critical time gap is the least amount of 50 km/h (14 m/sec) the gap distance thus ranges from 56 to 119
successive vehicle arrival time in which a driver will attempt a m; at 90 km/h (25 m/sec) the corresponding distances are 100
merging or crossing maneuver. There are five different gap to 213 m.
acceptance situations. These are:
(1) Left turn across opposing traffic, no traffic control 3.12.2 Merging
(2) Left turn across opposing traffic, with traffic control
(permissive green) In merging into traffic on an acceleration ramp on a freeway or
(3) Left turn onto two-way facility from stop or yield similar facility, the Situation (5) data for a four lane facility at
controlled intersection 90 km/h with a one second allowance for the ramp provides a
(4) Crossing two-way facility from stop or yield controlled baseline estimate of gap acceptance: 4.5 seconds. Theoretically
intersection as short a gap as three car lengths (14 meters) can be accepted
(5) Turning right onto two-way facility from stop or yield if vehicles are at or about the same speed, as they would be in
controlled intersection merging from one lane to another. This is the minimum,
however, and at least twice that gap length should be used as a
nominal value for such lane merging maneuvers.
3.13 Stopping Sight Distance

The minimum sight distance on a roadway should be sufficient empirically derived estimates now available in Fambro et al.
to enable a vehicle traveling at or near the design speed to stop (1994) for both these parts of the SSD equation are expressed
before reaching a "stationary object" in its path, according to the in percentile levels of drivers who could be expected to (1)
AASHTO Policy on Geometric Design (AASHTO 1990). It respond and (2) brake in the respective distance or shorter.
goes on to say that sight distance should be at least that required Since PRT and braking distance that a driver may achieve in a
for a "below-average" driver or vehicle to stop in this distance. given vehicle are not highly correlated, i.e., drivers that may be
very fast to initiate braking may be very conservative in the
Previous sections in this chapter on perception-response time actual braking maneuver, or may be strong brakers. PRT does
(Section 3.3.1) and braking performance (Section 3.2) provide not predict braking performance, in other words.
the raw materials for estimating stopping sight distance. The
time-honored estimates used in the AASHTO Green Book Very often, the engineer will use "worst case" considerations in
(AASHTO 1990) and therefore many other engineering a design analysis situation. What is the "reasonable" worst case
resources give a flat 2.5 sec for PRT, and then the linear for achieving the AASHTO "below average" driver and vehicle?
deceleration equation (Equation 3.13) as an additive model. Clearly, a 99th percentile PRT and a 99th percentile braking
This approach generates standard tables that are used to estimate distance gives an overly conservative 99.99 combined percentile
stopping sight distance (SSD) as a function of coefficients of
friction and initial speed at inception of the maneuver. The
3 - 26
3. HUMAN FACTORS
Table 3.12
Critical Gap Values for Unsignalized Intersections
Average Speed of Traffic
50 km/h 90 km/h
Maneuver Control
Number of Traffic Lanes, Major Roadway
2 4 2 4
1 None 5.0 5.5 5.5 6.0
2 Permissive 5.0 5.5 5.5 6.0

Green1
3 Stop 6.5 7.0 8.0 8.5
3 Yield 6.0 6.5 7.0 7.5
4 Stop 6.0 6.5 7.5 8.0
4 Yield 5.5 6.0 6.5 7.0
52 Stop 5.5 5.5 6.5 6.5
52 Yield 5.0 5.0 5.5 5.5

1
During Green Interval
2
If curve radius >15 m or turn angle <60 subtract 0.5 seconds.
If acceleration lane provided, subtract 1.0 seconds.
All times are given in seconds.
All Maneuvers: if population >250,000, subtract 0.5 seconds

if restricted sight distance, add 1.0 seconds
Maximum subtraction is 1.0 seconds
Maximum critical gap 8.5 seconds
level--everybody will have an SSD equal to or shorter than this For example, on a dry level roadway, using Equation 3.12, at a
somewhat absurd combination. The combination of 90th velocity of 88 km/h, the SSD components would be:
percentile level of performance for each segment yields a
combined percentile estimate of 99 percent (i.e., 0.10 of each PRT: 1.57 x 24.44 = 38.4 m
distribution is outside the envelope, and their product is 0.01,
therefore 1.00 - 0.01 = 0.99). A realistic worst case ( 99th Braking Distance: 82.6 m
percentile) combination to give SSD would, from previous
sections of this chapter be: SSD: 38 + 83 = 121 m
PRT: 1.57 sec (Table 3.4) For comparison, the standard AASHTO SSD for a dry level
roadway, using a nominal 0.65 for f, the coefficient of friction,
Braking deceleration: -0.37 g (Section 3.10.2) would be:
3 - 27
3. HUMAN FACTORS
PRT: 2.50 x 24.44 = 61.1 m These two estimates are comparable, but the first estimate has
an empirical basis for it. The analyst can assume other
Braking Distance: 47.3 m combinations of percentiles (for example, 75th percentile
performance in combination yields an estimate of the 94th
SSD: 61 + 47 = 108 m percentile). It is always possible, of course, to assume different
levels of percentile representation for a hypothetical driver, e.g.,
50th percentile PRT with 95th percentile braking performance.
3.14 Intersection Sight Distance

The AASHTO Policy on Geometric Design (AASHTO 1990)
identifies four differences cases for intersection sight distance 3.14.3 Case III: Stop Control on
considerations. From the viewpoint of traffic flow theory, the
Secondary Roadway
question may be posed, "How long is a driver going to linger at
an intersection before he or she begins to move?" Only the first
Hostetter et al. (1986) note that "for a large percentage of trials
three cases will be discussed here, since signalized intersections
at intersections with reasonable sign distance triangles, drivers
(Case IV) have been discussed in Section 3.5.1.
completed monitoring of the crossing roadway before coming
to a stop." Their solution to this dilemma was to include three
measures of PRT. They start at different points but terminate
3.14.1 Case I: No Traffic Control with the initiation of an accelerator input. One of the PRT's
starts with the vehicle at rest. The second begins with the first
The driver initiates either acceleration or deceleration based head movement at the stop. The third begins with the last head
upon his or her perceived gap in intersecting traffic flow. The movement in the opposite direction of the intended turn or
principles given in Section 3.13 apply here. PRT for this toward the shorter sight distance leg (for a through maneuver).
situation should be the same as for conditions of no surprise None of the three takes into account any processing the driver
outlined in Section 3.3.1. AASHTO (1990) gives an allowance might be doing prior to the stop at the intersection.
of three seconds for PRT, which appears to be very conservative
under these circumstances. Their findings were as follows (Table 3.13):
Table 3.13
3.14.2 Case II: Yield Control for PRTs at Intersections
Secondary Roadway 4-way T-Intersection
This is a complex situation. McGee et al. (1983) could not find Mean 85th Mean 85th
reliable data to estimate the PRT. A later, follow-up study by
Hostetter et al. (1986) considered the PRT to stretch from the PRT 1 2.2 sec 2.7 2.8 3.1
time that the YIELD sign first could be recognized as such to PRT 2 1.8 2.6 1.9 2.8
the time that the driver either began a deceleration maneuver or
speeded up to clear the intersection in advance of cross traffic. PRT 3 1.6 2.5 1.8 2.5
But decelerations often started 300 m or more from the
intersection, a clear response to the sign and not to the traffic
Thus a conservative estimate of PRT, i.e., time lag at an
ahead. PRTs were thus in the range of 20 to over 30 sec. with
intersection before initiating a maneuver, would be somewhat
much variability and reflect driving style rather than
in excess of three seconds for most drivers.
psychophysical performance.
3 - 28
3. HUMAN FACTORS
3.15 Other Driver Performance Characteristics

3.15.1 Speed Limit Changes 3.15.3 Real-Time Driver Information Input
Gstalter and Hoyos (1988) point out the well-known With the advent of Intelligent Transportation Systems (ITS),
phenomenon that drivers tend to adapt to sustained speed over driver performance changes associated with increased
a period of time, such that the perceived velocity (not looking information processing work load becomes a real possibility.
at the speedometer) lessens. In one study cited by these authors, ITS may place message screens, collision avoidance displays
drivers drove 32 km at speeds of 112 km/h. Subjects were then and much more in the vehicle of the future. Preliminary
instructed to drop to 64 km/h. The average speed error turned studies of the effects of using such technology in a traffic stream
out to be more than 20 percent higher than the requested speed. are just now appearing in the literature. There is clearly much
A similar effect undoubtedly occurs when posted speed limits more to come. For a review of some of the human factors
change along a corridor. In the studies cited, drivers were aware implications of ITS, see Hancock and Pansuraman (1992) and
of the "speed creep" and attempted (they said) to accommodate any recent publications by Peter Hancock of The University of
for it. When drivers go from a lower speed to a higher one, they Minnesota.
also adapt, such that the higher speed seems higher than in fact
it is, hence errors of 10 to 20 percent slower than commanded Drivers as human beings have a very finite attentional resource
speed occur. It takes several minutes to re-adapt. Hence speed capacity, as summarized in Dulas (1994). Resources can be
adjustments on a corridor should not be modeled by a simple allocated to additional information processing tasks only at the
step function, but rather resemble an over damped first-order cost of decreasing the efficiency and accuracy of those tasks.
response with a time constant of two minutes or more. When the competing tasks use the same sensory modality and
similar resources in the brain, increases in errors becomes
dramatic. To the extent that driving is primarily a psychomotor
3.15.2 Distractors On/Near Roadway task at the skill-based level of behavior, it is relatively immune
to higher-level information processing, if visual perception is
One of the problems that militates against smooth traffic flow not a dominant factor.
on congested facilities is the "rubber neck" problem. Drivers
passing by accident scenes, unusual businesses or activities on But as task complexity increases, say, under highly congested
the road side, construction or maintenance work, or other urban freeway conditions, any additional task becomes very
occurrences irrelevant to the driving task tend to shift sufficient disruptive of performance. This is especially true of older
attention to degrade their driving performance. In Positive drivers. Even the use of cellular telephones in traffic has been
Guidance terms (Lunenfeld and Alexander 1990) such a found to be a potent disrupter of driving performance (McKnight
situation on or near the roadway is a temporary violation of and McKnight 1993). A study described by Dulas (1994) found
expectancy. How to model the driver response to such that drivers using a touch screen CRT at speeds of 64 km tended
distractions? In the absence of specific driver performance data to increase lane deviations such that the probability of lane
on distractors, the individual driver response could be estimated excursion was 0.15. Early and very preliminary studies indicate
by injecting a sudden accelerator release with consequent that close attention to established human engineering principles
deceleration from speed discussed in Section 3.11.2. This for information display selection and design should result in real-
response begins as the distraction comes within a cone of 30 time information systems that do not adversely affect driver
degrees centered around the straight-ahead direction on a performance.
tangent, and the outer delineation of the curve on a horizontal
curve. A possible increase in the amplitude of lane excursions
could also occur, similar to the task-loaded condition in the
Dulas (1994) study discussed in Section 3.9.1.2.
3 - 29
3. HUMAN FACTORS
References
American Association of Highway and Transportation Fitts, P. M. (1954). The Information Capacity of the Human
Officials (1990). A Policy on Geometric Design of Motor System in Controlling the Amplitude of Movement.
Highways and Streets. Washington, DC. Journal of Experimental Psychology, 47, pp. 381-391.
Ang, A. H-S., and W. H. Tang (1975). Probability Concepts Fox, M. (1989). Elderly Driver's Perceptions of Their
in Engineering Planning and Design Vol. 1: Basic Driving Abilities Compared to Their Functional Visual
Principles. New York: John Wiley and Sons. Perception Skills and Their Actual Driving Performance.
Berman, A. H. (1994). A Study of Factors Affecting Foot Taira, E. (Ed.). Assessing the Driving Ability of the
Movement Time in a Braking Maneuver. Unpublished MS Elderly New York: Hayworth Press.
Thesis, Texas A&M University. Garner, K.C. (1967). Evaluation of Human Operator
Blackwell, O. and H. Blackwell (1971). Visual Performance Coupled Dynamic Systems. In Singleton, W. T., Easterby,
Data for 156 Normal Observers of Various Ages. Journal R. S., and Whitfield, D. C. (Eds.). The Human Operator
of Illumination Engineering Society 1, pp. 3-13. in Complex Systems, London: Taylor and Francis.
Boff, K. R. and J. E. Lincoln (1988). Engineering Data Godthelp, H. (1986). Vehicle Control During Curve Driving.
Compendium. USAF, H. G. Armstrong Medical Research Human Factors, 28,2, pp. 211-221.
Laboratory, Wright-Patterson AFB, Ohio, (Section 8.1). Godthelp, H. and J. Shumann (1994). Use of Intelligent
Brackett, R. Q. and R. J. Koppa (1988). Preliminary Study Accelerator as an Element of a Driver Support System.
of Brake Pedal Location Accuracy. Proceedings of the Abstracted in Driver Performance Data Book Update:
Human Factors and Ergonomics Society 32nd Meeting. Older Drivers and IVHS Circular 419, Transportation
Chang, M-S, C. J. Messer, and A. J. Santiago (1985). Timing Research Board, National Research Council, Washington,
Traffic Signal Change Intervals Based on Driver DC.
Behavior. Transportation Research Record 1027, Greene, F. A. (1994). A Study of Field and Laboratory
Transportation Research Board, National Research Legibility Distances for Warning Symbol Signs.
Council, Washington, DC. Unpublished doctoral dissertation, Texas A&M University.
Drury, C. G. (1975). Application of Fitt's Law to Foot Pedal Gstalter, H. and C. G. Hoyos (1988). Human Factors Aspects
Design. Human Factors 12, pp. 559-561. of Road Traffic Safety. In Peters, G.A. and Peters, B.J.,
Dudek, C. L. (1990). Guidelines on the Use of Changeable Automotive Engineering and Litigation Volume 2, New
Message Signs. FHWA-TS-90-043, Federal Highway York: Garland Law Publishing Co.
Administration, Washington, DC. Hancock, P. A. and R. Parasuraman (1992). Human Factors
Dulas, R. J. (1994). Lane Deviation as a Result of Visual and Safety in the Design of Intelligent Vehicle - Highway
Workload and Manual Data Entry. Unpublished doctoral Systems (IVHS).
dissertation, Texas A&M University. Hoffman, E. R. (1991). Accelerator to Brake Movement
Evans, L. and R. Rothery (1973). Experimental Times and A Comparison of Hand and Foot Movement
Measurements of Perceptual Thresholds in Car Times. Ergonomics 34, pp. 277-287 and pp. 397-406.
Following. Highway Research Record 454, Transportation Hooper, K. G. and H. W. McGee (1983). Driver Perception-
Research Board, Washington, DC. Reaction Time: Are Revisions to Current Specifications
Fambro, D. et al. (1994). Determination of Factors Affecting in Order? Transportation Research Record 904,
Stopping Sight Distance. Working Paper I: Braking Transportation Research Board, National Research
Studies (DRAFT) National Cooperative Highway Research Council, Washington, DC, pp. 21-30.
Program, Project 3-42. Hostetter, R., H. McGee, K. Crowley, E. Seguin, and G. Dauber,
Federal Highway Administration (1983). Traffic Control (1986). Improved Perception-Reaction Time Information
Device Handbook.Washington, DC. for Intersection Sight Distance. FHWA/RD-87/015,
Federal Highway Administration, Washington, DC.
3 - 30
3. HUMAN FACTORS
Hulbert, S. (1988). The Driving Task. In Peters, G.A. and McKnight, A. J. and A. S. McKnight (1993). The Effect
Peters, B.J., Automotive Engineering and Litigation of Cellular Phone Use Upon Driver Attention. Accident
Volume 2, New York: Garland Law Publishing Co. Analysis and Prevention 25 (3), pp. 259-265.
Institute of Transportation Engineers (1992). Traffic McPherson, K., J. Michael, A. Ostrow, and P. Shaffron
Engineering Handbook, Washington, DC. (1988). Physical Fitness and the Aging Driver - Phase I.
IVHS America Publication B-571, Society of Automotive AAA Foundation for Traffic Safety, Washington, DC.
Engineers, (1992). Intelligent Vehicle-Highway Systems McRuer, D. T. and R. H. Klein (1975). Automobile
(IVHS) Safety and Human Factors Considerations. Controllability - Driver/Vehicle Response for Steering
Warrendale, PA. Control. Systems Technology Inc. TR-1040-1-I. DOT-
Jacobs, R. J., A. W. Johnston, and B. L. Cole, (1975). The HS-801-407, National Highway Traffic Safety
Visibility of Alphabetic and Symbolic Traffic Signs. Administration, Washington, DC.
Australian Board Research, 5, pp. 68-86. Moir, A. and D. Jessel (1991). Brain Sex. New York:
Johansson, G. and K. Rumar (1971). Driver's Brake Reaction Carol Publishing Co.
Times. Human Factors, 12 (1), pp. 23-27. Mortimer, R. G. (1988). Rear End Crashes. Chapter 9 of
Knight, J. L. (1987). Manual Control of Tracking in Peters, G.A. and Peters, B.J., Automotive Engineering and
Salvendy, G. (ed) Handbook of Human Factors. New Litigation,Vol. 2, New York: Garland Law Publishing Co.
York: John Wiley and Sons. NHTSA Driver Performance Data Book (1987).
Koppa, R. J. (1990). State of the Art in Automotive Adaptive R. L. Henderson (ed.) DOT HS 807 121, National Highway
Equipment. Human Factors, 32 (4), pp. 439-455. Traffic Safety Administration, Washington, DC.
Koppa, R. J., M. McDermott, et al., (1980). Human Factors NHTSA Driver Performance Data Book (1994).
Analysis of Automotive Adaptive Equipment for Disabled Transportation Reserach Circular 419, Transportation
Drivers. Final Report DOT-HS-8-02049, National Research Board, National Research Council, Washington,
Highway Traffic Safety Administration, Washington, DC. DC.
Kroemer, K. H. E., H. B. Kroemer, and K. E. Kroemer-Ebert Neuman, T. R. (1989). New Approach to Design for
(1994). Ergonomics: How to Design for Ease and Stopping Sight Distance. Transportation Research Record,
Efficiency Englewood Cliffs, NJ: Prentice-Hall. 1208, Transportation Research Board, National Research
Lerner, Neil, et al. (1995). Literature Review: Older Driver Council, Washington, DC.
Perception-Reaction Time for Intersection Sight Distance O'Leary, A. and R. Atkins (1993). Transportation Needs of
and Object Detection, Comsis Corp., Contract DTFH 61- the Older Driver. Virginia Transportation Research
90-00038 Federal Highway Administration, Washington, Council MS-5232 (FHWA/VA-93-R14).
DC, Final Report. Odeh, R. E. (1980). Tables for Normal Tolerance Limits,
Lunenfeld, H. and G. J. Alexander (1990). A User's Guide Sampling Plans, and Screening. New York: arcel Dekker,
to Positive Guidance (3rd Edition) FHWA SA-90-017, Inc.
Federal Highway Administration, Washington, DC. Ordy, J. , K. Brizzee, and H. Johnson (1982 ). Cellular
Marmor, M. (1982). Aging and the Retina. In R. Sekuler, Alterations in Visual Pathways and the Limbic System:
D. Kline, and K. Dismukes (eds.). Aging and Human Implications for Vision and Short Term Memory. ibid.
Visual Function, New York: Alan R. Liss. Picha, D. (1992). Determination of Driver Capability
McGee, H. W. (1989). Reevaluation of Usefulness and in the Detection and Recognition of an Object. In
Application of Decision Sight Distance. Transportation Kahl, K. B. Investigation of Appropriate Objects for Safe
Research Record 1208, TRB, NRC, Washington, DC. Stopping Distance. Unpublished Master's Thesis, Texas
McGee, H., K. Hooper, W. Hughes, and W. Benson (1983). A&M University.
Highway Design and Operational Standards Affected by Price, D. (1988). Effects of Alcohol and Drugs.
Driver Characteristics. FHWA RD 78-78, Federal In Peters, G. A. and Peters, B. J., Automotive Engineering
Highway Administration, Washington, DC. and Litigation Volume 2, New York: Garland Law
Publishing Co.
3 - 31
3. HUMAN FACTORS
Reason, J. (1990). Human Error. New York: Cambridge Transportation Research Board, National Research Council
University Press. (1988). Transportation in an Aging Society. Special
Schiff, W. (1980). Perception: An Applied Approach. Report 218, Washington, DC.
Boston: Houghton Mifflin. Transportation Research Board, National Research Council
Sekuler, R. and R. Blake (1990). Perception. 2nd Ed. New (1993). Human Factors Research in Highway Safety,
York: McGraw-Hill. TRB Circular 414, Washington, DC.
Sheridan, T. B. (1962). The Human Operator in Control Triggs, T. J. (1988). Speed Estimation, Chapter 18, Vol. II,
Instrumentation. In MacMillan, R. H., Higgins, T. J., and Peters, G. A. and Peters, B. Automotive Engineering and
Naslin, P. (Eds.) Litigation, op. cit.
Simmala, H. (1981). Driver/Vehicle Steering Response Weir, David, (1976). Personal communication at a meeting,
Latencies. Human Factors 23, pp. 683-692. CA.
Smiley, A. M. (1974). The Combined Effect of Alcohol Wierwille, W. W., J. G. Casali, and B. S. Repa (1983).
and Common Psychoactive Drugs: Field Studies with an Driver Steering Reaction Time to Abrupt-onset
Instrumented Automobile. Technical Bulletin ST-738, Crosswinds as Measured in a Moving Base Driving
National Aeronautical Establishment, Ottawa, Canada. Simulator. Human Factors 25, pp. 103-116.
Taoka, G. T. (1989). Brake Reaction Times of Unalerted Wortman, R. H. and J. S. Matthias (1983). Evaluation of
Drivers. ITE Journal, Washington, DC. Driver Behavior at Signalized Intersections.
Transportation Research Board (1985). Highway Capacity Transportation Research Record 904, TRB, NRC,
Manual, Special Report 209, Washington, DC. Washington, DC.
Transportation Research Board (1987). Zero Alcohol and
Other Options - Limits for Truck and Bus Drivers, Special
Report 216, Transportation Research Board, National
Research Council, Washington, DC.
3 - 32
CAR FOLLOWING MODELS
BY RICHARD W. ROTHERY6
6
Senior Lecturer, Civil Engineering Department, The University of Texas, ECJ Building 6.204, Austin, TX
78712
= Numerical coefficients tc = Collision time

a ,m =
#
Generalized sensitivity coefficient T = Reaction time
af (t) = Instantaneous acceleration of a following U #
= Speed of a lead vehicle
vehicle at time t Uf = Speed of a following vehicle
a (t)
#
= Instantaneous acceleration of a lead Uf = Final vehicle speed
vehicle at time t Uf = Free mean speed, speed of traffic near zero
= Numerical coefficient concentration
C = Single lane capacity (vehicle/hour) Ui = Initial vehicle speed
- = Rescaled time (in units of response time, Urel = Relative speed between a lead and
T) following vehicle
= Short finite time period u (t)
#
= Velocity profile of the lead vehicle of a
F = Amplitude factor platoon
= Numerical coefficient V = Speed
k = Traffic stream concentration in vehicles Vf = Final vehicle speed
per kilometer 7 = Frequency of a monochromatic speed
kj = Jam concentration oscillation
km = Concentration at maximum flow ẍf(t) = Instantaneous acceleration of a following
kf = Concentration where vehicle to vehicle vehicle at time t
interactions begin ẍ#(t) = Instantaneous speed of a lead vehicle at
kn = Normalized concentration time t
L = Effective vehicle length ẍf(t) = Instantaneous speed of a following vehicle
L-1 = Inverse Laplace transform at time t
= Proportionality factor x #(t) = Instantaneous speed of a lead vehicle at
i = Sensitivity coefficient, i = 1,2,3,... time t
ln(x) = Natural logarithm of x x f(t) = Instantaneous speed of a following vehicle
q = Flow in vehicles per hour at time t
qn = Normalized flow x#(t) = Instantaneous position of a lead vehicle at
<S> = Average spacing rear bumper to rear time t
bumper xf(t) = Instantaneous position of the following
Si = Initial vehicle spacing vehicle at time t
Sf = Final vehicle spacing xi(t) = Instantaneous position of the ith vehicle at
So = Vehicle spacing for stopped traffic time t
S(t) = Inter-vehicle spacing z(t) = Position in a moving coordinate system
S = Inter-vehicle spacing change
T = Average response time <x> = Average of a variable x
To = Propagation time for a disturbance
t = Time 6 = Frequency factor
4.
CAR FOLLOWING MODELS
It has been estimated that mankind currently devotes over 10 roadways. The speed-spacing relations that were obtained from
million man-years each year to driving the automobile, which on these studies can be represented by the following equation:
demand provides a mobility unequaled by any other mode of
transportation. And yet, even with the increased interest in
traffic research, we understand relatively little of what is S
VV 2 (4.2)
involved in the "driving task". Driving, apart from walking,
talking, and eating, is the most widely executed skill in the world
today and possibly the most challenging. where the numerical values for the coefficients, , , and take
on various values. Physical interpretations of these coefficients
Cumming (1963) categorized the various subtasks that are are given below:
involved in the overall driving task and paralleled the driver's
role as an information processor (see Chapter 3). This chapter = the effective vehicle length, L
focuses on one of these subtasks, the task of one vehicle = the reaction time, T
following another on a single lane of roadway (car following). = the reciprocal of twice the maximum average
This particular driving subtask is of interest because it is deceleration of a following vehicle
relatively simple compared to other driving tasks, has been
successfully described by mathematical models, and is an In this case, the additional term, V2, can provide sufficient
important facet of driving. Thus, understanding car following spacing so that if a lead vehicle comes to a full stop
contributes significantly to an understanding of traffic flow. Car instantaneously, the following vehicle has sufficient spacing to
following is a relatively simple task compared to the totality of come to a complete stop without collision. A typical value
tasks required for vehicle control. However, it is a task that is empirically derived for would be 0.023 seconds 2/ft . A less
commonly practiced on dual or multiple lane roadways when conservative interpretation for the non-linear term would be:
passing becomes difficult or when traffic is restrained to a single
lane. Car following is a task that has been of direct or indirect
0.5(af 1 a 1)#
(4.3)
interest since the early development of the automobile.
One aspect of interest in car following is the average spacing, S, where aƒ and a are the average maximum decelerations of the
#
that one vehicle would follow another at a given speed, V. The following and lead vehicles, respectively. These terms attempt
interest in such speed-spacing relations is related to the fact that to allow for differences in braking performances between
nearly all capacity estimates of a single lane of roadway were vehicles whether real or perceived (Harris 1964).
based on the equation:
For = 0, many of the so-called "good driving" rules that have
C = (1000) V/S (4.1) permeated safety organizations can be formed. In general, the
speed-spacing Equation 4.2 attempts to take into account the
where physical length of vehicles; the human-factor element of
C = Capacity of a single lane perception, decision making, and execution times; and the net
(vehicles/hour) physics of braking performances of the vehicles themselves. It
V = Speed (km/hour) has been shown that embedded in these models are theoretical
S = Average spacing rear bumper to rear estimates of the speed at maximum flow, (/)0.5; maximum
bumper in meters flow, [ + 2( )0.5]-1; and the speed at which small changes in
traffic stream speed propagate back through a traffic stream,
The first Highway Capacity Manual (1950) lists 23 (/) 0.5 (Rothery 1968).
observational studies performed between 1924 and 1941 that
were directed at identifying an operative speed-spacing relation The speed-spacing models noted above are applicable to cases
so that capacity estimates could be established for single lanes of where each vehicle in the traffic stream maintains the same or
nearly the same constant speed and each vehicle is attempting to

$5 2//2:,1* 2'(/6
maintain the same spacing (i.e., it describes a steady-state traffic 1958 by Herman and his associates at the General Motors
stream). Research Laboratories. These research efforts were microscopic
approaches that focused on describing the detailed manner in
Through the work of Reuschel (1950) and Pipes (1953), the which one vehicle followed another. With such a description,
dynamical elements of a line of vehicles were introduced. In the macroscopic behavior of single lane traffic flow can be
these works, the focus was on the dynamical behavior of a approximated. Hence, car following models form a bridge
stream of vehicles as they accelerate or decelerate and each between individual "car following" behavior and the
driver-vehicle pair attempts to follow one another. These efforts macroscopic world of a line of vehicles and their corresponding
were extended further through the efforts of Kometani and flow and stability properties.
Sasaki (1958) in Japan and in a series of publications starting in
4.1 Model Development

Car following models of single lane traffic assume that there is vehicle characteristics or class of
a correlation between vehicles in a range of inter-vehicle characteristics and from the driver's vast
spacing, from zero to about 100 to 125 meters and provides an repertoire of driving experience. The
explicit form for this coupling. The modeling assumes that each integration of current information and
driver in a following vehicle is an active and predictable control catalogued knowledge allows for the
element in the driver-vehicle-road system. These tasks are development of driving strategies which
termed psychomotor skills or perceptual-motor skills because become "automatic" and from which evolve
they require a continued motor response to a continuous series "driving skills".
of stimuli.
Control: The skilled driver can execute control
The relatively simple and common driving task of one vehicle commands with dexterity, smoothness, and
following another on a straight roadway where there is no coordination, constantly relying on feedback
passing (neglecting all other subsidiary tasks such as steering, from his own responses which are
routing, etc.) can be categorized in three specific subtasks: superimposed on the dynamics of the system's
counterparts (lead vehicle and roadway).
Perception: The driver collects relevant information
mainly through the visual channel. This It is not clear how a driver carries out these functions in detail.
information arises primarily from the motion The millions of miles that are driven each year attest to the fact
of the lead vehicle and the driver's vehicle. that with little or no training, drivers successfully solve a
Some of the more obvious information multitude of complex driving tasks. Many of the fundamental
elements, only part of which a driver is questions related to driving tasks lie in the area of 'human
sensitive to, are vehicle speeds, accelerations factors' and in the study of how human skill is related to
and higher derivatives (e.g., "jerk"), inter- information processes.
vehicle spacing, relative speeds, rate of
closure, and functions of these variables (e.g., The process of comparing the inputs of a human operator to that
a "collision time"). operator's outputs using operational analysis was pioneered by
the work of Tustin (1947), Ellson (1949), and Taylor (1949).
Decision These attempts to determine mathematical expressions linking
Making: A driver interprets the information obtained by input and output have met with limited success. One of the
sampling and integrates it over time in order to primary difficulties is that the operator (in our case the driver)
provide adequate updating of inputs. has no unique transfer function; the driver is a different
Interpreting the information is carried out 'mechanism' under different conditions. While such an approach
within the framework of a knowledge of has met with limited success, through the course of studies like

$5 2//2:,1* 2'(/6
these a number of useful concepts have been developed. For where is a proportionality factor which equates the stimulus
example, reaction times were looked upon as characteristics of function to the response or control function. The stimulus
individuals rather than functional characteristics of the task itself. function is composed of many factors: speed, relative speed,
In addition, by introducing the concept of "information", it has inter-vehicle spacing, accelerations, vehicle performance, driver
proved possible to parallel reaction time with the rate of coping thresholds, etc.
with information.
Do all of these factors come into play part of the time? The
The early work by Tustin (1947) indicated maximum rates of the question is, which of these factors are the most significant from
order of 22-24 bits/second (sec). Knowledge of human an explanatory viewpoint. Can any of them be neglected and still
performance and the rates of handling information made it retain an approximate description of the situation being
possible to design the response characteristics of the machine for modeled?
maximum compatibility of what really is an operator-machine
system. What is generally assumed in car following modeling is that a
driver attempts to: (a) keep up with the vehicle ahead and (b)
The very concept of treating an operator as a transfer function avoid collisions.
implies, partly, that the operator acts in some continuous
manner. There is some evidence that this is not completely These two elements can be accomplished if the driver maintains
correct and that an operator acts in a discontinuous way. There a small average relative speed, Urel over short time periods, say
is a period of time during which the operator having made a t, i.e.,
"decision" to react is in an irreversible state and that the response
1 t t/2
must follow at an appropriate time, which later is consistent with <U# Uf>
<Urel> U (t)dt
t 2t t/2 rel
(4.5)
the task.
The concept of a human behavior being discontinuous in is kept small. This ensures that ‘collision’ times:
carrying out tasks was first put forward by Uttley (1944) and
S(t)
has been strengthened by such studies as Telfor's (1931), who tc
(4.6)
demonstrated that sequential responses are correlated in such a Urel
way that the response-time to a second stimulus is affected
significantly by the separation of the two stimuli. Inertia, on the
other hand, both in the operator and the machine, creates an are kept large, and inter-vehicle spacings would not appreciably
appearance of smoothness and continuity to the control element. increase during the time period, t. The duration of the t will
depend in part on alertness, ability to estimate quantities such as:
In car following, inertia also provides direct feedback data to the spacing, relative speed, and the level of information required for
operator which is proportional to the acceleration of the vehicle. the driver to assess the situation to a tolerable probability level
Inertia also has a smoothing effect on the performance (e.g., the probability of detecting the relative movement of an
requirements of the operator since the large masses and limited object, in this case a lead vehicle) and can be expressed as a
output of drive-trains eliminate high frequency components of function of the perception time.
the task.
Because of the role relative-speed plays in maintaining relatively
Car following models have not explicitly attempted to take all of large collision times and in preventing a lead vehicle from
these factors into account. The approach that is used assumes 'drifting' away, it is assumed as a first approximation that the
that a stimulus-response relationship exists that describes, at argument of the stimulus function is the relative speed.
least phenomenologically, the control process of a driver-vehicle
unit. The stimulus-response equation expresses the concept that From the discussion above of driver characteristics, relative
a driver of a vehicle responds to a given stimulus according to a speed should be integrated over time to reflect the recent time
relation: history of events, i.e., the stimulus function should have the form
like that of Equation 4.5 and be generalized so that the stimulus
Response = Stimulus (4.4)

$5 2//2:,1* 2'(/6
at a given time, t, depends on the weighted sum of all earlier where

values of the relative speed, i.e.,
(t T)
0, for tCT (4.9)
< U# Uf >
<Urel >
)(t t )Urel(t )dt
t t/2
2t t/2 (4.7)
(t T)
1, for t
T (4.10)
where (t) is a weighing function which reflects a driver's
estimation, evaluation, and processing of earlier information
(Chandler et al. 1958). The driver weighs past and present and
information and responds at some future time. The consequence
(t T)dt
1
of using a number of specific weighing functions has been 2o
examined (Lee 1966), and a spectral analysis approach has been
used to derive a weighing function directly from car following
data (Darroch and Rothery 1969).
For this case, our stimulus function becomes
The general features of a weighting function are depicted in
Figure 4.1. What has happened a number of seconds ( 5 sec) Stimulus(t) = U (t - T) - Uf (t - T)
#
(4.11)
in the past is not highly relevant to a driver now, and for a short
time ( 0.5 sec) a driver cannot readily evaluate the information which corresponds to a simple constant response time, T, for a
available to him. One approach is to assume that driver-vehicle unit. In the general case of (t), there is an
average response time, T , given by
)(t)
(t T) (4.8)
T(t)
t )(t )dt
t
20 (4.12)
Past Future
W eighting
function
Now
Time
Figure 4.1
Schematic Diagram of Relative Speed Stimulus
and a Weighting Function Versus Time (Darroch and Rothery 1972).

$5 2//2:,1* 2'(/6
The main effect of such a response time or delay is that the driver
is responding at all times to a stimulus. The driver is observing . .
the stimulus and determining a response that will be made some ẍf(tT)
[x#(t) x f(t)] (4.15)
time in the future. By delaying the response, the driver obtains
"advanced" information.
Equation 4.15 is a first approximation to the stimulus-response
For redundant stimuli there is little need to delay response, apart equation of car-following, and as such it is a grossly simplified
from the physical execution of the response. Redundancy alone description of a complex phenomenon. A generalization of car
can provide advance information and for such cases, response following in a conventional control theory block diagram is
times are shorter. shown in Figure 4.1a. In this same format the linear car-
following model presented in Equation 4.15 is shown in Figure
The response function is taken as the acceleration of the 4.1b. In this figure the driver is represented by a time delay and
following vehicle, because the driver has direct control of this a gain factor. Undoubtedly, a more complete representation of
quantity through the 'accelerator' and brake pedals and also car following includes a set of equations that would model the
because a driver obtains direct feedback of this variable through dynamical properties of the vehicle and the roadway
inertial forces, i.e., characteristics. It would also include the psychological and
physiological properties of drivers, as well as couplings between
Response (t) = af (t) = ẍ f (t) (4.13) vehicles, other than the forward nearest neighbors and other
driving tasks such as lateral control, the state of traffic, and
where xi(t) denotes the longitudinal position along the roadway emergency conditions.
of the ith vehicle at time t. Combining Equations4.11 and 4.13
into Equation 4.4 the stimulus-response equation becomes For example, vehicle performance undoubtedly alters driver
(Chandler et al. 1958): behavior and plays an important role in real traffic where mixed
traffic represents a wide performance distribution, and where
. . appropriate responses cannot always be physically achieved by
ẍf(t)
[x#(t T) x f(t T)] (4.14)
a subset of vehicles comprising the traffic stream. This is one
area where research would contribute substantially to a better
understanding of the growth, decay, and frequency of
or equivalently disturbances in traffic streams (see, e.g., Harris 1964; Herman
and Rothery 1967; Lam and Rothery 1970).
Figure 4.1a
Block Diagram of Car-Following.

$5 2//2:,1* 2'(/6
Figure 4.1b
Block Diagram of the Linear Car-Following Model.
4.2 Stability Analysis

In this section we address the stability of the linear car following 4.2.1 Local Stability
equation, Equation 4.15, with respect to disturbances. Two
particular types of stabilities are examined: local stability and In this analysis, the linear car following equation, (Equation
asymptotic stability. 4.15) is assumed. As before, the position of the lead vehicle and
the following vehicle at a time, t, are denoted by x (t) and xf (t),
#
Local Stability is concerned with the response of a following respectively. Rescaling time in units of the response time, T,
vehicle to a fluctuation in the motion of the vehicle directly in using the transformation, t = -T, Equation 4.15 simplifies to
front of it; i.e., it is concerned with the localized behavior
between pairs of vehicles. . .
ẍf(-1)
C[(x#(-) x f(-))] (4.16)
Asymptotic Stability is concerned with the manner in which a
fluctuation in the motion of any vehicle, say the lead vehicle of where C = T. The conditions for the local behavior of the
a platoon, is propagated through a line of vehicles. following vehicle can be derived by solving Equation 4.16 by the
method of Laplace transforms (Herman et al. 1959).
The analysis develops criteria which characterize the types of
possible motion allowed by the model. For a given range of The evaluation of the inverse Laplace transform for Equation
model parameters, the analysis determines if the traffic stream 4.16 has been performed (Chow 1958; Kometani and Sasaki
(as described by the model) is stable or not, (i.e., whether 1958). For example, for the case where the lead and following
disturbances are damped, bounded, or unbounded). This is an vehicles are initially moving with a constant speed, u, the
important determination with respect to understanding the solution for the speed of the following vehicle was given by
applicability of the modeling. It identifies several characteristics Chow where denotes the integral part of t/T. The complex
with respect to single lane traffic flow, safety, and model validity. form of Chow's solution makes it difficult to describe various
If the model is realistic, this range should be consistent with physical properties (Chow 1958).
measured values of these parameters in any applicable situation
where disturbances are known to be stable. It should also be
v n
consistent with the fact that following a vehicle is an extremely
- (n))T
n 1
t
common experience, and is generally stable. x n(t)

u ( 1)n & (u0(t -) u)dt

0 (n)T (n 1)!)!

&$5 )2//2:,1* 02'(/6
However, the general behavior of the following vehicle's motion particular, it demonstrates that in order for the following vehicle
can be characterized by considering a specific set of initial not to over-compensate to a fluctuation, it is necessary that C
conditions. Without any loss in generality, initial conditions are 1/e. For values of C that are somewhat greater, oscillations
assumed so that both vehicles are moving with a constant speed, occur but are heavily damped and thus insignificant. Damping
u. Then using a moving coordinate system z(t) for both the lead occurs to some extent as long as
and following vehicles the formal solution for the acceleration C < %/2.
of the following vehicle is given more simply by:
These results concerning the oscillatory and non-oscillatory
behavior apply to the speed and acceleration of the following
L 1[C(C se s) 1s (4.16a) vehicle as well as to the inter-vehicle spacing. Thus, e.g., if C
e-1, the inter-vehicle spacing changes in a non-oscillatory manner
by the amount S , where
where L-1 denotes the inverse Laplace transform. The character
of the above inverse Laplace transform is determined by the S
1
(V U) (4.18)
singularities of the factor (C + ses)-1 since Cs2Z (s) is a regular
#

function. These singularities in the finite plane are the simple
poles of the roots of the equation when the speeds of the vehicle pair changes from U to V. An
C se
0
s
(4.17) important case is when the lead vehicle stops. Then, the final
speed, V, is zero, and the total change in inter-vehicle spacing is
- U/ .
Similarly, solutions for vehicle speed and inter-vehicle spacings
can be obtained. Again, the behavior of the inter-vehicle spacing In order for a following vehicle to avoid a 'collision' from
is dictated by the roots of Equation 4.17. Even for small t, the initiation of a fluctuation in a lead vehicle's speed the inter-
character of the solution depends on the pole with the largest real vehicle spacing should be at least as large as U/. On the other
part, say , s0 = a 0 + ib 0, since all other poles have considerably hand, in the interests of traffic flow the inter-vehicle spacing
larger negative real parts so that their contributions are heavily should be small by having as large as possible and yet within
damped. the stable limit. Ideally, the best choice of is (eT)-1.
Hence, the character of the inverse Laplace transform has the The result expressed in Equation 4.18 follows directly from
at tb t
form e 0 e 0 . For different values of C, the pole with the Chow's solution (or more simply by elementary considerations).
largest real part generates four distinct cases: Because the initial and final speeds for both vehicles are U and
V, respectively, we have
a) if C e 1(0.368), then a00, b0
0 , and the

motion is non-oscillatory and exponentially x¨f(tT)dt
V U (4.19)
damped. 20
b) if e- 1 < C < % / 2, then a0< 0, b 0 > 0 and the and from Equation 4.15 we have
motion is oscillatory with exponential damping.
if C = % / 2 , then a0 = 0, b0, > 0 and the motion is . .

c)

[x l(t) x f(t)]dt
S
oscillatory with constant amplitude. 20
d) if C > % / 2 then a0 > 0, b0 > 0 and the motion is or
oscillatory with increasing amplitude.
S
2 [x #(t) x. f(t)]dt
V U (4.20)
The above establishes criteria for the numerical values of C 0
which characterize the motion of the following vehicle. In

&$5 )2//2:,1* 02'(/6
as given earlier in Equation 4.18. Figure 4.2 illustrates the solutions for C= e-1, where the lead
vehicle reduces its speed and then accelerates back to its original
In order to illustrate the general theory of local stability, the speed. Since C has a value for the locally stable limit, the
results of several calculations using a Berkeley Ease analog acceleration and speed of the following vehicle, as well as the
computer and an IBM digital computer are described. It is inter-vehicle spacing between the two vehicles are non-
interesting to note that in solving the linear car following oscillatory.
equation for two vehicles, estimates for the local stability
condition were first obtained using an analog computer for In Figure 4.3, the inter-vehicle spacing is shown for four other
different values of C which differentiate the various type of values of C for the same fluctuation of the lead vehicle as shown
motion. in Figure 4.2. The values of C range over the cases of oscillatory
Note: Vehicle 2 follows Vehicle 1 (lead car) with a time lag T=1.5 sec and a value of C=e-1(0.368), the limiting value for local
stability. The initial velocity of each vehicle is u
Figure 4.2
Detailed Motion of Two Cars Showing the
Effect of a Fluctuation in the Acceleration of the Lead Car (Herman et al. 1958).

&$5 )2//2:,1* 02'(/6
Note: Changes in car spacings from an original constant spacing between two cars for the noted values of C. The a cceleration
profile of the lead car is the same as that shown in Figure 4.2.
Figure 4.3
Changes in Car Spacings from an
Original Constant Spacing Between Two Cars (Herman et al. 1958).
motion where the amplitude is damped, undamped, and of inter-vehicle spacing. For m = 1, we obtain the linear car
increasing amplitude. following equation.
For the values of C = 0.5 and 0.80, the spacing is oscillatory and Using the identical analysis for any m, the equation whose roots
heavily damped. determine the character of the motion which results from

For C = 1.57 ( ), Equation 4.21 is
2
the spacing oscillates with constant amplitude. For C = 1.60, the C s me s
0 (4.22)
motion is oscillatory with increasing amplitude.
Local Stability with Other Controls. Qualitative arguments can None of these roots lie on the negative real axis when m is even,
be given of a driver's lack of sensitivity to variation in relative therefore, local stability is possible only for odd values of the
acceleration or higher derivatives of inter-vehicle spacings mth derivative of spacing: relative speed, the first derivative of
because of the inability to make estimates of such quantities. It relative acceleration (m = 3), etc. Note that this result indicates
is of interest to determine whether a control centered around that an acceleration response directly proportional to inter-
such derivatives would be locally stable. Consider the car vehicle spacing stimulus is unstable.
following equation of the form
dm
x¨f(-1)
C [x#(-) xf(-)] (4.21) 4.2.2 Asymptotic Stability
dt m
In the previous analysis, the behavior of one vehicle following
another was considered. Here a platoon of vehicles (except for
for m= 0,1,2,3..., i.e., a control where the acceleration of the the platoon leader) follows the vehicle ahead according to the
following vehicle is proportional to the mth derivative of the

$5 2//2:,1* 2'(/6
linear car following equation, Equation 4.15. The criteria which decreases with increasing n if
necessary for asymptotic stability or instability were first
investigated by considering the Fourier components of the speed
7 7
1( )22( )sin(7,) > 1
fluctuation of a platoon leader (Chandler et al. 1958).
The set of equations which attempts to describe a line of N i.e. if

identical car-driver units is:
7 > 2sin(7,)
ẍn 1(tT)
[x n(t) x n 1(t)] (4.23)
The severest restriction on the parameter arises from the low

where n =0,1,2,3,...N. frequency range, since in the limit as 7 0, must satisfy the
inequality
Any specific solution to these equations depends on the velocity
profile of the lead vehicle of the platoon, u0(t), and the two , < 1 [lim70(7,)/sin(7,)] (4.27)
2
parameters and T. For any inter-vehicle spacing, if a
disturbance grows in amplitude then a 'collision' would
eventually occur somewhere back in the line of vehicles. Accordingly, asymptotic stability is insured for all frequencies
where this inequality is satisfied.
While numerical solutions to Equation 4.23 can determine at
what point such an event would occur, the interest is to For those values of 7 within the physically realizable frequency
determine criteria for the growth or decay of such a disturbance. range of vehicular speed oscillations, the right hand side of the
Since an arbitrary speed pattern can be expressed as a linear inequality of 4.27 has a short range of values of 0.50 to about
combination of monochromatic components by Fourier analysis, 0.52. The asymptotic stability criteria divides the two parameter
the specific profile of a platoon leader can be simply represented domain into stable and unstable regions, as graphically
by one component, i.e., by a constant together with a illustrated in Figure 4.4.
monochromatic oscillation with frequency, 7 and amplitude, fo
, i.e., The criteria for local stability (namely that no local oscillations
occur when , e-1) also insures asymptoticstability. It has also
uo(t)
ao f o e i7t (4.24) been shown (Chandler et al. 1958) that a speed fluctuation can
be approximated by:
4%n
and the speed profile of the nth vehicle by xn 1(t)u0(t)
1
T
½
2
un(t)
ao f n e i7t [t n/] (4.28)
(4.25) exp
4n/(1/2 ,)
Substitution of Equations 4.24 and 4.25 into Equation 4.23 Hence, the speed of propagation of the disturbance with respect
yields: to the moving traffic stream in number of inter-vehicle
separations per second, n/t, is .
un(t)
ao F(7,,,,n)e i6(7,,,,n) (4.26)
That is, the time required for the disturbance to propagate
between pairs of vehicles is -1, a constant, which is independent
where the amplitude factor F (7, , ,, n) is given by of the response time T. It is noted from the above equation that
in the propagation of a speed fluctuation the amplitude of the
7 7
[1( )22( )sin(7,)] n/2

$5 2//2:,1* 2'(/6
1.00
c=0.52
Unstable
0.75
0.50
Stable
c=0.50
0.25
0 0.5 1.0 1.5 2.0

T (sec)
Figure 4.4
Regions of Asymptotic Stability (Rothery 1968).
disturbance grows as the response time, T, approaches 1/(2) through the line of vehicles even though this case is at the
until instability is reached. Thus, while asymptotic limit.
T < 0.5 ensures stability, short reaction times increase the range
of the sensitivity coefficient, , that ensures stability. From a This results from the fact that the disturbance is not a single
practical viewpoint, small reaction times also reduce relatively Fourier component with near zero frequency. However,
large responses to a given stimulus, or in contrast, larger instability is clearly exhibited in the third case of Figure 4.5
response times require relatively large responses to a given where C = 0.75 and in Figure 4.6 where C = 0.8. In the case
stimulus. Acceleration fluctuations can be correspondingly shown in Figure 4.6, the trajectories of each vehicle in a platoon
analyzed (Chandler et al. 1958). of nine are graphed with respect to a coordinate system moving
with the initial platoon speed u. Asymptotic instability of a
platoon of nine cars is illustrated for the linear car following
4.2.1.1 Numerical Examples equation, Equation 4.23, where C = 0.80. For t = 0, the vehicles
are all moving with a velocity u and are separated by a distance
In order to illustrate the general theory of asymptotic stability as of 12 meters. The propagation of the disturbance, which can be
outlined above, the results of a number of numerical calculations readily discerned, leads to "collision" between the 7th and 8th
are given. Figure 4.5 graphically exhibits the inter-vehicle cars at about t = 24 sec. The lead vehicle at t = 0 decelerates
spacings of successive pairs of vehicles versus time for a platoon for 2 seconds at 4 km/h/sec, so that its speed changes from u to
of vehicles. Here, three values of C were used: C = 0.368, 0.5, u -8 km/h and then accelerates back to u. This fluctuation in the
and 0.75. The initial fluctuation of the lead vehicle, n = 1, was speed of the lead vehicle propagates through the platoon in an
the same as that of the lead vehicle illustrated in Figure 4.2. This unstable manner with the inter-vehicle spacing between the
disturbance consists of a slowing down and then a speeding up seventh and eighth vehicles being reduced to zero at about 24.0
to the original speed so that the integral of acceleration over time sec after the initial phase of the disturbance is generated by the
is zero. The particularly stable, non-oscillatory response is lead vehicle of the platoon.
evident in the first case where C = 0.368 (1/e), the local
stability limit. As analyzed, a heavily damped oscillation occurs In Figure 4.7 the envelope of the minimum spacing that occurs
in the second case where C = 0.5, the asymptotic limit. Note that between successive pairs of vehicles is graphed versus time
the amplitude of the disturbance is damped as it propagates

$5 2//2:,1* 2'(/6
Note: Diagram uses Equation 4.23 for three values of C. The fluctuation in acceleration of the lead car, car number 1, is the
same as that shown in Fig. 4.2 At t=0 the cars are separated by a spacing of 21 meters.
Figure 4.5
Inter-Vehicle Spacings of a Platoon of Vehicles
Versus Time for the Linear Car Following Model (Herman et al. 1958).
where the lead vehicle's speed varies sinusoidally with a point. Here, the numerical solution yields a maximum and
frequency 7 =2%/10 radian/sec. The envelope of minimum minimum amplitude that is constant to seven significant places.
inter-vehicle spacing versus vehicle position is shown for three
values of . The response time, T, equals 1 second. It has been
shown that the frequency spectrum of relative speed and 4.2.1.2 Next-Nearest Vehicle Coupling
acceleration in car following experiments have essentially all
their content below this frequency (Darroch and Rothery 1973). In the nearest neighbor vehicle following model, the motion of
each vehicle in a platoon is determined solely by the motion of
The values for the parameter were 0.530, 0.5345, and the vehicle directly in front. The effect of including the motion
0.550/sec. The value for the time lag, T, was 1 sec in each case. of the "next nearest neighbor" vehicle (i.e., the car which is two
The frequency used is that value of 7 which just satisfies vehicles ahead in addition to the vehicle directly in front) can be
the stability equation, Equation 4.27, for the case where ascertained. An approximation to this type of control, is the
= 0.5345/sec. This latter figure serves to demonstrate not only model
the stability criteria as a function of frequency but the accuracy
of the numerical results. A comparison between that which is ẍn2(t,)
1[x n1(t) x n2(t)]2[x n(t) x n2] (4.29)
predicted from the stability analysis and the numerical solution
for the constant amplitude case (=0.5345/sec) serves as a check

$5 2//2:,1* 2'(/6
Note: Diagram illustrates the linear car following equation, eq. 4.23, where C=080.
Figure 4.6
Asymptotic Instability of a Platoon of Nine Cars (Herman et al. 1958).
Figure 4.7
Envelope of Minimum Inter-Vehicle Spacing Versus Vehicle Position (Rothery 1968).

$5 2//2:,1* 2'(/6
1
(12), < (7,)/sin(7,)] (4.30) (12), >
1
(4.31)
2 2
which in the limit 7 0 is This equation states that the effect of adding next nearest
neighbor coupling to the control element is, to the first order, to
increase 1 to (1 + 2). This reduces the value that 1 can have
and still maintain asymptotic stability.
4.3 Steady-State Flow

This section discusses the properties of steady-state traffic flow also follows from elementary considerations by integration of
based on car following models of single-lane traffic flow. In Equation 4.32 as shown in the previous section (Gazis et al.
particular, the associated speed-spacing or equivalent speed- 1959). This result is not directly dependent on the time lag, T,
concentration relationships, as well as the flow-concentration except that for this result to be valid the time lag, T, must allow
relationships for single lane traffic flow are developed. the equation of motion to form a stable stream of traffic. Since
vehicle spacing is the inverse of traffic stream concentration, k,
The Linear Case. The equations of motion for a single lane of the speed-concentration relation corresponding to Equation 4.33
traffic described by the linear car following model are given by: is:
x¨n1(t,)
[x n(t) x n1(t)] (4.32) kf
ki 1(Uf Ui)
1 1
(4.34)
where n = 1, 2, 3, ....
The significance of Equations 4.33 and 4.34 is that:
In order to interrelate one steady-state to another under this
control, assume (up to a time t=0) each vehicle is traveling at a 1) They link an initial steady-state to a second arbitrary
speed Ui and that the inter-vehicle spacing is S i. Suppose that steady-state, and
at t=0, the lead vehicle undergoes a speed change and increases
or decreases its speed so that its final speed after some time, t, is 2) They establish relationships between macroscopic traffic
U f . A specific numerical solution of this type of transition is stream variables involving a microscopic car following
exhibited in Figure 4.8. parameter, .
In this example C = T=0.47 so that the stream of traffic is In this respect they can be used to test the applicability of the car
stable, and speed fluctuations are damped. Any case where the following model in describing the overall properties of single
asymptotic stability criteria is satisfied assures that each lane traffic flow. For stopped traffic, Ui = 0, and the
following vehicle comprising the traffic stream eventually corresponding spacing, So, is composed of vehicle length and
reaches a state traveling at the speed Uf . In the transition from "bumper-to-bumper" inter-vehicle spacing. The concentration
a speed Ui to a speed U f , the inter-vehicle spacing S changes corresponding to a spacing, So, is denoted by k j and is frequently
from Si to Sf , where referred to as the 'jam concentration'.
Sf
Si 1(Uf Ui) (4.33) Given k,j then Equation 4.34 for an arbitrary traffic state defined
by a speed, U, and a concentration, k, becomes
This result follows directly from the solution to the car following
equation, Equation 4.16a or from Chow (1958). Equation 4.33

$5 2//2:,1* 2'(/6
Note: A numerical solution to Equation 4.32 for the inter-vehicle spacings of an 11- vehicle platoon going from one steady-sta te to
another (T = 0.47). The lead vehicle's speed decreases by 7.5 meters per second.
Figure 4.8
Inter-Vehicle Spacings of an Eleven Vehicle Platoon (Rothery 1968).
versus a normalized concentration together with the

corresponding theoretical steady-state result derived from
U
(k 1 kj )
1
(4.35) Equation 4.35, i.e.,
k
q
Uk
(1 ) (4.36)
A comparison of this relationship was made (Gazis et al. 1959) kj
with a specific set of reported observations (Greenberg 1959) for
a case of single lane traffic flow (i.e., for the northbound traffic
flowing through the Lincoln Tunnel which passes under the The inability of Equation 4.36 to exhibit the required qualitative
Hudson River between the States of New York and New Jersey). relationship between flow and concentration (see Chapter 2) led
This comparison is reproduced in Figure 4.9 and leads to an to the modification of the linear car following equation (Gazis et
estimate of 0.60 sec -1 for . This estimate of implies an upper al. 1959).
bound for T 0.83 sec for an asymptotic stable traffic stream
using this facility. Non-Linear Models. The linear car following model specifies
an acceleration response which is completely independent of
While this fit and these values are not unreasonable, a vehicle spacing (i.e., for a given relative velocity, response is the
fundamental problem is identified with this form of an equation same whether the vehicle following distance is small [e.g., of the
for a speed-spacing relationship (Gazis et al. 1959). Because it order of 5 or 10 meters] or if the spacing is relatively large [i.e.,
is linear, this relationship does not lead to a reasonable of the order of hundreds of meters]). Qualitatively, we would
description of traffic flow. This is illustrated in Figure 4.10 expect that response to a given relative speed to increase with
where the same data from the Lincoln Tunnel (in Figure 4.9) is smaller spacings.
regraphed. Here the graph is in the form of a normalized flow,

$5 2//2:,1* 2'(/6
Note: The data are those of (Greenberg 1959) for the Lincoln Tunnel. The curve represents a "least squares fit" of Equation 4.35
to the data.
Figure 4.9
Speed (miles/hour) Versus Vehicle Concentration (vehicles/mile).(Gazis et al. 1959).
Note: The curve corresponds to Equation 4.36 where the parameters are those from the "fit" shown in Figure 4.9.
Figure 4.10
Normalized Flow Versus Normalized Concentration (Gazis et al. 1959).

&$5 )2//2:,1* 02'(/6
In order to attempt to take this effect into account, the linear therefore the flow-concentration relationship, does not describe
model is modified by supposing that the gain factor, , is not a the state of the traffic stream.
constant but is inversely proportional to vehicle spacing, i.e.,
A solution to any particular set of equations for the motion of a

1 /S(t)
1 /[xn(t) xn1(t)] (4.37) traffic stream specifies departures from the steady-state. This is
not the case for simple headway models or hydro-dynamical
where 1 is a new parameter - assumed to be a constant and approaches to single-lane traffic flow because in these cases any
which shall be referred to as the sensitivity coefficient. Using small speed change, once the disturbance arrives, each vehicle
Equation 4.37 in Equation 4.32, our car following equation is: instantaneously relaxes to the new speed, at the 'proper' spacing.
1
ẍn1(t,)
[x (t) x n1(t)] (4.38) This emphasizes the shortcoming of these alternate approaches.
[xn(t) xn1(t)] n They cannot take into account the behavioral and physical
aspects of disturbances. In the case of car following models, the
initial phase of a disturbance arrives at the nth vehicle
for n = 1,2,3,... downstream from the vehicle initiating the speed change at a
time (n-1)T seconds after the onset of the fluctuation. The time
As before, by assuming the parameters are such that the traffic it takes vehicles to reach the changed speed depends on the
stream is stable, this equation can be integrated yielding the parameter , for the linear model, and 1, for the non-linear
steady-state relation for speed and concentration: model, subject to the restriction that -1 > T or 1 < S/T,
respectively.
u
1ln (kj /k) (4.39)
These restrictions assure that the signal speed can never precede
and for steady-state flow and concentration: the initial phase speed of a disturbance. For the linear case, the
restriction is more than satisfied for an asymptotic stable traffic
stream. For small speed changes, it is also satisfied for the non-
q
1kln(kj /k) (4.40) linear model by assuming that the stability criteria results for the
linear case yields a bound for the stability in the non-linear case.
where again it is assumed that for u=0, the spacing is equal to Hence, the inequality , /S*<0.5 provides a sufficient stability
an effective vehicle length, L = k-1. These relations for steady- condition for the non-linear case, where S* is the minimum
state flow are identical to those obtained from considering the spacing occurring during a transition from one steady-state to
traffic stream to be approximated by a continuous compressible another.
fluid (see Chapter 5) with the property that disturbances are
propagated with a constant speed with respect to the moving Before discussing a more general form for the sensitivity
medium (Greenberg 1959). For our non-linear car following coefficient (i.e., Equation 4.37), the same reported data
equation, infinitesimal disturbances are propagated with speed (Greenberg 1959) plotted in Figures 4.9 and 4.10 are graphed in
1 . This is consistent with the earlier discussion regarding the Figures 4.11 and 4.12 together with the steady-state relations
speed of propagation of a disturbance per vehicle pair. (Equations 4.39 and 4.40 obtained from the non-linear model,
Equation 4.38). The fit of the data to the steady-state relation via
It can be shown that if the propagation time, ,0, is directly the method of "least squares" is good and the resulting values for
proportional to spacing (i.e., ,0 S), Equations 4.39 and 4.40 1 and k j are 27.7 km/h and 142 veh/km, respectively.
are obtained where the constant ratio S /,o is identified as the Assuming that this data is a representative sample of this
constant l. facility's traffic, the value of 27.7 km/h is an estimate not only of
the sensitivity coefficient for the non-linear car following model
These two approaches are not analogous. In the fluid analogy but it is the 'characteristic speed' for the roadway under
case, the speed-spacing relationship is 'followed' at every instant consideration (i.e., the speed of the traffic stream which
before, during, and after a disturbance. In the case of car maximizes the flow).
following during the transition phase, the speed-spacing, and

&$5 )2//2:,1* 02'(/6
Note: The curve corresponds to a "least squares" fit of Equation 4.39 to the data (Greenberg 1959).
Figure 4.11
Speed Versus Vehicle Concentration (Gazis et al. 1959).
Note: The curve corresponds to Equation 4.40 where parameters are those from the "fit" obtained in Figure 4.11.
Figure 4.12
Normalized Flow Versus Normalized Vehicle Concentration (Edie et al. 1963).

&$5 )2//2:,1* 02'(/6
The corresponding vehicle concentration at maximum flow, i.e.,

when u = 1 , is e-l kj. This predicts a roadway capacity of U
Uf for 0kkf (4.44)
1 e-l kj of about 1400 veh/h for the Lincoln Tunnel. A noted
undesirable property of Equation 4.40 is that the tangent dq/dt
is infinite at k = 0, whereas a linear relation between flow and and
concentration would more accurately describe traffic near zero
concentration. This is not a serious defect in the model since car k kf
following models are not applicable for low concentrations U
Uf exp (4.45)
where spacings are large and the coupling between vehicles are
km
weak. However, this property of the model did suggest the
following alternative form (Edie 1961) for the gain factor, where kf corresponds to a concentration where vehicle to
vehicle interactions begin to take place so that the stream speed
.

2xn1(t,)/[xn(t) xn1(t)]2 begins to decrease with increasing concentration. Assuming that
interactions take place at a spacing of about 120 m, kf would
have a value of about 8 veh/km. A "kink" of this kind was
This leads to the following expression for a car following model: introduced into a linear model for the speed concentration
relationship (Greenshields 1935).
.
2xn1(t,) . . (4.41) Greenshields' empirical model for a speed-concentration relation
ẍn1(t,)
[x (t) xn1(t)] is given by
[xn(t) xn1(t)] 2 n
U
Uf (1 k/kj) (4.46)
As before, this can be integrated giving the following steady-
state equations:
k/km where Uf is a “free mean speed” and kj is the jam concentration.
U
Uf e (4.42)
It is of interest to question what car following model would
correspond to the above steady-state equations as expressed by
and Equation 4.46. The particular model can be derived in the
following elementary way (Gazis et al. 1961). Equation 4.46 is
k/km rewritten as
q
Uf ke (4.43)
U
Uf (1 L/S) (4.47)
where U f is the "free mean speed", i.e., the speed of the traffic
stream near zero concentration and km is the concentration when Differentiating both sides with respect to time obtains
the flow is a maximum. In this case the sensitivity coefficient, 2
can be identified as km-1. The speed at optimal flow is e-1Uf
which, as before, corresponds to the speed of propagation of a U
(Uf L/S 2)S (4.48)
disturbance with respect to the moving traffic stream. This
model predicts a finite speed, Uf , near zero concentration.
which after introduction of a time lag is for the (n+1) vehicle:
Ideally, this speed concentration relation should be translated to
the right in order to more completely take into account
Uf L
observations that the speed of the traffic stream is independent x¨n1(t,)
[x n(t) x n1(t)] (4.49)
of vehicle concentration for low concentrations, .i.e. [xn(t) xn1(t)]2

$5 2//2:,1* 2'(/6
The gain factor is: constants consistent with the physical restrictions and where
fp(x), (p = m or #), is given by
Uf L
(4.50) f p(x)
x 1 p (4.54)
[xn(t) xn1(t)]2
for p C 1 and
The above procedure demonstrates an alternate technique at
arriving at stimulus response equations from relatively
f p(x)
#nx (4.55)
elementary considerations. This method was used to develop
early car following models (Reuschel 1950; Pipes 1951). The
technique does pre-suppose that a speed-spacing relation reflects for p = 1. The integration constant b is related to the "free mean
detailed psycho-physical aspects of how one vehicle follows speed" or the "jam concentration" depending on the specific
another. To summarize the car-following equation considered, values of m and #. For m > 1, #C 1, or m =1, # >1
we have:
b
f m(Uf ) (4.56)
x¨n1(t,)
[x n(t) x n1(t)] (4.51)
and
where the factor, , is assumed to be given by the following:
b
af l(L) (4.57)
A constant, = 0;
A term inversely proportional to the spacing, = 1/S; for all other combinations of m and #, except # < 1 and
A term proportional to the speed and inversely m = 1.
proportional to the spacing squared, = 2U/S2; and
A term inversely proportional to the spacing squared, For those cases where # < 1 and m = 1 it is not possible to satisfy
= 3 / S 2. either the boundary condition at k = 0 or kj and the integration
constant can be assigned arbitrarily, e.g., at km, the concentration
These models can be considered to be special cases of a more at maximum flow or more appropriately at some 'critical'
general expression for the gain factor, namely: concentration near the boundary condition of a "free speed"
determined by the "kink" in speed-concentration data for the

a ,mx nm1(t,)/[xn(t) xn1(t)]
#
#
(4.52) particular facility being modeled. The relationship between km
and kj is a characteristic of the particular functional or model
being used to describe traffic flow of the facility being studied
where a ,m is a constant to be determined experimentally. Model
# and not the physical phenomenon involved. For example, for the
specification is to be determined on the basis of the degree to two models given by # = 1, m = 0, and # = 2, m = 0, maximum
which it presents a consistent description of actual traffic flow occurs at a concentration of e-l kj and kj / 2 , respectively.
phenomena. Equations 4.51 and 4.52 provide a relatively broad Such a result is not physically unrealistic. Physically the
framework in so far as steady-state phenomena is concerned question is whether or not the measured value of q max occurs at
(Gazis et al. 1961). or near the numerical value of these terms, i.e., km = e-1kj or kj/2
for the two examples cited.
Using these equations and integrating over time we have
f m(U)
a f l(S)b (4.53)
Using Equations 4.53, 4.54, 4.55, 4.56, 4.57, and the definition
of steady-state flow, we can obtain the relationships between
speed, concentration, and flow. Several examples have been
where, as before, U is the steady-state speed of the traffic stream, given above. Figures 4.13 and 4.14 contain these and additional
S is the steady-state spacing, and a and b are appropriate examples of flow versus concentration relations for various

$5 2//2:,1* 2'(/6
Note: Normalized flow versus normalized concentration corresponding to the steady-state solution of Equations 4.51 and 4.52
for m=1 and various values of #.
Figure 4.13
Normalized Flow Versus Normalized Concentration (Gazis et al. 1963).
Figure 4.14
Normalized Flow versus Normalized Concentration Corresponding to the Steady-State
Solution of Equations 4.51 and 4.52 for m=1 and Various Values of # (Gazis 1963).

$5 2//2:,1* 2'(/6
values of # and m. These flow curves are normalized by letting considering the perceptual factors that are related to the car
qn = q/qmax, and kn = k/kj. following task (Pipes and Wojcik 1968; Fox and Lehman 1967;
Michaels 1963). As was discussed earlier, the case for m = 0,
It can be seen from these figures that most of the models shown # = 1 generates a steady-state relation that can be developed bya
here reflect the general type of flow diagram required to agree fluid flow analogy to traffic (Greenberg 1959) and led to the
with the qualitative descriptions of steady-state flow. The reexamination of car following experiments and the hypothesis
spectrum of models provided are capable of fitting data like that that drivers do not have a constant gain factor to a given relative-
shown in Figure 4.9 so long as a suitable choice of the speed stimulus but rather that it varies inversely with the vehicle
parameters is made. spacing, i.e., m= 0, # =1 (Herman et al. 1959). A generalized
equation for steady-state flow (Drew 1965) and subsequently
The generalized expression for car following models, Equations tested on the Gulf Freeway in Houston, Texas led to a model
4.51 and 4.52, has also been examined for non-integral values where m = 0 and # = 3/2.
for m and # (May and Keller 1967). Fittingdata obtained on the
Eisenhower Expressway in Chicago they proposed a model with As noted earlier, consideration of a "free-speed" near low
m = 0.8 and # = 2.8. Various values for m and # can be identified concentrations led to the proposal and subsequent testing of the
in the early work on steady-state flow and car following . model m = 1, # = 2 (Edie 1961). Yet another model, m = 1,
# = 3 resulted from analysis of data obtained on the Eisenhower
The case m = 0, # = 0 equates to the "simple" linear car following Expressway in Chicago (Drake et al. 1967). Further analysis of
model. The case m = 0, # = 2 can be identified with a model this model together with observations suggest that the sensitivity
developed from photographic observations of traffic flow made coefficient may take on different values above a lane flow of
in 1934 (Greenshields 1935). This model can also be developed about 1,800 vehicles/hr (May and Keller 1967).
4.4 Experiments And Observations

This section is devoted to the presentation and discussion of derived from car following models for steady-state flow are
experiments that have been carried out in an effort to ascertain examined.
whether car following models approximate single lane traffic
characteristics. These experiments are organized into two Finally, the degree to which any specific model of the type
distinct categories. examined in the previous section is capable of representing a
consistent framework from both the microscopic and
The first of these is concerned with comparisons between car macroscopic viewpoints is examined.
following models and detailed measurements of the variables
involved in the driving situation where one vehicle follows
another on an empty roadway. These comparisons lead to a 4.4.1 Car Following Experiments
quantitative measure of car following model estimates for the
specific parameters involved for the traffic facility and vehicle The first experiments which attempted to make a preliminary
type used. evaluation of the linear car following model were performed a
number of decades ago (Chandler et al. 1958; Kometani and
The second category of experiments are those concerned with Sasaki 1958). In subsequent years a number of different tests
the measurement of macroscopic flow characteristics: the study with varying objectives were performed using two vehicles,
of speed, concentration, flow and their inter-relationships for three vehicles, and buses. Most of these tests were conducted on
vehicle platoons and traffic environments where traffic is test track facilities and in vehicular tunnels.
channeled in a single lane. In particular, the degree to which this
type of data fits the analytical relationships that have been

$5 2//2:,1* 2'(/6
In these experiments, inter-vehicle spacing, relative speed, speed 5) Miscellaneous Experiments. Several additional car
of the following vehicle, and acceleration of the following following experiments have been performed and reported
vehicles were recorded simultaneously together with a clock on as follows:
signal to assure synchronization of each variable with every
other. a) Kometani and Sasaki Experiments. Kometani and
Sasaki conducted and reported on a series of experiments
These car following experiments are divided into six specific that were performed to evaluate the effect of an additional
categories as follows: term in the linear car following equation. This term is
related to the acceleration of the lead vehicle. In
1) Preliminary Test Track Experiments. The first particular, they investigated a model rewritten here in the
experiments in car following were performed by (Chandler following form:
et al. 1958) and were carried out in order to obtain
x¨n1(t,)
[x n(t) x n1(t)]ẍn(t) (4.58)
estimates of the parameters in the linear car following
model and to obtain a preliminary evaluation of this model.
Eight male drivers participated in the study which was This equation attempts to take into account a particular
conducted on a one-mile test track facility. driving phenomenon, where the driver in a particular state
realizes that he should maintain a non-zero acceleration
2) Vehicular Tunnel Experiments. To further establish the even though the relative speed has been reduced to zero or
validity of car following models and to establish estimates, near zero. This situation was observed in several cases in
the parameters involved in real operating environments tests carried out in the vehicular tunnels - particularly
where the traffic flow characteristics were well known, a when vehicles were coming to a stop. Equation 4.58
series of experiments were carried out in the Lincoln, above allows for a non-zero acceleration when the relative
Holland, and Queens Mid-Town Tunnels of New York speed is zero. A value of near one would indicate an
City. Ten different drivers were used in collecting 30 test attempt to nearly match the acceleration of the lead driver
runs. for such cases. This does not imply that drivers are good
estimators of relative acceleration. The conjecture here is
3) Bus Following Experiments. A series of experiments that by pursuing the task where the lead driver is
were performed to determine whether the dynamical undergoing a constant or near constant acceleration
characteristics of a traffic stream changes when it is maneuver, the driver becomes aware of this qualitatively
composed of vehicles whose performance measures are after nullifying out relative speed - and thereby shifts the
significantly different than those of an automobile. They frame of reference. Such cases have been incorporated
were also performed to determine the validity and measure into models simulating the behavior of bottlenecks in
parameters of car following models when applied to heavy tunnel traffic (Helly 1959).
vehicles. Using a 4 kilometer test track facility and 53-
passenger city buses, 22 drivers were studied. b) Experiments of Forbes et al. Several experiments
using three vehicle platoons were reported by Forbes
4) Three Car Experiments. A series of experiments were et al. (1957). Here a lead vehicle was driven by one of the
performed to determine the effect on driver behavior when experimenters while the second and third vehicles were
there is an opportunity for next-nearest following and of driven by subjects. At predetermined locations along the
following the vehicle directly ahead. The degree to which roadway relatively severe acceleration maneuvers were
a driver uses the information that might be obtained from executed by the lead vehicle. Photographic equipment
a vehicle two ahead was also examined. The relative recorded these events with respect to this moving
spacings and the relative speeds between the first and third reference together with speed and time. From these
vehicles and the second and third vehicles together with recordings speeds and spacings were calculated as a
the speed and acceleration of the third vehicle were function of time. These investigators did not fit this data
recorded. to car following models. However, a partial set of this data
was fitted to vehicle following models by another
investigator (Helly 1959). This latter set consisted of six

$5 2//2:,1* 2'(/6
tests in all, four in the Lincoln Tunnel and two on an for which the correlation coefficient is a maximum and typically
open roadway. falls in the range of 0.85 to 0.95.
c) Ohio State Experiments. Two different sets of The results from the preliminary experiments (Chandler et al.
experiments have been conducted at Ohio State 1958) are summarized in Table 4.1 where the estimates are
University. In the first set a series of subjects have been given for , their product; C = T, the boundary value for
studied using a car following simulator (Todosiev 1963). asymptotic stability; average spacing, < S >; and average speed,
An integral part of the simulator is an analog computer . The average value of the gain factor is 0.368 sec-1. The
which could program the lead vehicle for many different average value of T is close to 0.5, the asymptotic stability
driving tasks. The computer could also simulate the boundary limit.
performance characteristics of different following vehicles.
These experiments were directed toward understanding the
manner in which the following vehicle behaves when the Table 4.1 Results from Car-Following Experiment
lead vehicle moves with constant speed and the
measurement of driver thresholds for changes in spacing,
Driver <S> T
relative speed, and acceleration. The second set of
experiments were conducted on a level two-lane state 1 0.74 sec -1 19.8 36 1.04
highway operating at low traffic concentrations (Hankin m/sec m
and Rockwell 1967). In these experiments the purpose
was "to develop an empirically based model of car 2 0.44 16 36.7 0.44
following which would predict a following car's 3 0.34 20.5 38.1 1.52
acceleration and change in acceleration as a function of
observed dynamic relationships with the lead car." As in 4 0.32 22.2 34.8 0.48
the earlier car following experiments, spacing and relative
5 0.38 16.8 26.7 0.65
speed were recorded as well as speed and acceleration of
the following vehicle. 6 0.17 18.1 61.1 0.19
d) Studies by Constantine and Young. These studies 7 0.32 18.1 55.7 0.72
were carried out using motorists in England and a 8 0.23 18.7 43.1 0.47
photographic system to record the data (Constantine and
Young 1967). The experiments are interesting from the
vantage point that they also incorporated a second
photographic system mounted in the following vehicle and Using the values for and the average spacing <S > obtained for
directed to the rear so that two sets of car following data each subject a value of 12.1 m/sec (44.1 km/h) is obtained for an
could be obtained simultaneously. The latter set collected estimate of the constant a1, 0 (Herman and Potts 1959). This
information on an unsuspecting motorist. Although latter estimate compares the value for each driver with that
accuracy is not sufficient, such a system holds promise. driver's average spacing, <S > , since each driver is in somewhat
different driving state. This is illustrated in Figure 4.15. This
approach attempts to take into account the differences in the
4.4.1.1 Analysis of Car Following Experiments estimates for the gain factor or a0,0, obtained for different
drivers by attributing these differences to the differences in their
The analysis of recorded information from a car following respective average spacing. An alternate and more direct
experiment is generally made by reducing the data to numerical approach carries out the correlation analysis for this model using
values at equal time intervals. Then, a correlation analysis is an equation which is the discrete form of Equation 4.38 to obtain
carried out using the linear car following model to obtain a direct estimate of the dependence of the gain factor on spacing,
estimates of the two parameters, and T. With the data in S(t).
discrete form, the time lag ,T, also takes on discrete values. The
time lag or response time associated with a given driver is one

$5 2//2:,1* 2'(/6
Figure 4.15
Sensitivity Coefficient Versus the Reciprocal of the Average Vehicle Spacing (Gazis et al. 1959).
Vehicular Tunnel Experiments. Vehicular tunnels usually have versus the reciprocal of the average vehicle spacing for the tests
roadbeds that are limited to two lanes, one per direction. conducted in the Lincoln and Holland tunnels, respectively.
Accordingly, they consist of single-lane traffic where passing is Figure 4.17, the gain factor, , versus the reciprocal of the
prohibited. In order to investigate the reasonableness of the non- average spacing for the Holland Tunnel tests. The straight line
linear model a series of tunnel experiments were conducted. is a "least-squares" fit through the origin. The slope, which is an
Thirty test runs in all were conducted: sixteen in the Lincoln estimate of a1,0 and equals 29.21 km/h. Figure 4.18 graphs the
Tunnel, ten in the Holland Tunnel and four in the Queens Mid- gain factor, , versus the reciprocal of the average spacing for
Town Tunnel. Initially, values of the parameters for the linear the Lincoln Tunnel tests. The straight line is a "least-squares" fit
model were obtained, i.e., = a0,0 and T. These results are through the origin. These results yield characteristic speeds,
shown in Figure 4.16 where the gain factor, = a0,0 versus the a 1,0 , which are within ± 3km/h for these two similar facilities.
time lag, T, for all of the test runs carried out in the vehicular Yet these small numeric differences for the parameter al,0
tunnels. The solid curve divides the domain of this two properly reflect known differences in the macroscopic traffic
parameter field into asymptotically stable and unstable regions. flow characteristics of these facilities.
It is of interest to note that in Figure 4.16 that many of the drivers The analysis was also performed using these test data and the
fall into the unstable region and that there are drivers who have non-linear reciprocal spacing model. The results are not
relatively large gain factors and time lags. Drivers with strikingly different (Rothery 1968). Spacing does not change
relatively slow responses tend to compensatingly have fast significantly within any one test run to provide a sensitive
movement times and tend to apply larger brake pedal forces measure of the dependency of the gain factor on inter-vehicular
resulting in larger decelerations. spacing for any given driver (See Table 4.2). Where the
variation in spacings were relatively large (e.g., runs 3, 11, 13,
Such drivers have been identified, statistically, as being involved and 14) the results tend to support the spacing dependent model.
more frequently in "struck-from-behind accidents" (Babarik This time-dependent analysis has also been performed for seven
1968; Brill 1972). Figures 4.17 and 4.18 graph the gain factor additional functions for the gain factor for the same fourteen

$5 2//2:,1* 2'(/6
Table 4.2
Comparison of the Maximum Correlations obtained for the Linear and Reciprocal Spacing Models for the Fourteen
Lincoln Tunnel Test Runs
Number r0,0 rl,0 <S> (m) S (m) Number r0,0 rl,0 <S> (m) S (m)
1 0.686 0.459 13.4 4.2 8 0.865 0.881 19.9 3.4
2 0.878 0.843 15.5 3.9 9 0.728 0.734 7.6 1.8
3 0.77 0.778 20.6 5.9 10 0.898 0.898 10.7 2.3
4 0.793 0.748 10.6 2.9 11 0.89 0.966 26.2 6.2
5 0.831 0.862 12.3 3.9 12 0.846 0.835 18.5 1.3
6 0.72 0.709 13.5 2.1 13 0.909 0.928 18.7 8.8
7 0.64 0.678 5.5 3.2 14 0.761 0.79 46.1 17.6
Figure 4.16
Gain Factor, , Versus the Time Lag, T, for All of the Test Runs (Rothery 1968).

&$5 )2//2:,1* 02'(/6
Note: The straight line is a "least-squares" fit through the origin. The slope, which is an estimate of a1,0, equals 29.21 km/h.
Figure 4.17
Gain Factor, , Versus the Reciprocal of the
Average Spacing for Holland Tunnel Tests (Herman and Potts 1959).
Note: The straight line is a "least-squares" fit through the origin. The slope, is an estimate of a1,0, equals 32.68 km/h.
Figure 4.18
Gain Factor, ,Versus the Reciprocal of the
Average Spacing for Lincoln Tunnel Tests (Herman and Potts 1959).

&$5 )2//2:,1* 02'(/6
little difference from one model to the other. There are definite of the cases when that factor is introduced and this model (#=1;
trends however. If one graphs the correlation coefficient for a m=0) provided the best fit to the data. The principle results of
given #, say #=1 versus m; 13 of the cases indicate the best fits the analysis are summarized in Figure 4.19 where the sensitivity
are with m = 0 or 1. Three models tend to indicate marginal coefficient a 0,0 versus the time lag, T, for the bus following
superiority; they are those given by (#=2; m=1), (#=1; m=0) and experiments are shown. All of the data points obtained in these
(#=2; m=0). results fall in the asymptotically stable
Bus Following Experiments. For each of the 22 drivers tested, region, whereas in the previous automobile experiments
the time dependent correlation analysis was carried out for the approximately half of the points fell into this region. In Figure
linear model (#=0; m=0), the reciprocal spacing model (#=1; 4.19, the sensitivity coefficient, a0,0 , versus the time lag, T, for
m=0), and the speed, reciprocal-spacing-squared model (#=2; the bus following experiments are shown. Some drivers are
m=1). Results similar to the Tunnel analysis were obtained: high represented by more than one test. The circles are test runs by
correlations for almost all drivers and for each of the three drivers who also participated in the ten bus platoon experiments.
models examined (Rothery et al. 1964). The solid curve divides the graph into regions of asymptotic
stability and instability. The dashed lines are boundaries for the
The correlation analysis provided evidence for the reciprocal regions of local stability and instability.
spacing effect with the correlation improved in about 75 percent
Table 4.3
Maximum Correlation Comparison for Nine Models, a ,m, for Fourteen Lincoln Tunnel Test Runs.
#
Driver r(0,0) r(1,-1) r(1,0) r(1,1) r(1,2) r(2,-1) r(2,0) r(2,1) r(2,2)
1 0.686 0.408 0.459 0.693 0.721 0.310 0.693 0.584 0.690
2 0.878 0.770 0.843 0.847 0.746 0.719 0.847 0.827 0.766
3 0.770 0.757 0.778 0.786 0.784 0.726 0.786 0.784 0.797
4 0.793 0.730 0.748 0.803 0.801 0.685 0.801 0.786 0.808
5 0.831 0.826 0.862 0.727 0.577 0.805 0.728 0.784 0.624
6 0.720 0.665 0.709 0.721 0.709 0.660 0.720 0.713 0.712
7 0.640 0.470 0.678 0.742 0.691 0.455 0.745 0.774 0.718
8 0.865 0.845 0.881 0.899 0.862 0.818 0.890 0.903 0.907
9 0.728 0.642 0.734 0.773 0.752 0.641 0.773 0.769 0.759
10 0.898 0.890 0.898 0.893 0.866 0.881 0.892 0.778 0.865
11 0.890 0.952 0.966 0.921 0.854 0.883 0.921 0.971 0.940
12 0.846 0.823 0.835 0.835 0.823 0.793 0.835 0.821 0.821
13 0.909 0.906 0.923 0.935 0.927 0.860 0.935 0.928 0.936
14 0.761 0.790 0.790 0.771 0.731 0.737 0.772 0.783 0.775

&$5 )2//2:,1* 02'(/6
Note: For bus following experiments - Some drivers are represented by more than one test. The circles are test runs by drivers
who also participated in the ten bus platoons experiments. The solid curve divides the graph into regions of asymptotic stability
and instability. The dashed lines are boundaries for the regions of local stability and instability.
Figure 4.19
Sensitivity Coefficient, a0,0 ,Versus the Time Lag, T (Rothery et al. 1964).
The results of a limited amount of data taken in the rain suggest Figures 4.20 and 4.21 graph the values of a0,0 for all test runs
that drivers operate even more stably when confronted with wet versus <S>-1 and <S>-2, respectively. In Figure 4.20, the
road conditions. These results suggest that buses form a highly sensitivity coefficient versus the reciprocal of the average
stable stream of traffic. spacing for each bus following experiment, and the "least-
squares" straight line are shown. The slope of this regression is
The time-independent analysis for the reciprocal-spacing model an estimate of the reciprocal spacing sensitivity coefficient. The
and the speed-reciprocal-spacing-squared model uses the time solid dots and circles are points for two different test runs.
dependent sensitivity coefficient result, a0,0 , the average speed,
, and the average spacing, <S>, for eachof the car In Figure 4.21, the sensitivity coefficient versus the ratio of the
following test cases in order to form estimates of a1,0 and a2,1, average speed to the square of the average spacing for each bus
i.e. by fitting following experiment and the "least-square" straight line are
shown. The slope of this regression is an estimate of the speed-
a1,0
a0,0
reciprocal spacing squared sensitivity coefficient. The solid dots
<S> and circles are data points for two different test runs. The slope
of the straight line in each of these figures give an estimate of
and their respective sensitivity coefficient for the sample population.
For the reciprocal spacing model the results indicate an estimate

a0,0
a2,1 for a1,0 = 52.8 ± .05 m/sec. (58 ± 1.61 km/h) and for the speed-
<S>2 reciprocal spacing squared model a2,1 = 54.3 ± 1.86 m. The
errors are one standard deviation.
respectively (Rothery et al. 1964).

$5 2//2:,1* 2'(/6
Note: The sensitivity coefficient versus the reciprocal of the average spacing for each bus following experiment. The least squares
straight line is shown. The slope of this regression is an estimate of the reciprocal spacing sensitivity coefficient. The solid dots and
circles are data points for two different test runs.
Figure 4.20
Sensitivity Coefficient Versus the Reciprocal of the Average Spacing (Rothery et al. 1964).
Note: The sensitivity coefficient versus the ratio of the average speed to the square of the average spacing for each bus following
experiment. The least squares straight line is shown. The slope of this regression is an estimate of the speed-reciprocal spacing
squared sensitivity coefficient. The solid dots and circles are data points for two different test runs.
Figure 4.21
Sensitivity Coefficient Versus the Ratio of the Average Speed (Rothery et al. 1964).

$5 2//2:,1* 2'(/6
Three Car Experiments. These experiments were carried out in external situation remains constant. The task does not change
an effort to determine, if possible, the degree to which a driver appreciably from cycle to cycle. Accordingly, response times
is influenced by the vehicle two ahead, i.e., next nearest can be reduced and even canceled because of the cyclic nature
interactions (Herman and Rothery 1965). The data collected in of the task.
these experiments are fitted to the car following model:
By the driver recognizing the periodic nature of the task or that
ẍn2(t,)
1[x n1(t) x n2(t)]2[x n(t) x n2(t)] (4.61)
the motion is sustained over a period of time ( 13 sec for the
acceleration phase and 3 sec for the deceleration phase) the
This equation is rewritten in the following form: driver obtains what is to him/her advanced information.
Accordingly, the analysis of these experiments resulted in short
x¨n2(t,)
l [x n1(t) x n2(t)] [x n(t) x n2(t)] (4.62) response times 0.73 sec for low speed (20-40 km/h.) tests and
0.54 sec for high speed (40-80 km/h.) tests. The results also
produced significantly large gain factors. All of the values
where obtained for each of the drivers for T, exceeded the asymptotic
stability limit. Significantly better fits of the data can be made

2/1
using a model which includes the acceleration of the lead vehicle
(See Equation 4.58) relative to the linear model which does not
A linear regression analysis is then conducted for specific values contain such a term. This is not surprising, given the task of
of the parameter . For the case = 0 there is nearest neighbor following the lead vehicle's motion as described above.
coupling only and for >> 1 there is next nearest
A partial set of the experiments conducted by Forbes et al.
neighbor coupling only. Using eight specific values of (0, (1958) were examined by Helly (1959), who fitted test runs to
0.25, 0.50, 1, 5, 10, 100, and ) and a mean response time of the linear vehicle model, Equation 4.41, by varying and T to
1.6 sec, a maximum correlation was obtained when = 0. minimize the quantity:
However, if response times are allowed to vary, modest
M [x
improvements can be achieved in the correlations. N
Exp.
n t) x nTheor..(j. t)]2
(j. (4.63)
j
1
While next nearest neighbor couplings cannot be ruled out
entirely from this study, the results indicate that there are no
significant changes in the parameters or the correlations when where the data has been quantitized at fixed increments of t,
following two vehicles and that the stimulus provided by the N t is the test run duration, x nsu Exp. (j. t) is the experimentally
nearest neighbor vehicle, i.e., the 'lead' vehicle, is the most measured values for the speed of the following vehicle at time
significant. Models incorporating next nearest neighbor j t, and x n Theor..(j. t) is the theoretical estimate for the speed of
interactions have also been used in simulation models (Fox and the following vehicle as determined from the experimentally
Lehman 1967). The influence of including such interactions in measured values of the acceleration of the following vehicle and
simulations are discussed in detail by those authors. the speed of the lead vehicle using the linear model. These
results are summarized in Table 4.4.
Miscellaneous Car Following Experiments. A brief discussion
of the results of three additional vehicle following experiments Ohio State Simulation Studies. From a series of experiments
are included here for completeness. conducted on the Ohio State simulator, a relatively simple car
following model has been proposed for steady-state car
The experiments of Kometani and Sasaki (1958) were car following (Barbosa 1961). The model is based on the concept
following experiments where the lead vehicle's task was closely of driver thresholds and can be most easily described by means
approximated by: "accelerate at a constant rate from a speed u to of a 'typical' recording of relative speed versus spacing as
a speed u' and then decelerate at a constant rate from the speed
u' to a speed u." This type of task is essentially 'closed' since the

$5 2//2:,1* 2'(/6
Table 4.4 Results from Car Following Experiments large or the spacing becoming too small. At point "A," after a
time lag, the driver initiates this deceleration and reduces the
relative speed to zero. Since drivers have a threshold level
Driver # T a00 rmax below which relative speed cannot be estimated with accuracy,
the driver continues to decelerate until he becomes aware of a
1 1.0 0.7 0.86 positive relative speed because it either exceeds the threshold at
2 0.5 1.3 0.96 this spacing or because the change in spacing has exceeded its
threshold level. At point "2," the driver makes the decision to
3 0.6 0.8 0.91 accelerate in order not to drift away from the lead vehicle. This
decision is executed at point "B" until point "3" is reach and the
4 0.5 1.0 0.87
cycle is more or less repeated. It was found that the arcs, e.g.,
5 0.7 1.1 0.96 AB, BC, etc. are "approximately parabolic" implying that
accelerations can be considered roughly to be constant. These
6 0.5 1.0 0.86 accelerations have been studied in detail in order to obtain
estimates of relative speed thresholds and how they vary with
respect to inter-vehicle spacing and observation times (Todosiev
shown in Figure 4.22. At point "1," it is postulated that the 1963). The results are summarized in Figure 4.23. This driving
driver becomes aware that he is moving at a higher speed than task, following a lead vehicle traveling at constant speed, was
the lead vehicle and makes the decision to decelerate in order to also studied using automobiles in a driving situation so that the
avoid having either the negative relative speed becoming too pertinent data could be collected in a closer-to-reality situation
and then analyzed (Rothery 1968).
Figure 4.22
Relative Speed Versus Spacing (Rothery 1968).

$5 2//2:,1* 2'(/6
Figure 4.23
Relative Speed Thresholds Versus Inter-Vehicle Spacing for
Various Values of the Observation Time. (Rothery 1968).
The interesting element in these latter results is that the character speed stimulus is positive or negative. This effect can be taken
of the motion as exhibited in Figure 4.22 is much the same. into account by rewriting our basic model as:
However, the range of relative speeds at a given spacing that
were recorded are much lower than those measured on the x n1(t,)
i[x n(t) x n1(t)] (4.64)
simulator. Of course the perceptual worlds in these two tests are
considerably different. The three dimensional aspects of the test
track experiment alone might provide sufficient additional cues where i = + or - depending on whether the relative speed is
to limit the subject variables in contrast to the two dimensional greater or less than zero.
CRT screen presented to the 'driver' in the simulator. In any
case, thresholds estimated in driving appear to be less than those A reexamination of about forty vehicle following tests that were
measured via simulation. carried out on test tracks and in vehicular tunnels indicates,
without exception, that such an asymmetry exists (Herman and
Asymmetry Car Following Studies. One car following Rothery 1965). The average value of - is 10 percent greater
experiment was studied segment by segment using a model than +. The reason for this can partly be attributed to the fact
where the stimulus included terms proportional to deviations that vehicles have considerably different capacities to accelerate
from the mean inter-vehicle spacing, deviations from the mean and decelerate. Further, the degree of response is likely to be
speed of the lead vehicle and deviations from the mean speed of different for the situations where vehicles are separating
the following car (Hankin and Rockwell 1967). An interesting compared to those where the spacing is decreasing. This effect
result of the analysis of this model is that it implied an creates a special difficulty with car following models as is
asymmetry in the response depending on whether the relative discussed in the literature (Newell 1962; Newell 1965). One of

$5 2//2:,1* 2'(/6
the principal difficulties is that in a cyclic change in the lead The associated units for these estimates are ft/sec, ft2/sec, and
vehicle's speed - accelerating up to a higher speed and then miles/car, respectively. As illustrated in this table, excellent
returning to the initial speed, the asymmetry in acceleration and agreement is obtained with the reciprocal spacing model. How
deceleration of the following car prevents return to the original well these models fit the macroscopic data is shown in Figure
spacing. With n such cycles, the spacing continues to increase 4.26, where the speed versus vehicle concentration data is
thereby creating a drifting apart of the vehicles. A relaxation graphed together with the curves corresponding to the steady-
process needs to be added to the models that allows for this state speed-concentration relations for the various indicated
asymmetry and also allows for the return to the correct spacing. models. The data appears in Figure 4.24 and 4.25.
The curves are least square estimates. All three models provide
4.4.2 Macroscopic Observations: a good estimate of the characteristic speed (i.e., the speed at
optimum flow, namely 19, 24, and 23 mi/h for the reciprocal
Single Lane Traffic
spacing, reciprocal spacing squared, and speed reciprocal
spacing squared models, respectively).
Several data collections on single lane traffic have been carried
out with the specific purpose of generating a large sample from
Edie's original motivation for suggesting the reciprocal spacing
which accurate estimates of the macroscopic flow characteristics
speed model was to attempt to describe low concentration, non-
could be obtained. With such a data base, direct comparisons
congested traffic. The key parameter in this model is the "mean
can be made with microscopic, car following estimates -
free speed", i.e., the vehicular stream speed as the concentration
particularly when the car following results are obtained on the
goes to zero. The least squares estimate from the macroscopic
same facility as the macroscopic data is collected. One of these
data is 26.85 meters/second.
data collections was carried out in the Holland Tunnel (Edie et
al. 1963). The resulting macroscopic flow data for this 24,000
Edie also compared this model with the macroscopic data in the
vehicle sample is shown in Table 4.5.
concentration range from zero to 56 vehicles/kilometer; the
reciprocal spacing model was used for higher concentrations
The data of Table 4.5 is also shown in graphical form, Figures
(Edie 1961). Of course, the two model fit is better than any one
4.24 and 4.25 where speed versus concentration and flow versus
model fitted over the entire range, but marginally (Rothery
concentration are shown, respectively. In Figure 4.24, speed
1968). Even though the improvement is marginal there is an
versus vehicle concentration for data collected in the Holland
apparent discontinuity in the derivative of the speed-
Tunnel is shown where each data point represents a speed class
concentration curve. This discontinuity is different than that
of vehicles moving with the plotted speed ± 1.61 m/sec. In
which had previously been discussed in the literature. It had
Figure 4.25, flow versus vehicle concentration is shown; the
been suggested that there was an apparent break in the flow
solid points are the flow values derived from the speed classes
concentration curve near maximum flow where the flow drops
assuming steady-state conditions. (See Table 4.5 and Figure
suddenly (Edie and Foote 1958; 1960; 1961). That type of
4.24.) Also included in Figure 4.25 are one-minute average flow
discontinuity suggests that the u-k curve is discontinuous.
values shown as encircled points. (See Edie et al. 1963). Using
this macroscopic data set, estimates for three sensitivity
However, the data shown in the above figures suggest that the
coefficients are estimated for the particular car following models
curve is continuous and its derivative is not. If there is a
that appear to be of significance. These are: a1,0, a 2,1, and a 2,0.
discontinuity in the flow concentration relation near optimum
These are sometimes referred to as the Reciprocal Spacing
flow it is considerably smaller for the Holland Tunnel than has
Model, Edie's Model, and Greenshields' Model, respectively.
been suggested for the Lincoln Tunnel. Nonetheless, the
The numerical values obtained are shown and compared with the
apparent discontinuity suggests that car following may be
microscopic estimates from car following experiments for these
bimodal in character.
same parameters.

$5 2//2:,1* 2'(/6
Table 4.5 Macroscopic Flow Data
Speed Average Spacing Concentration Number of

(m/sec) (m) (veh/km) Vehicles
2.1 12.3 80.1 22
2.7 12.9 76.5 58
3.3 14.6 67.6 98
3.9 15.3 64.3 125
4.5 17.1 57.6 196
5.1 17.8 55.2 293
5.7 18.8 52.6 436
6.3 19.7 50 656
6.9 20.5 48 865
7.5 22.5 43.8 1062
8.1 23.4 42 1267
8.7 25.4 38.8 1328
9.3 26.6 37 1273
9.9 27.7 35.5 1169
10.5 30 32.8 1096
11.1 32.2 30.6 1248
11.7 33.7 29.3 1280
12.3 33.8 26.8 1162
12.9` 43.2 22.8 1087
13.5 43 22.9 1252
14.1 47.4 20.8 1178
14.7 54.5 18.1 1218
15.3 56.2 17.5 1187
15.9 60.5 16.3 1135
16.5 71.5 13.8 837
17.1 75.1 13.1 569
17.7 84.7 11.6 478
18.3 77.3 12.7 291
18.9 88.4 11.1 231
19.5 100.4 9.8 169
20.1 102.7 9.6 55
20.7 120.5 8.1 56

$5 2//2:,1* 2'(/6
Note: Each data point represents a speed class of vehicles moving with the plotted speed ± 1 ft/sec (See Table 4.4).
Figure 4.24
Speed Versus Vehicle Concentration (Edie et al. 1963).
Note: The solid points are the flow values derived from the speed classes assuming steady-state condition. Also included in
Figure 4.25 are one minute average flow values shown as encircled points.
Figure 4.25
Flow Versus Vehicle Concentration (Edie et al. 1963).

&$5 )2//2:,1* 02'(/6
Figure 4.26
Speed Versus Vehicle Concentration (Rothery 1968).
Table 4.6 flow variables without resorting to using two different

Parameter Comparison expressions or two different sets of parameters to one expression
(Holland Tunnel Data) has been made by Hall (1987). More recently, Acha-Daza and
Hall (1994) have reported an analysis of freeway data using
catastrophe theory which indicates that such an approach can
Microscopic Macroscopic effectively be applied to traffic flow. Macroscopic data has also
Parameters Estimates Estimates
been reported on single lane bus flow. Here platoons of ten
a1,0 26.8 27.8 buses were studied (Rothery et al. 1964).
a2, 0 0.57 0.12 Platoons of buses were used to quantify the steady-state stream
a2,1 (123)-1
(54) -1 properties and stability characteristics of single lane bus flow.
Ideally, long chains of buses should be used in order to obtain
the bulk properties of the traffic stream and minimize the end
effects or eliminate this boundary effect by having the lead
vehicle follow the last positioned vehicle in the platoon using a
A totally different approach to modeling traffic flow variables circular roadway. These latter type of experiments have been
which incorporates such discontinuities can be found in the carried out at the Road Research Laboratory in England
literature. Navin (1986) and Hall (1987) have suggested that (Wardrop 1965; Franklin 1967).
catastrophe theory (Thom 1975; Zeeman 1977) can be used as
a vehicle for representing traffic relationships. Specifically, In the platoon experiments, flow rates, vehicle concentration,
Navin followed the two regime approach proposed by Edie and and speed data were obtained. The average values for the speed
cited above and first suggested that traffic relations can be and concentration data for the ten bus platoon are shown in
represented using the cusp catastrophe. A serious attempt to Figure 4.28 together with the numerical value for the parameter
apply such an approach to actual traffic data in order to represent

&$5 )2//2:,1* 02'(/6
Figure 4.27
Flow Versus Concentration for the Lincoln and Holland Tunnels.
Figure 4.28
Average Speed Versus Concentration
for the Ten-Bus Platoon Steady-State Test Runs (Rothery 1968).

&$5 )2//2:,1* 02'(/6
a1,0 = 53 km/h which is to be compared to that obtained from the speed of clusters of three or more buses on seven different
two bus following experiments discussed earlier namely, 58 highways distributed across the United States were measured
km/h. Given these results, it is estimated that a single lane of and concluded that a maximum flow for buses would be
standard size city buses is stable and has the capacity of over approximately 1300 buses/hour and that this would occur at
65,000 seated passengers/hour. An independent check of this about 56 km/h.
result has been reported (Hodgkins 1963). Headway times and
4.5 Automated Car Following

All of the discussion in this chapter has been focused on manual Mechanical Laboratory (Oshima et al. 1965), the Transportation
car following, on what drivers do in following one or more other Road Research Laboratory (Giles and Martin 1961; Cardew
vehicles on a single lane of roadway. Paralleling these studies, 1970), Ford Motor Corporation (Cro and Parker 1970) and the
research has also focused on developing controllers that would Japanese Automobile Research Institute (Ito 1973). During the
automatically mimic this task with specific target objectives in past several decades three principal research studies in this arena
mind. stand out: a systems study of automated highway systems
conducted at General Motors from 1971-1981, a long-range
At the 1939 World's Fair, General Motors presented program on numerous aspects of automated highways conducted
conceptually such a vision of automated highways where at The Ohio State University from 1964-1980, and the Program
vehicles were controlled both longitudinally (car following) and on Advanced Technology for the Highway (PATH) at the
laterally thereby freeing drivers to take on more leisurely University of California, Berkeley from about 1976 to the
activities as they moved at freeway speeds to their destinations. present. Three overviews and detailed references to milestones
In the intervening years considerable effort has been extended of these programs can be found in the literature: Bender (1990),
towards the realization of this transportation concept. One prime Fenton and Mayhan (1990), and Shladover et al. (1990),
motivation for such systems is that they are envisioned to provide respectively.
more efficient utilization of facilities by increasing roadway
capacity particularly in areas where constructing additional The car following elements in these studies are focused on
roadway lanes is undesirable and or impractical, and in addition, developing controllers that would replace driver behavior, carry
might improve safety levels. The concept of automated out the car following task and would satisfy one or more
highways is one where vehicles would operate on both performance and/or safety criteria. Since these studies have
conventional roads under manual control and on specially essentially been theoretical, they have by necessity required the
instrumented guideways under automatic control. Here we are difficult task of modeling vehicle dynamics. Given a controller
interested in automatic control of the car following task. Early for the driver element and a realistic model representation of
research in this arena was conducted on both a theoretical and vehicle dynamics a host of design strategies and issues have been
experimental basis and evaluated at General Motors Corporation addressed regarding inter-vehicular spacing control, platoon
(Gardels 1960; Morrison 1961; Flory et al. 1962), Ohio State configurations, communication schemes, measurement and
University (Fenton 1968; Bender and Fenton 1969; Benton timing requirements, protocols, etc. Experimental verifications
et al. 1971; Bender and Fenton 1970), Japan Governmental of these elements are underway at the present time by PATH.

$5 2//2:,1* 2'(/6
4.6 Summary and Conclusions

Historically, the subject of car following has evolved over the this chapter has addressed in order. First, it provides a
past forty years from conceptual ideas to mathematical model mathematical model of a relative common driving task and
descriptions, analysis, model refinements resulting from provides a scientific foundation for understanding this aspect of
empirical testing and evaluation and finally extensions into the driving task; it provides a means for analysis of local and
advanced automatic vehicular control systems. These asymptotic stability in a line of vehicles which carry implications
developments have been overlapping and interactive. There with regard to safety and traffic disruptions and other dynamical
have been ebbs and flows in both the degree of activity and characteristics of vehicular traffic flow; it provides a steady state
progress made by numerous researchers that have been involved description of single lane traffic flow with associated road
in the contributions made to date. capacity estimates; it provides a stepping stone for extension into
advance automatic vehicle control systems; and finally, it has and
The overall importance of the development of the subject of car will undoubtedly continue to provide stimulus and
following can be viewed from five vantage points, four of which encouragement to scientists working in related areas of traffic
theory.
References
Acha-Daza, J. A. and F. L. Hall (1994). Application of Chow, T. S. (1958). Operational Analysis of a Traffic
Catastrophe Theory to Traffic Flow Variables. Dynamics Problem. Operations Research, 6(6),
Transportation Research - B, 28B(3). Elsevier Science pp. 165-184.
Ltd., pp. 235-250. Constantine, T. and A. P. Young (1967). Traffic Dynamics:
Babarik, P. (1968). Automobile Accidents and Driver Car Following Studies. Traffic Engineering and Control 8,
Reaction Pattern. Journal of Applied Psychology, 52(1), pp. 551.
pp. 49-54. Cumming, R. W. (1963). The Analysis of Skills in Driving.
Barbosa, L. (1961). Studies on Traffic Flow Models. Reports Journal of the Australian Road Research Board 1, pp. 4.
No. 202A-1. The Ohio State University Antenna Darroch, J. N. and R. W. Rothery (1973). Car Following and
Laboratory. Spectral Analysis. Proceedings of the 5th International
Bender, J. G. (1971). An Experimental Study of Vehicle Symposium on the Theory of Traffic Flow and
Automatic Longitudinal Control. IEEE Transactions on Transportation. Ed. Newell, G. F., American Elsevier
Vehicular Technology, VT-20, pp. 114-123. Publishing Co., New York.
Bender, J. G. (1991). An Overview of Systems Studies of Drake, J. S., J. L. Schofer, and A. D. May, Jr. (1967). A
Automated Highway Systems. IEEE Transactions on Statistical Analysis of Speed Density Hypotheses. Highway
Vehicular Technology 40(1). IEEE Vehicular Technology Research Record 154, pp. 53-87.
Society, pp. 82-99. Drew, D. R. (1965). Deterministic Aspects of Freeway
Bender, J. G. and R. E. Fenton (1969). A Study of Automatic Operations and Control. Highway Research Record, 99,
Car Following. IEEE Transactions on Vehicular pp. 48-58.
Technology, VT-18, pp. 134-140. Edie, L. C. (1961). Car-Following and Steady State Theory
Cardew, K. H. F. (1970). The Automatic Steering of Vehicles for Non-Congested Traffic. Operations Research 9(1),
-An Experimental System Fitted to a Citroen Car. pp. 66-76.
RL 340, Road Research Laboratories. Edie, L. C. and E. Baverez (1967). Generation and
Chandler, F. E., R. Herman, and E. W. Montroll, (1958). Propagation of Stop-Start Waves. Vehicular Traffic
Traffic Dynamics: Studies in Car Following, Operations Science Proceedings of the 3rd International Symposium on
Research, 6, pp. 165-184. the Theory of Traffic Flow. L.C. Edie, R. Herman and R.W.
Rothery (Eds.). American Elsevier, New York.

$5 2//2:,1* 2'(/6
Edie, L. C. and R. S. Foote, (1958). Traffic Flow in Tunnels. Greenburg, H. (1959). An Analysis of Traffic Flow.
Proceedings of the Highway Research Board, 37, Operations Research 7(I), pp. 255-275.
pp. 334-344. Greenshields, B. D. (1935). A Study of Traffic Capacity.
Edie, L. C. and R. S. Foote (1960). Effect of Shock Waves on Proceedings of the Highway Research Board, 14, 468.
Tunnel Traffic Flow. Proceedings of the Highway Research Hall, F. L. (1987). An Interpretation of Speed-Flow
Board, 39, pp. 492-505. Concentration Relationships Using Catastrophe Theory.
Edie, L. C., and R. S. Foote (1961). Experiments on Single Transportation Research, 21A, pp. 335-344.
Lane Flow in Tunnels. Theory of Traffic Flow Proceedings Hankin, A. and T. H. Rockwell (1967). A Model of Car
of the Theory of Traffic Flow. Ed. R. Herman. Elsevier Following Derived Empirically by Piece-Wise Regression
Pub. Co., Amsterdam, pp. 175-192. Analysis. Vehicular Traffic Science Proceedings of the 3rd
Edie, L. C., R. S. Foote, R. Herman, and R. W. Rothery International Symposium on the Theory of Traffic Flow.
(1963). Analysis of Single Lane Traffic Flow. Traffic L.C. Edie, R. Herman and R.W. Rothery (Eds.). American
Engineering, 21, 27. Elsevier, New York.
Ellson, D. G. (1949). The Application of Operational Analysis Harris, A. J. (1964). Following Distances, Braking Capacity
to Human Motor Behavior. Psychology Review 9, and the Probability of Danger of Collision Between
pp. 56. Vehicles. Australian Road Research Board, Proceedings 2,
Fenton, R. E. (1968). One Approach to Highway Automation. Part 1, pp. 496-412.
Proceedings IEEE 56; pp. 556-566. Helly, W. (1959). Dynamics of Single-Lane Vehicular Traffic
Fenton, R. E. and R. J. Mayhan (1991). Automated Highway Flow. Research Report No. 2, Center for Operations
Studies at The Ohio State University - An Overview. IEEE Research, MIT. Cambridge, Mass.
Transaction on Vehicular Technology 40(1); IEEE Herman, R., E. W. Montroll, R. B. Potts and R. W. Rothery
Vehicular Technology Society, pp. 100-113. (1958). Traffic Dynamics: Analysis of Stability in Car
Flory, L. E. (1962). Electronic Techniques in a System of Following. Operations Research, E. 17, pp. 86-106.
Highway Vehicle Control. RCA Review 23, pp. 293-310. Herman, R. and R. B. Potts (1959). Single Lane Traffic
Forbes, T. W., M. J. Zagorski, E. L. Holshouser, and W. A. Theory and Experiment. Proceedings Symposium on
Deterline (1959). Measurement of Driver Reaction to Theory of Traffic Flow. Ed. R. Herman, Elsevier
Tunnel Conditions. Proceedings of the Highway Research Publications Co., pp. 120-146.
Board 37, pp. 345-357. Herman, R. and R. W. Rothery (1962). Microscopic and
Gardels, K. (1960). Automatic Car Controls for Electronic Macroscopic Aspects of Single Lane Traffic Flow.
Highways. GM Research Laboratories Report GMR-276, Operations Research, Japan, pp. 74.
Warren, MI. Herman, R. and R. W. Rothery (1965). Car Following and
Gazis, D. C., R. Herman, and R. B. Potts (1959). Car Steady-State Flow. Proceedings of the 2nd International
Following Theory of Steady State Traffic Flow. Operations Symposium on the Theory of Traffic Flow. Ed J. Almond,
Research 7(4), pp. 499-505. O.E.C.D., Paris.
Gazis, D. C., R. Herman, and R. W. Rothery (1961). Non- Herman, R. and R. W. Rothery (1969). Frequency and
Linear Follow the Leader Models of Traffic Flow. Amplitude Dependence of Disturbances in a Traffic
Operations Research, 9, pp. 545-567. Stream. Proceedings of 4th International Symposium on the
Gazis, D. C., R. Herman, and R. W. Rothery (1963). Analytical Theory of Traffic Flow, Ed. W. Leutzbach and P. Baron.
Methods in Transportation: Mathematical Car-Following Bonn, Germany.
Theory of Traffic Flow. Journal of the Engineering Highway Capacity Manual (1950). U.S. Government Printing
Mechanics Division, ASCE Proc. Paper 3724 89(Paper Office, Washington, DC.
372), pp. 29-46. Hodgkins, E. A. (1963). Effect of Buses on Freeway Capacity.
Giles, C. G. and J. A. Martin (1961). Cable Installation for Presentation at the 43rd Annual Meeting of the Highway
Vehicle Guidance Investigation in the New Research Track Research Board 33. Highway Research Abstracts, pp. 69.
at Crowthorne. Rep RN/4057/CGG, Road Research Ito, T. (1973). An Automatic Drivers System of Automobiles
Laboratory, Crowthorne, England. by Guidance Cables. Society of Automotive Engineering
(730127).

$5 2//2:,1* 2'(/6
Kometani, E. and T. Suzaki (1958). On the Stability of Traffic Shladover, S. E., C. A. Desoer, J. D. Hedrick, M. Tomizuka,
Flow. J. Operations Research, Japan 2, pp. 11-26. J. Walrand, W. B. Zhang, D. H. McMahon, H. Peng,
Lam, T. and R. W. Rothery (1970). Spectral Analysis of S. Shiekholeslam, and N. McKeown (1991). Automated
Speed fluctuations on a Freeway. Transportation Science Vehicle Control Developments in the PATH Program.
4(3). IEEE Transactions on Vehicular Technology 40(1),
Lee, G. (1966). A Generalization of Linear Car-Following pp. 114-130.
Theory. Operations Research 14(4), pp. 595-606. Taylor, F. V. (1949). Certain Characteristics of the Human
May, A. D. and H.E.M. Keller (1967). Non-Integer Car Serve. Electrical Engineering 68, pp. 235.
Following Models. Highway Research Record 199, Telfor, G. W. (1931). The Refractory Phase of Voluntary and
pp. 19-32. Associative Responses. Journal of Experimental
Michaels, R. M. (1963). Perceptual Factors in Car Psychology, 14, pp. 1.
Following. Proceedings of the 2nd International Thorn, R. (1975). Structural Stability and Morphogenesis.
Symposium on the Theory of Road Traffic Flow (London, An English Translation by D.H. Fowler. John Wiley &
England), OECD. Sons, New York.
Navin, F. P. D. (1986). Traffic Congestion Catastrophes. Todosiev, E. P. (1963). The Action Point Model of the Driver
Transportation Planning and Technology 11, pp. 19-25. Vehicle System. Report No. 202A-3. Ohio State
Newell, G. F. (1961). Nonlinear Effects in the Dynamics of University, Engineering Experiment Station, Columbus,
Car Following. Operations Research 9(2), pp. 209-229. Ohio.
Newell, G. F. (1962). Theories of Instability in Dense Tuck, E. (1961). Stability of Following in Two Dimensions.
Highway Traffic. J. Operations Research Society of Japan Operations Research, 9(4), pp. 479-495.
5(1), pp. 9-54. Tustin, A. (1947). The Nature of the Operator Response in
Oshima, R. (1965). Control System for Automobile Driving. Manual Control and its Implication for Controller Design.
Proceedings of the Tokyo 1FAC Symposium, pp. 347-357. J.I.E.E. 92, pp. 56.
Pipes, L. A. (1951). A Proposed Dynamic Analogy of Traffic. Unwin, E. A. and L. Duckstein (1967). Stability of
ITTE Report, Institute of Transportation and Traffic Reciprocal-Spacing Type Car Following Models.
Engineering, University of California, Berkeley. Transportation Science, 1, pp. 95-108.
Pipes, L. A. (1953). An Operational Analysis of Traffic Uttley, A. (1941). The Human Operator as an Intermittent
Dynamics. Journal of Applied Physics 24, pp. 271-281. Servo. Report of the 5th Meeting of Manual Tracking
Reuschel, A. (1950). Fahrzeugbewegungen in der Kolonne Panel, GF/171.SR1A.
Beigleichformig beschleunigtem oder vertzogerten Wardrop, J. G. (1965). Experimental Speed/Flow Relations in
Leitfahrzeub, Zeit. D. Oster. Ing. U. Architekt Vereines Ed. a Single Lane. Proceedings of the 2nd International
(Vehicle Movements in a Platoon with Uniform Symposium on the Theory of Road Traffic Flow. Ed. J.
Acceleration or Deceleration of the Lead Vehicle), pp. 50- Almond O.E.C.D.
62 and 73-77. Zeeman, E. C. (1977). Catastrophe Theory. Selected Papers
Rothery, R. W., R. Silver, R. Herman and C. Torner (1964). 1972-1977. Addison-Wesley, London.
Analysis of Experiments on Single Lane Bus Flow.
Operations Research 12, pp. 913.

CONTINUUM FLOW MODELS
REVISED BY H. MICHAEL ZHANG*

ORIGINAL TEXT BY REINHART KUHNE7
PANOS MICHALOPOULOS8
*
Professor, Department of Civil and Environmental Engineering, University of California Davis, One
Shields Avenue, Davis, CA 95616
7
Managing Director, Steierwald Schonharting und Partner, Hessbrühlstr. 21c 7 70565 Stuttgart, Germany
8
Professor, Department of Civil Engineering, University of Minnesota Institute of Technology, 122 Civil
Engineering Building, 500 Pillsbury Drive S.E., Minneapolis, MN 55455-0220
a = dimensionless traffic parameter L = distance

A = stop-start wave amplitude L = length of periodic interval
= sensitivity coefficient ld = logarithmus dualis
b = net queue length at traffic signal lo = characteristic length
c = g + r = cycle length = wave length of stop-start waves
c0 = coefficient μ0 = dynamic viscosity
2
c0 = constant, independent of density k μ = viscosity term
ds = infinitesimal time n = current time step
t, x = the time and space increments respectively N = normalization constant
such that x/t > free flow speed n i, n 2 = exponents
i, i+1 = deviations Ni = number of cars (volume)
= state vector 7 = eigenvalue
i, i+1 = state vector at position i, i+1 p = probability
f(x, v, t ) = vehicular speed distribution function q = actual traffic volume, flow
f = relative truck portion, kpass = k Q0 = net flow rate
f0 = equilibrium speed distribution qa = average flow rate
= fluctuating force as a stochastic quantity qaka = arrival flow and density conditions
g = effective green interval qni = capacity flow
n
gj = is the generation (dissipation) rate at node j at r = effective red interval
t = t0 + nt; if no sinks or sources exist gj =
n
)0 = quantity
0 and the last term of Equation 5.28 vanishes T = oscillation time
gmin = minimum green time required for t = time
undersaturation t0 = the initial time
h = average space headway - = relaxation time as interaction time lag
i = station u = speed
j = node Ue(k) = equilibrium speed-density relation
k = density ue(k n ij) = equilibrium speed
k-, k+ = density downstream, upstream shock uf = free-flow speed of the approach under
k0 = operating point consideration
k10 = equilibrium density ug = group velocity
KA = constant value umax - umin = speed range
ka = density within L2 uw = shock wave speed
kbumper = density "bumper to bumper" uz = spatial derivative of profile speed
kd, qd = density, flow downstream v(k) = viscosity
ku, qu = density, flow upstream vg = values of the group velocity
khom = vehicle density in homogeneous flow W(q) = distribution of the actual traffic volume
kj = jam density of the approach under values q
consideration x = space
n n
kj ,qj = density and flow rate on node j at xh = estimated queue length
t = t0 + nt xi, ti, yi = coordinates at point i
km = density conditions Xij = length of any line ij
kpass = density "bumper to bumper" for 100% y = street width
passenger cars y(t) = queue length at any time point t
kref = reference state yij = queue length from i to j assuming a positive
ktruck = density "bumper to bumper" for 100% trucks direction opposite to x, i.e. from B to A
z = x - U, t, collective coordinate
x = shockspeed
Revised Chapter 5 of Traffic Flow Theory Monograph
Original text by Reinhart Kühne and Panos Michalopoulos

Revised by H. M. Zhang
1 Conservation and traffic waves

There have been two kinds of traffic flow theories presented thus far in this monograph. Chapter
2 discussed the relations among aggregated quantities of flow rate, concentration and mean speeds
for stationary vehicular traffic (these relations are also referred to as the equations of state in
traffic flow), and Chapter 3 described the dynamic evolution of microscopic quantities of vehicle
spacing, headway and speeds in single-lane traffic through the follow-the-leader type of models.
In this Chapter we study another kind of traffic flow theory—continuum traffic flow theory—
that describes the temporal-spatial evolution of macroscopic flow quantities as mentioned earlier.
Continuum theories are natural extensions to the first two kinds of theories because, one one
hand, they are closely related to the equations of state and car-following theories, and on the
other hand, they overcome some of the drawbacks of the first (stationarity) and second (too much
detail and limited observability) kinds of theories. The quantities and features that continuum
theories describe, such as traffic concentration and shock waves, are mostly observable with current
surveillance technology, which makes it easier to validate and calibrate these models.
Many theories exist in the continuum description of traffic flow. All of them share two funda-
mental relations: one is the conservation of vehicles and the other is the flow-concentration-speed
relation
q = ku. (1)
The q − k − u relation is true by choice, i.e., one defines flow rate (q), concentration (k) and mean
speed (u) in such a way that (1) always holds (see Chapter 2 for definitions). Conservation of
vehicles, on the other hand, is true regardless of how q, k, u are defined, and can be expressed in
different forms. Here we present three forms of the conservation law using variables in (1).
The first integral form of the conservation law. Consider a stretch of highway between x1 and
x2 (x1 < x2 ). At time t the traffic concentration on this section is k(x, t), and traffic flows into
the section at a rate of q(x1 , t), and out of the section at a rate of q(x2 , t). The total number of
5-1
vehicles in this section at time t is then

x2
k(x, t)dx.
x1
Suppose no entries and exits exist between x1 and x2 , then by conservation the rate of change in
the number of vehicles
∂ x2
k(x, t)dx.
∂t x1
should equal to the net flow into this section
q(x1 , t) − q(x2, t),
that is, x2
∂
k(x, t)dx = q(x1 , t) − q(x2 , t), (2)
∂t x1
This is the first integral form of the conservation law.
The second integral form of the conservation law. The conservation of vehicles also means that,
in the x −t plane, the net number of vehicles passing through any closed curve is zero provided that
no sources and sinks are there in the enclosed region (Fig. 2). Note that the number of vehicles
across any segment dl of curve C is −kdx + qdt, the total number of vehicles across the enclosed
region is therefore the line integral of the vector field (−k, q), and we have the second integral form
of the conservation law:
−kdx + qdt = 0. (3)
C
(3) implies that we can construct a scalar function N (x, t) with the properties
∂N ∂N
q= , k=− .
∂t ∂x
This function is the cumulative count of vehicles passing location x at time t, provided that N (x, t =
0) = 0. The fact that such a function exists is another consequence of the conservation law.
The differential form of the conservation law. By applying the divergence theorem to the second
integral form we obtain

∂k ∂q
−kdx + qdt = + dxdt = 0,
C D ∂t ∂x
which leads to the differential form of the conservation law:
∂k ∂q
+ = 0. (4)
∂t ∂x
The three different forms of the conservation law are not completely equivalent. Note that in
the differential form, both k and q are required to be differentiable with respect to time t and space
x. In contrast, it is perfectly fine in the integral forms that these variables are discontinuous in
space and/or time. This is an important distinction because a special kind of traffic waves, called
shock waves, are prevalent in vehicular traffic flow and play an essential role in the continuum
5-2
description of traffic flow. A shock wave is a drastic change in traffic concentration (and/or speed,
flow rate) that propagates through a traffic stream. Examples of shock waves include the stoppage
of traffic in front of a red light, and traffic slow-downs caused by an accident. In reality a shock
always has a profile (that is, a transition region of non-zero width) that usually spans a few vehicle
lengths, but this can be practically treated as a discontinuity when compared with the length of
roads in consideration.
Because of the existence of shocks, it is necessary to expand the solution space of (4) to include
the so-called weak solutions. In mathematical sense a weak solution is a function (k, q)(x, t) that
satisfies (4) everywhere except along a certain path x(t). On x(t) (k, q)(x, t) are discontinuous but
obey the integral forms of the conservation law. A shock wave solution to (4) is a weak solution.
Moreover, its path x(t) is governed by the following equation:
ẋ(t) = [q]/[k]. (5)
where [k] = kr − kl , [q] = qr − ql , ẋ(t) is the speed of the shock, also denoted as s, and (kl,r , ql,r ) are
the traffic states immediately to the right and left of the shock path x(t), respectively. Expression
(5) is also known as the Rankine-Hugoniot (R-H) condition (LeVeque 1992) , which is a consequence
of the conservation law.
Expression (5) has a simple geometric interpretation in the (k, q) plane, also known as the k − q
phase plane. It is simply the slope of the segment that connects the two phase points (kl , ql ) and
(kr , qr ) (Fig. 1(a)). Suppose that a unique curve connects (kl,r , ql,r ), and kr → kl (i.e., the shock
is weak), then s approaches the slope of the tangent of the connecting curve at (kl , ql ) (Fig. 1(b)).
We call this speed the speed of traffic sound waves, and give it a special notation c(k). This is the
speed that information is propagated in homogeneous traffic. It plays a central role in the basic
kinematic wave model of traffic flow. It must be pointed out that a unique q − k curve is not
necessary to compute the shock speed. This can be done as long as both the densities and flow
rates on both sides of the shock are known.
Next we provide two derivations of (5) from the conservation law. The first derivation is based
on the first integral form of the conservation law. Suppose that traffic states on both sides of the
shock are constant states (k1,2 , q1,2 ), and suppose that two observers located at x1 (t) and x2 (t) on
the two sides of the shock (i.e., x1 < x(t) < x2 ) travel at precisely the speed of the shock, s. We
apply the conservation law (2) to the region between the two observers after doing a coordinate

transformation x = x − st to obtain:
x
∂ 2

k(x, t)dx = [q1 − k1 s] − [q2 − k2 s].
∂t x1

The integral, however, can be evaluated independently in [x1 , xl ) and (xr , x2 ], where xl,r are points

immediate to the right/left of shock path x (t). This integral turns out to be

k1 (xl − x1 ) + k2 (x2 − xr ),
5-3
Figure 1: Geometric representation of shocks, sound waves, and traffic speeds in the k − q phase
plane
a constant, whose derivative is zero. Therefore we have
[q1 − k1 s] − [q2 − k2 s] = 0,
which leads to (5) after rearranging terms.

The second derivation of the Rankine-Hugoniot condition comes from the second integral form.
As shown in Fig.2, a shock path x(t) breaks the closed curve C into two parts Cl and Cr . Let Γl,r
be the right/left edges of x(t), respectively, then the conservation law applies to each of the three
enclosures C, Cl ∪ Γl , and Cr ∪ Γr :
−kdx + qdt = 0. (6)
C

−kdx + qdt = 0. (7)
Cl ∪Γl

−kdx + qdt = 0. (8)
Cr ∪Γr
Adding (7) and (8) together and rearranging terms we obtain

−kdx + qdt + −kdx + qdt + −kdx + qdt = 0.
C Γl Γr
The first integral is zero, and the second and third integrand become
(−kl,r s + ql,r )dt
on Γl,r . Taking account the direction of integration we obtain the following integral equation:
t2
{(−kl s + ql ) − (−kr s + qr )}dt = 0.
t1
5-4
This implies that

(−kl s + ql ) − (−kr s + qr ) = 0,
which also leads to the Rankine-Hugoniot shock condition.
Figure 2: Field representation of shocks and conservation of flow
The differential conservation law (4), supplemented by the Rankine-Hugoniot condition (5), is
the corner stone of all continuum vehicular traffic theories. Yet it is not a complete theory, nor is it
unique to traffic flow. In fact, the differential conservation law is obeyed by many kinds of material
fluids, such as gas flow in a pipe or water flow in a channel. As such, it does not capture the unique
characters of vehicular traffic flow, a special kind of “fluid”. It must be supplemented by additional
relations to form a complete and useful traffic flow theory. Some of the relations between macro-
scopic traffic variables are already available and have been discussed in Chapter 2. These are the
binary relations between flow rate, traffic concentration and mean travel speed. Although these re-
lations are found among stationary traffic, they can also be used as a first approximation of dynamic
flow-concentration-speed relations. This leads to the first, and the simplest continuum theory of
traffic flow—the kinematic wave theory developed independently by Lighthill and Whitham (1955),
and Richards (1956) (The LWR model). A more accurate relation between flow-concentration or
speed-concentration is one that accounts for the dynamic behavior of drivers—anticipation and
inertia. The usage of such a relation leads to the so-called higher-order models. We will study in
this chapter these two kinds of continuum theories in great deal. In the sections to follow, we’ll
first study the properties of the LWR model and its Riemann problem, then study the properties
of a special class of higher-order models and their respective Riemann problems. Next we develop
numerical approximations of both types of models, and finally we provide some examples of the
use of the numerical approximations.
5-5
2 The kinematic wave model of LWR

2.1 The LWR model and characteristics
The LWR model assumes that the relation observed in stationary flow, q = f∗ (k), also applies to
dynamic traffic. With such a relation, the conservation law becomes:
kt + f∗ (k)x = 0. (9)
and the Rankine-Hugoniot condition becomes
s = [f∗ ]/[k].
This is the celebrated kinematic wave model of LWR. It is one of the simplest nonlinear scalar
conservation laws in physics and engineering.
Experimental evidence indicates that f∗ (k) is usually smooth and concave (refer to Chapter 2),
and satisfies the following boundary conditions:
f∗ (0) = f∗ (kj ) = 0,

f∗ (0) = vf > 0, f∗ (kj ) = cj < 0.
where kj is the jam concentration at which vehicles grind to a halt, usually ranging from 260 vpm
to 330 vpm (in passenger car units), cj is the speed of traffic sound waves at jam condition, taking
values in the range of [-10 mph, -20 mph], and vf is free-flow travel speed, ranging from 55mph
to 75mph on freeways. All these parameter values are readily computable from conventional field
measurements.
With the introduction of traffic wave speed ( when no confusion arises we use traffic wave and
traffic sound waves interchangably) c(k) = f∗ (k), the LWR model also reads
kt + c(k)kx = 0. (10)
Now let us look at what changes in concentration an observer sees when he travels at the speed of
c(k), that of traffic waves.
dk dx
= kt + kx = kt + c(k)kx = 0.
dt dt
That is to say, he sees no changes in density at all if he travels at the wave speed. As a result, if he
knows the initial concentration at a point ξ, he’ll know the concentration at any point on the path
ẋ(t) = c(k), x(0) = ξ. (11)
This path is called the characteristic curve of the LWR model, and the wave speed c(k) is also
known as the characteristic speed of the LWR model. Because k is constant along the characteristic
curve, the characteristic of (9) is therefore a straight line.
One must not confuse over traffic sound wave speeds, shock wave speeds, characteristic speeds,
and vehicle speeds. Traffic sound wave speeds and characteristic speeds in the LWR model are
5-6
identical, which are the slopes of the tangential lines on the flow-concentration curve (also known
as the fundamental diagram of traffic flow). Shock wave speeds are the slopes of the secant that
connecting any two traffic states on f∗ (k) , and vehicular speeds are the slopes of the rays from the
origin to points on f∗ (k) (Fig. 1). Because of the concave shape of f∗ (k), we have the following
relations among the various speeds:
v = f∗ /k ≡ v∗ (k) ≥ c(k), s < v∗ (kl,r ),
that is to say, all waves, including sound and shock waves, travel no faster than traffic. In other
words, information in LWR traffic is propagated against the traffic stream. This property of
information propagation in the LWR model is known as the anisotropic property of traffic flow.
The anisotropic property may not hold if the fundamental diagram is not concave (Zhang 2000c).
2.2 The Riemann problem and entropy solutions

The LWR model is a well-posed hyperbolic partial differential equation (pde) 1 and can be solved
with proper initial/boundary data. In fact, analytical solutions can be obtained for a special kind
of problem called Riemann problem, whose initial data—the so-called Reimann data—are two
constant states kl,r ≥ 0 separated by a single jump:

kl , x < 0,
k(x, t = 0) =
kr , x > 0
The analytical solution to the Riemann problem can either be a shock:

kl , x < st,
k(x, t) = (12)
kr , x > st.
or a smooth expansion wave (also known as a rarefaction wave):

⎧
⎪
⎪ kl , x < cl t,
⎨
x

k(x, t) = (f∗ )−1 t , cl t ≤ x ≤ cr t,
(13)
⎪
⎪
⎩ kr , x > cr t
The condition for it to be a shock is governed by the so-called entropy condition, which states that
cl > cr . (14)

Otherwise the solution will be an expansion wave. Because of the concavity condition f∗ (k) < 0,
(14) implies that kl < kr , that is, shocks arising from the LWR theory are compressive. Moreover,
they reach vehicles from upstream because s < vl,r . Examples of the two kinds of solutions to a
Riemann problem are shown in Figs. 3& 4.
For general initial conditions, it is usually tedious, if not difficult, to obtain analytical solutions
of the LWR equation, and numerical approximations are often sought, where the Riemann problem
1
A pde is hyperbolic if its characteristic speeds are real (as compared to imaginary).
5-7
Figure 3: A shock solution
Figure 4: A rarefaction solution
plays a key role in developing some of the most efficient and accurate numerical schemes. This will
be discussed in Section 4 together with the treatment of boundary conditions. There are special
cases where an analytical solution is still straightforward to obtain. One case involves a special kind
of initial condition—k(x, 0) is piece-wise constant and and increasing, and another case involves a
special kind of fundamental diagram—the triangular shaped fundamental diagram that has only
two wave speeds (Koshi et al 1983, Newell 1993). In the former case, a series of Riemann problems
can be solved with consideration of wave interactions, and in the latter case, expansion waves are
replaced by acceleration shocks, and it is much simpler to construct the solutions with shocks than
expansion waves (see Newell 1993).
2.3 Applications
Same as Sections 5.1.3 & 5.1.4 in the original text.
5-8
2.4 Extensions to the LWR model

The LWR model we presented applies to traffic on a homogeneous highway with no entering and
exiting traffic. In reality all highways have entries and exits and variable geometric features.
Fortunately the LWR equation can be easily extended to model such inhomogeneities.
Suppose the net inflow to a road section is s(x, t)dxdt from sources such as freeway ramps, then
from conservation we have
kt + f∗ (k)x = r(x, t), (15)
This is the LWR model with sources. These source flows can be entering or exiting flows from
ramps, or both.
When road geometries change, such as the drop of a lane at some locations, they often generate
disturbances that cause traffic breakdowns. The effect of such inhomogeneities on traffic flow can be
captured to some degree with a location-dependent fundamental diagram, f∗ (k, x). Because vehicle
conservation still holds, applying the conservation principle to an inhomogeneous road leads to the
following extended LWR model:
kt + f∗ (k, x)x = 0. (16)
If a road segment has both sources and geometric inhomogeneities, the following model applies:
kt + f∗ (k, x)x = r(x, t). (17)
Like the homogeneous LWR model, these extensions are also non-linear hyperbolic pdes, whose
characteristics are
ẋ = f∗k (k, x). (18)
These characteristics, however, are no longer straight lines as in the homogeneous model. This is
because along the characteristics, we have
⎧
⎪
⎪ r, for (15)
⎨
k̇ = kt + kx ẋ = kt + f∗k kx = −f∗x , for (16) (19)
⎪
⎪
⎩ r − f , for (17)
∗x
(18) and (19) form a system of ordinary differential equations (ode), and can be solved iteratively
using any ode solver such as the Runge-Kutta method. This way of solving the inhomogeneous
LWR models is known as the method of characteristics (Courant and Hilbert 1962).
The basic LWR model and its various extensions we have covered treat traffic across lanes as
homogeneous, which is somewhat restrictive because alternate motion of traffic in different lanes is
often observed in heavily congested traffic. This limitation is easily overcome by modeling traffic
evolution in each lane. Here the conservation principle still applies, but a continuum of sources
exist for each lane. These sources are the exchange of flow between adjacent lanes. Without loss
of generality, assume a two-lane highway with traffic (k1 (x, t), q1 (x, t)) and (k2 (x, t), q2 (x, t)). Let
Q1 be the net amount of traffic entering lane 1 from lane 2 in unit time and distance, and Q1 be
5-9
the net amount of traffic entering lane 2 from lane 1 in unit time and distance, then the extended
LWR model for a two-lane highway is (Dressler 1949)
k1t + q1x = Q1 , (20)
k2t + q2x = Q2 . (21)
where q1 = f1∗ (k1 ) and q2 = f2∗ (k2 ).
From conservation we know the source flows Q1 , Q2 observe
Q1 + Q2 = 0
and experiences tell us that the change of flow between lanes is related to the differential of lane
speeds/densities, for drivers always like to move into the faster lane in congested traffic. An example
of Q1 , Q2 was proposed by Gazis et al. (1962):
0 0
Q1,2 = α[(k2,1 − k1,2 ) − (k2,1 − k1,2 )] (22)
0 are equilibrium densities of lanes 1 & 2, respectively (Gazis et al. 1962). These can be
where k1,2
observed experimentally.
The above lane-specific LWR model is a system of quasi-linear PDEs and can be expressed in
vector forms
k1 f1∗(k1 ) Q1 (k1 , k2 )
+ = , (23)
k2 f2∗(k2 ) Q2 (k1 , k2 )
t x
or
k1 f1∗,k1 0 k1 Q1 (k1 , k2)
+ = . (24)
k2 t 0 f2∗,k2 k2 x Q2 (k1 , k2)
For a system
of quasi-linear PDEs, the characteristics are the eigenvalues of the Jacobian ma-
f1∗,k1 0
trix DF ≡ , which contains the partial derivatives of the flow vector F (U ) ≡
0 f2∗,k2

f1∗ (k1 ) k1
with respect to the state vector U ≡ .
f2∗ (k2 ) k2
This system is said to be strictly hyperbolic if all its characteristics are real and distinct. It

is easy to see that the characteristics of (23) are f1∗ (k1 ) and f2∗(k2 ) because the dynamics of the
two lanes are nearly decoupled (the only coupling comes from the source term). The system is

strictly hyperbolic under the condition that f1∗ (k1 ) = f2∗ (k2 ) and f1,2∗ < 0. Thus this system can
be solved in the same way as other inhomogeneous models through the method of characteristics.
Only this time one solves a system of four odes instead of two:

ẋ1 = f1∗ (k1 ),

ẋ2 = f2∗ (k2 ),
k̇1 = Q1 (k1 , k2 ),
k̇2 = Q2 (k1 , k2 ).
It should be noted that both the inhomogeneous models and the multiple-lane models can be
solved numerically using the finite difference procedures developed in Section 5.
5-10
2.5 Limitations of the LWR model

The kinematic wave model of LWR is only a first approximation of the real traffic flow process and
is deficient in describing a number of traffic features of potential importance. The include (i) driver
differences, (ii) shock structure, (iii) forward moving waves in queued up traffic and (iv) traffic
instability (Daganzo 1997).
The LWR model is inadequate for modeling light traffic because it does not recognize the
segregation of fast and slow drivers in the traffic stream owing to passing. An adequate model for
traffic where significant amount of passing takes place should track both fast and slow drivers and
their interactions, not by lumping both types of drivers together through a single descriptor such as
density k(x, t). When drivers travel at roughly the same speed, however, the LWR model can still
be used for light traffic, where density is independent of concentration and flow increases linearly
with density.
The second deficiency of the LWR model lies in its treatment of shock waves. The shocks in
the LWR model has no width, i.e., no transition region. A vehicle enters a shock thus dropping its
speed in no time, implying an infinite deceleration. In reality, shocks always have a structure, i.e., a
transition region of a few vehicles’ length in which vehicles decelerate at finite rates. This deficiency
can be addressed through introducing higher-order approximations of traffic dynamics (Kühne
1984,1989) or using a microscopic description (Newell 1961). It is one of the author’s (Zhang)
opinion that as long as the accurate knowledge of vehicle acceleration is not of main interest, the
lack of a shock structure is not a major failing of the LWR theory, because in comparison with
the space scale one is modeling, a few vehicle lengths of shock width can be practically considered
as nil. To most applications, the important matter is whether the LWR theory gives a reasonable
estimate of shock speeds. Empirical evidence indicates that it does (Chapter 4 in Daganzo 1997).

The LWR theory has only one family of waves which travels at a speed of f∗ (k). These waves

always travel against the traffic stream because f∗ (k) < v∗ (k). In particular, f∗ (k) < 0 when
concentration k exceeds a critical value k∗ at which flow rate is maximal. An early study of tunnel
traffic, however, revealed that a different type of waves exists in real traffic (Edie and Baverez
1967). In analyzing the tunnel traffic data, Edie and Baverez (1967) noted that “ small changes
in flow may not propagate at a speed equal to the slope of the tangent to a steady-state q-k curve
as suggested by the hydrodynamic wave theories of traffic flow. Instead, they are carried along at
about stream speed or only slightly less than stream speed right up to saturation flows, at which
level they suddenly reverse directions.” This observation indicates that apart from the family of
waves that travels against the traffic stream, there’s at least another family of traffic waves that
travels with the traffic stream, even in congested traffic. The findings of Edie and Baverez (1967)
prompted Newell (1965) to suggest a fundamental diagram with multiple branches—one branch for
free flow, one for acceleration flow and another for deceleration flow. The extended LWR theory
with this fundamental diagram was able to explain not only the forward wave motion in queued-up
traffic, but also the instability found in tunnel traffic.
In comparison with other deficiencies of the LWR theory, the fourth deficiency, namely its
5-11
inability to model traffic instability, has more serious consequences, for traffic instability is at the
very heart of the traffic congestion phenomenon. The LWR model is always stable in the sense
that traffic disturbances, small or large, are always dampened. In other words, a driver who obeys
the LWR driving law always responds to stimulus properly, i.e., he always manages to change
his speed in the right amount of time and with the right magnitude that he simply absorbs the
disturbance. In fact the reaction time of a LWR driver is zero and the rate of adjustment he makes
is infinite. In reality, a driver responds to traffic events with a time delay, and not always precisely.
As a result, some disturbances in real traffic may get magnified as they propagate through the
traffic stream, causing traffic break-downs (stop-and-go) that could last for several hours. Such
stop-and-go traffic patterns exhibit, in physical space (i.e., x − t domain), periodic oscillations with
amplitude-dependent oscillation time, and in phase space (i.e., q − k or k − u domain), hysteresis
loops and wide scatter of data points.
Evidences of traffic instability and the resulting stop-and-go flow pattern are found on highways
around the globe. Perhaps the most impressive measurements of transients and stop-start wave
formation are gained from European freeways. Due to space restrictions, there are numerous
freeways with two lanes per highway in Europe. These freeways, often equipped with a dense
measurement grid not only for volume and occupancy but also for speed detection, show stable stop-
start waves lasting in some cases for more than three hours. Measurement data exists for Germany
(Leutzbach 1991), the Netherlands (Verweij 1985), and Italy (Ferrari 1989). First we examine the
measurements from German highways. The German data were collected from the Autobahn A5
Karlsruhe-Basel at 617 km by the Institute of Transport Studies at Karlsruhe University (Kühne
1987). Each measurement point is a mean value of a two-minute ensemble actuated every 30 sec.
These data were collected during a holiday when no trucks used that stretch of road.
All the data sets have traffic densities over the critical density and show signs of instability (i.e.,
stop-start waves with more or less regular shape and of long duration - in some series up to 12
traffic breakdowns). These data can characterized in the time domain by their oscillation times T
and magnitudes A. These characteristic features derived from the data shown in Figures 5.9a,b
and 5.9c,d (in the original text) (Kühne 1987,Michalopoulos and Pisharody 1980) are listed in
the following table:
Table 1: Oscillation time and magnitudes of stop-and-go traffic from German measurement
oscillation time T 16 min 15 min 7.5 min 5 min

amplitude A 70km/h 70 km/h 40 km/h 25 km/h
measurement figure 2a 2b 3a 3b
These characteristic values show a proportionally between amplitude and oscillation time. This
strong dependence is a sign for the non-linear and anharmonic character of stop-start waves. In
the case of harmonic oscillations, the amplitude is independent of the oscillation time as the linear
5-12
pendulum shows. Obviously, the proportionally holds only for the range between traffic flow at a
critical lane speed of about 80 km/h (= speed corresponding to the critical density kc . 25 veh/km)
and creeping with jam speed of about 10 km/h. For oscillations covering the whole range between
free-flow speed and complete gridlock, saturation effects will reduce the proportionally.
Examples of stop-start waves from other locations, such as the Netherlands (Verweij 1985),
Japan (Koshi et al. 1983), Italy (Ferrari 1989) and the U.S.A. also show similar characters as
those observed on German highways. Such oscillations, when viewed from the q − k phase plane,
show wider scatter of data points in the congested regime. Embedded in the scatter are sharp
drops (often referred to as the capacity drop) and hysteresis loops (e.g., Treiterer and Myers 1974)
of irregular shapes that differ from the equilibrium phase curve q = f∗ (k) and cannot be simply
explained away by stochastic arguments. All but the first deficiencies of the LWR model can be
addressed, to various degrees of success, through the introduction of higher-order approximations
or “dynamic” fundamental diagrams. This leads to two classes of traffic flow models—higher-order
and lower-order continuum models. Higher-order models introduce a dynamic speed-concentration
or flow concentration relation that accounts for driver reaction time and anticipation of traffic
conditions ahead, i.e.,
v(x, t + τ ) = V∗ (k(x + Δx, t)),
q(x, t + τ ) = kV∗ (k(x + Δx, t)).
The approximations of these dynamic relations lead to evolution equations for travel speed or flow
rate (Payne 1971, Whitham 1974, Zhang 1998). Consequently the traffic model becomes a system
of partial differential equations and its solutions, when shown in the phase plane, deviate from
the equilibrium fundamental diagram, producing the scatter and forward waves in the congested
region. Lower-order models, on the other hand, models different traffic motions—acceleration,
deceleration and coasting—explicitly on the fundamental diagram. Their fundamental diagrams
f∗ (k) has multiple branches and connecting curves, each describes a particular kind of motion
(Newell 1965, Zhang 2001). For one reason or another, higher-order models are more widely studied
and used than lower-order models. Therefore we will focus our presentation on higher-order models
in the remaining text. Readers who are interested in lower-order traffic flow models can refer to
Newell (1965), Daganzo (1999a,b) and Zhang (2001).
3 Higher-order continuum models

The development of higher-order traffic flow models again originated from the seminal work of
Lighthill and Whitham (1955), in which they suggested the following higher-order extension of
their first-order model:
The LW model: qt + Cqx + T qtt − Dqxx = 0, (25)
where C is convection speed, T is a reaction time constant and D is the diffusion coefficient.
Because of a lack of strong experimental evidence in support of such an extension, higher-order
5-13
approximations of traffic flow were not pursued further till 1971, when Payne (1971) and later
Whitham (1974) derived a so-called ‘momentum equation’ from a car-following argument:
c20 v∗ (k) − v
kvt + vvx + kx = , (26)
k τ
where v∗ (k) is the equilibrium speed-concentration relation, c0 < 0 is the ‘sound’ speed, and is
given by c20 = μτ , where μ is often referred to as the anticipation coefficient and τ relaxation time
2 . Recently, Zhang (1998) proposed a new addition to the existing momentum equations:
v∗ (k) − v
vt + vvx + kv∗ (k)2 kx = , (27)
τ
which is structurally similar to the momentum equation of Payne (1971) and Whitham (1974).
On the left hand side of the momentum equations, the second term is the change of speed
due to convection, and the third term captures drivers’ adjustment to travel speeds owing to
anticipation. The term on the right hand side captures drivers’ affinity to equilibrium travel speeds.
In the momentum equations, the acceleration of a vehicle, expressed by the material derivative vt +
vvx , responds negatively to the increase of concentration downstream, and positively (negatively)
to travel speeds that are lower (higher) than the corresponding equilibrium speeds for the same
concentration. As a result, travel speed v in both momentum equations usually differs from the
equilibrium speed v∗ (k) under the same traffic condition, but this difference is reduced over time
because of relaxation effects. The parameter τ decides the strength of relaxation. In literature τ is
often interpreted as driver reaction time, whose value ranges from 1 sec to 1.8 sec.

Through the definition of a concentration dependent sound speed c(k) (= kv∗ (k), or c0 , or
− μτ ), both momentum equations can be expressed in a general form:
c2(k) v∗ (k) − v
vt + vvx + kx = . (28)
k τ
This evolution equation of travel speed, coupled with the continuity equation
kt + (kv)x = 0, (29)
forms what we call a generalized PW higher-order model. The general PW model comprises a
system of partial differential equations that can be compactly expressed using vector notation

k v k k 0
+ c2 (k) = v∗ −v
. (30)
v t k v v x τ
or
Ut + A(U )Ux = R(U ), (31)

k v k 0
where U = , A(U ) = c2 (k) and R(U ) = v∗ −v .
v k v τ

2 v∗
There are other versions of (26)that differ in values taken by μ. Payne (1971), for example, gives μ = − 2
5-14
This system is strictly hyperbolic because its characteristics, the eigenvalues of the Jacobian
matrix A(U), are real and distinctive
λ1,2 = v ± c(k), λ1 < λ2 . (32)
The same is true for the LW model, because, through introduction of two auxiliary variables w = qt
and z = qx the LW model can be transformed into a system of PDEs

w 0 1 w 0
+ =− , (33)
z −DT 0 z w + Cz
t x
whose characteristics are

D
λ1,2 = ∓ , λ1 < λ2 .
T
The two higher-order models differ, however, in that the general PW model is genuinely nonlinear
while the LW model is linearly degenerate.
Although both the LWR model and the general PW model are hyperbolic, the latter has two
characteristics, one is always slower than traffic and the other always faster than traffic, owing to

c(k) < 0. This is constrasted to the single characteristic of the LWR model, λ∗ = v∗ (k) + kv∗ (k)
that is always slower than traffic. These differences have profound consequences on the behavior
of these models, which we shall discuss in the context of various kinds traffic waves.
3.1 Propagation of traffic sound waves in higher-order models

We first note that traffic sound waves—the propagation of small disturbances in homogeneous
traffic—travel at characteristic speeds in the LWR model. In higher-order models, there are two
families of characteristics, thus two characteristic speeds. If a disturbance is still propagated at
characteristic speeds, then it travels in both directions of traffic with different speeds, reaching
drivers from front and behind. This can be checked through writing the higher-order model in
question, here the LW model, in a special form
⎧ ⎛ ⎞⎛ ⎞⎫
⎨ D ⎠⎝ D ⎠⎬
(∂t + C∂x ) + ⎝∂t + ∂x ∂t − ∂x q = 0.
⎩ T T ⎭
where ∂t + () ∗ ∂x is called a wave operator.

From this special form, one can clearly see that the LW model possesses three families of waves:
the first order wave traveling at the convection speed c∗ = C (which is also the characteristic speed
of the corresponding first-order model qt + Cqx = 0), the slower second order wave traveling at the
first characteristic speed c1 = − D T and the faster second order wave traveling at the speed of the

second characteristic c2 = D T . For certain parameter values of D and T , the second characteristic
speed can be greater than the convection speed, indicating that fast waves reach traffic from behind
(the rear view mirror effect). In fact, this is also required for LW traffic to be stable. Otherwise
5-15
the first-order signals would violate the second-order signals, which leads to traffic instability. The
general stability condition for the above higher-order model is
c1 ≤ c ∗ ≤ c 2 , (34)
that is, the first order waves are sandwiched between the two second order waves. This stability
condition can be derived from Fourier stability analysis (e.g., Whitham 1974, Kühne 1984, Zhang
1999).
For the general PW model, small perturbations around equilibrium points (k0 , v0 ) would prop-
agate in the same way as the
waves in the LW model, only with different wave speeds. Instead of
c∗ = C, c1 = − T , c2 = D
D
T , here we have c∗ = λ∗ (k0 ), c1 = v0 + c(k0 ), and c2 = v0 − c(k0 ).
Because of c(k) < 0, fast waves in the general PW model also reaches vehicles from behind.
There are subtle differences among special cases of the general PW model. In the PW model,
where c(k) = c0 , the stability condition (34) can be violated and waves grow in magnitude and
eventually become shocks in the form of roll waves (Whitham 1974, Kuhne 1984) while in Zhang’s

model where c(k) = kv∗ (k) the stability condition is always satisfied and waves always damp in
magnitude (Zhang 1999). Thus, like the LWR model, Zhang’s model is also inherently stable.
Moreover, the asymptotic behavior of small perturbations of the kind k = k0 + ξ(x, t), v = v0 +
w(x, t) near equilibrium point (k0 , v0 ) (v0 = v∗ (k0 ) and ξ, w are small perturbations) in the PW and
Zhang models also differ. Such perturbations to Zhang’s model can be accurately approximated by
a convection equation
ξt + λ∗ (k0 )ξx = 0, (35)
while those to the PW model can be accurately approximated by a diffusion equation
ξt + λ∗ (k0 )ξx = D∗ ξxx , or (36)
ξt = D∗ ξx x (37)

if a moving coordinate x = x − λ∗ (k0 )t, t = t is used, where D∗ is the diffusion coefficient, and
is given by −(λ1(k0 , v0 ) − λ∗ (k0 ))(λ2 (k0 , v0 ) − λ∗ (k0 )), a positive constant (Whitham 1974, del
Castillo et al. 1994, and Zhang 1999). Driven by relaxation, any non-equilibrium state (k, v) in the
PW model will become closer and closer to its corresponding equilibrium state (k, v∗ (k)) with the
increase of time. This latter behavior, together with the properties of Eq. 37 (i.e., its solutions are
smooth), implies that at the end of a queue the PW model does backward smoothing to a sharp
density/speed profile, thus predicting possibly negative travel speeds (Daganzo 1995a). Zhang’s
model, however, is absent of this problem because its solutions are not diffusive.
3.2 Propagation of shock and expansion waves

When traffic conditions undergo sharp transitions, shocks or expansion waves arise. The LWR
theory possesses both types of waves and whether a particular kind of wave arises is determined
by the entropy condition. A LWR shocks has zero width and travels at a particular speed given by
5-16
the Rankine-Hugoniot condition s = [f∗ ]/[k], which is derived from the integral conservation law.
The general PW model also has similar properties. It has both kinds of waves, actually two for
each kind—one associated with the first characteristic and the other with the second characteristic.
The first is referred to as 1-shock or 1-rarefaction waves, and the second 2-shock or 2- rarefaction
waves. Whether a shock or expansion wave arises in a general PW solution is also dependent on
entropy conditions, and they are as follows (Zhang 2000a):
1-Shock, (E-H1): λ1 (Ur ) < s < λ1 (Ul ), s < λ2 (Ur ), (38)

2-Shock, (E-H2): λ2 (Ur ) < s < λ2 (Ul ), s > λ1 (Ul ), (39)
1-Rarefaction, (E-R1): λ1 (Ul ) < λ1 (Ur ), (40)
2-Rarefaction, (E-R2): λ2(Ul ) < λ2 (Ur ), (41)
And the speeds of shocks are also given by the Rankine-Hugoniot conditions derived from
integral conservation laws. A difficulty arises, however, from the fact that the momentum equation
is not a conservation law, it is simply an evolution equation for travel speeds. As a result, we do
not have a unique corresponding integral conservation law for the momentum equation, that is,
there are many different integral conservation forms that lead to the momentum equation. Two
such examples are:
x2
∂ v2 v2
v(x, t)dx = + φ(k) (x1 , t) − + φ(k) (x2 , t) (42)
∂t x1 2 2
x2
v∗ − v c2 (k)
+ dx, φ (k) = ,
x1 τ k
x2
∂ (kv)2 (kv)2
(kv)(x, t)dx = + ψ(k) (x1 , t) − + ψ(k) (x2 , t) (43)
∂t x1 k k
x2
kv∗ − kv
+ dx, ψ (k) = c2 (k).
x1 τ
We cannot tell, from physical principles, which integral form is “correct”. The selection of a
particular integral form should therefore be guided by field observations, i.e., choosing the one that
produces closest shock speeds to field measurements. For illustration purposes, we shall take the
simplest, i.e., the first integral form in our subsequent presentations. The same arguments can be
applied directly to other integral forms. The first integral momentum equation leads to following
conservative differential form:

v2 v∗ − v
vt + + φ(k) = .
2 x
τ
together with the conservation of mass
kt + (kv)x = 0,
5-17
we obtain the system of “conservation ” laws with a source term:
Ut + F (U)x = R(U ), (44)

2
t
where F (U ) = kv, v2 + φ(k) , and U, R are as defined before.
The shock speeds of the general PW model is not affected by the presence of the source term,
and is still given by the Rankine-Hugoniot condition:
s[U ] = [F (U )], (45)
or
v2
s[k] = [kv], s[v] = + φ(k) .
2
These equations imply a certain relation between k and v. This relation leads to the shock curves
(1-shock & 2-shock ) in the k − v phase plane (Zhang 2000a).
Expansion (or rarefaction) waves of the general PW model, however, are influenced by the
relaxation source term. The strength of this influence is determined by the relaxation time. In a
short time (compared to the relaxation time), one can neglect the effect of the source term and
obtain certain relations between k, v in an expansion wave solution. These are the 1-rarefaction and
2-rarefaction curves in the k − v phase plane (Zhang 2000a). Over time, however, the cumulative
effects of the source term, relaxation, build up into rarefaction wave solutions, and one addresses this
problem by modifying the solutions obtained without the source term. Also because of relaxation,
the general PW model either approaches to the LWR model or a viscous LWR model kt + (kv∗ )x =
D∗ kxx , depending on the choice of c(k).
Clearly one cannot ignore the effects of relaxation. Nevertheless, the study of the general PW
model without relaxation
Ut + F (U)x = 0 (46)
reveals much information about the properties of the general PW model with relaxation. And
computations of numerical solutions to the general PW model often rely on the Riemann solutions
of (46). Therefore we provide here the solutions of Riemann problems to the general PW model
without relaxation (46) (see Zhang 2000a for details).
The solution to the Riemann problem
Ut + F (U )x = 0, (47)

Ul , x < 0
U (x, t = 0) = (48)
Ur , x > 0
are of 8 kinds:
1. 1-shock:
2(kl − kr )(φ(kl ) − φ(kr ))
H1: vr − vl = − , kl < kr . (49)
kl + kr
5-18
2. 2-shock:
2(kl − kr )(φ(kl ) − φ(kr ))
H2: vr − vl = − , kl > kr . (50)
kl + kr
3. 1-rarefaction:
c(k)
R1: vr = dk, kl > kr (51)
kl k
4. 2-rarefaction:
c(k)
R2: vr = dk, kl < kr (52)
kl k
5. 1-shock + 2-shock:

2(kl − km )(φ(kl ) − φ(km ))
H1: vm − vl = − , kl < km . (53)
kl + km

2(km − kr )(φ(km ) − φ(kr ))
H2: vr − vm = − , km > kr . (54)
km + kr
6. 1-rarefaction + 2-rarefaction:

c(k)
R1: vm = dk, kl > km (55)
kl k

c(k)
R2: vr = dk, km < kr (56)
km k
7: 1-rarefaction + 2-shock:

2(kl − km )(φ(kl ) − φ(km ))
H1: vm − vl = − , kl < km . (57)
kl + km

2(km − kr )(φ(km ) − φ(kr ))
H2: vr − vm = − , km > kr . (58)
km + kr
8: 1-shock + 2-rarefaction:

2(kl − km )(φ(kl ) − φ(km ))
H1: vm − vl = − , kl < km . (59)
kl + km

c(k)
R2: vr = dk, km < kr (60)
km k
where Um = (km , vm )t is an intermediate state that provides the transition from a 1-wave to a
2-wave.
The transitions are most clearly seen from the phase diagram (Fig. 5). Note that for a state
Ul , there are four special curves (H1, R1, H2, R2) emanating from that point, dividing the quarter
plane of k − v into four regions: I, II, III, IV. If the downstream Ur in the Riemann problem falls
in region I, the solution would be of Type 6 ( R1+ R2); in region II, Type 7 (R1+ H2), in region
III, Type 5 (H1+H2), and in region IV, Type 8 (H1 + R2). Because of the entropy conditions, the
intermediate states always fall on a 1-wave curve (i.e., R1, H1). When Ur falls on a particular wave
curve, then the transition involves only one kind of wave and no intermediate state is produced.
These are the types 1-4 solutions.
5-19
Figure 5: Phase transition diagram in the solution of Riemann problems (adopted from Zhang
(2000a))
3.3 Traveling waves, instability and roll waves

Besides traffic sound waves, shock waves and expansion waves, there is another kind of waves in
the general PW theory. This is the so-called traveling waves (Whitham 1974, Zhang 1999). It is
a wave that has a smooth profile and travels at a constant speed, and takes the following form:
(k, v)(x, t) = (k, v)(χ), χ = x − st (Fig. 7(a)). For the traveling wave solution to arise, a certain
stability condition has to be met. When this condition is violated, another kind of waves, the roll
waves, may arise. A roll wave is a series of smooth, monotonic profiles separated by jumps (Fig.
6). The existence of roll waves is a direct consequence of traffic instability.
In this section we first obtain the traveling waves, then construct roll waves based on the
traveling wave solution. Recall that a traveling wave is a steady profile with a translation speed s:
(k, v)(x, t) = (k, v)(χ), χ = x − st. (61)
Substitute the traveling wave solution (61) into the general PW model, one obtains
−skχ + (kv)χ = 0, (62)

c2 (k) v∗ − v
−svχ + vvχ + kχ = , (63)
k τ
5-20
which after integration and further substitution of terms become

2 Q2
c (k) − 2 kχ = f∗(k) − Q − ks, (64)
k
where Q is an integration constant and
Q = k(v − s). (65)
Because the two states at x = ±∞, (k, v) = (k1,2, v1,2 ) both satisfy (65), one can compute the
speed of the traveling wave
k1 v1 − k2 v2
s= ,
k1 − k2
which is the same as the shock speed given by the Rankine-Hugonoit condition.
Let h(k) denote the right hand side of (64), then h” (k) = f ” (k) < 0. Moreover, h(k) crosses
the k−axis at most twice at ka , kb , ka ≤ kb . These crossing points corresponding to equilibrium
concentration values and in between them h(k) > 0 (Zhang 1999). Since both (k, v) = (k1,2 , v1,2 )
are equilibrium points, k1,2 are also roots of h(k). Moreover, we have k1 < k2 , or kχ > 0 from the
fact that f∗ (k) is strictly concave. Thus, for a smooth profile of k(χ) to exist, we must have
Q2
c2 (k) − > 0,
k2
which yields the following stability condition
v + c(k) < s < v − c(k). (66)
That is, the traveling wave must travel slower than the fast characteristic and faster than the slow
characteristic.
When the stability condition is met, the smooth traveling wave profile can be obtained by
integrating
2
dχ c2 (k) − Q
k 2
= , (67)
dk f∗ (k) − Q − ks
which is given by:
k c2 (η) − Q2
η2
χ= dη. (68)
f∗ (η) − Q − ηs
When the stability condition (66) is violated, that is, the traveling wave travels either faster
than the fast characteristic or slower than the slow characteristic, a smooth profile connecting k1
and k2 is no longer possible and a discontinuity (i.e., shock) must be inserted at the location where
the wave turns back on itself. This time one still obtains a monotonic profile but with a shock
separates two smooth pieces (Fig. 7(b)), and the speed of the shock is determined by the R-H
shock condition.
There is a special case when the stability condition is not met. This is the case where both the
numerator and denominator of (67) vanishes, which leads to non-smooth and periodic solutions to
5-21
Figure 6: Roll waves in the moving coordinate χ
Figure 7: Traveling waves and shocks in the PW model
5-22
the general PW model, often referred to as roll waves in hydraulic literature. Note that the physical
solution of these two algebraic equations is Q = −k0 c(k0 ), where Q is the constant flux measured
relative to the moving coordinate χ and k0 is the critical density that makes both the numerator
and denominator of (67) vanish. It is necessary that (k0 , v0 ) is an equilibrium state, which fixes the
shock speed s = v∗ (k0 ) + c(k0 ). The critical state also fixes the integration constant in the profile
equation (68). The roll waves can then be constructed, in the same manner as shown in Dressler
(1949), by piecing together the smooth profiles with shocks at appropriate locations, which in turn
are determined by the R-H shock conditions. Although the existence of roll wave solutions in the
PW model were known to transportation researchers not long after the development of the PW
model (e.g., Leutzbach 1985), no one to date has obtained specific roll wave solutions to compare
with real world observations of stop-start waves. With the development of Section 3.2, however,
this can now be done.
3.4 Summary and Discussions

The PW-like higher-order traffic flow models extend the basic kinematic wave traffic flow model
in two ways: they allow for non-equilibrium phase transitions and introduce instability. These
are achieved through the addition of a momentum equation that describes speed evolution. The
resulting flow obtained from these higher-order models, however, are not all that different than
those obtained from the basic kinematic wave model—shocks, for example, exist in both types of
models. Moreover, owning to relaxation, similar types of solutions (e.g. shocks or expansion waves)
of these two families of models become much similar in long time (typically 10τ ).
On the other hand, some significant differences also exist between the two family of models. For
one, higher-order models have two family of characteristics and waves as compared to one family of
the kinematic wave model. While the first family of waves in the higher-order models behave much
like those in the kinematic wave model, the second family behaves quite differently—they travel
faster than traffic, and reaches vehicles from behind. This property of the second characteristic
has led to doubts about the validity of higher-order traffic models and interesting discussions over
the pros and cons of higher-order approximations of traffic flow in general (e.g., Daganzo 1995a,
Papageorgiou 1998, Lebacque 1999, Zhang 2001). Because following vehicles usually cannot force
leading ones to speed up or slow down. fast-than-traffic waves are quite unrealistic when traffic is
on a single-lane highway. When traffic is flowing on a multi-lane highway and passing is allowed,
such faster-than-traffic waves do arise as a result of 1) lane-changing, or 2) averaging, or both (see
Zhang 2000c for details). Another difference between the two types of models is that higher-order
models contain traveling wave and roll wave solutions while the KW model does not. Roll wave
solutions are particularly interesting because of their similarity to observed stop-start waves.
Like the kinematic wave model, higher-order models can also be extended to model inhomo-
geneous roads. This is particularly straightforward with the PW model— one simply replace the
homogeneous equilibrium speed-density relation v∗ (k) with a space-dependent one v∗ (k, x). The
extension of higher-order models to model road networks, however, is not as straightforward and is
5-23
a worthwhile research topic for traffic flow researchers.

One of the motivations for developing the PW model was to remove shocks from the model
solutions. This, however, is not fulfilled by the higher-order models that we have covered up to
now, although the PW model does admit smooth traveling wave solutions. Higher-order space
derivatives, in the form of kxx or vxx must be introduced to obtain shocks with a structure. We
call these models diffusive or viscous (higher-order) models. In contrast, PW like models are called
inviscid higher-order models. The next section briefly introduces some of the popular viscous traffic
models, and a stochastic extension to a particular viscous higher-order model.
4 Diffusive, viscous and stochastic traffic flow models

4.1 Diffusive and viscous traffic flow models
The first diffusive traffic flow model, which was mentioned in the classical book of Whitham (1974)
was obtained by considering a flux that is dependent not only on vehicle concentration, but also
on the concentration gradient:
q = f∗ (k) − νkx ,
which, after substitution into the conservation equation, leads to
kt + f∗ (k)x = νkxx , (69)
where ν is a positive parameter.

This diffusive model, when f∗ (k) is quadratic, can be manipulated into the following form

ct + ccx = νcxx , c = f∗ (k), (70)
which is the well-known Burger’s equation.

Analytical solutions to the Burger’s equation with given initial data can be obtained through
the so-called Cole-Hopf transformation. We refer the reader to Whitham (1974, pp 96-112) for the
detailed solution formulas and only discuss the qualitative properties of various kinds of solutions
to the Burger’s equation. These are:
1. The solution to the initial value problem ct + ccx = νcxx , c(x, t = 0) = F (x) always exists,
and is smooth after t > 0.
2. When ν → 0, the solution approaches that of ct + ccx = 0.

cl x < 0
3. For initial data F (x) = cl > cr (A step function), the solution is a traveling
cr x > 0,
wave c(x − st), with s = c1 +c 2
2 . The width of the traveling wave, as measured by the range
ν
where 90% of the change in cl − cr occurs, is proportional to cl −c r
. As ν → 0, the width of
the traveling wave becomes nil and the traveling wave becomes a shock of ct + ccx = 0. Thus
the viscous model provides a shock structure to the LWR model.
5-24
4. For initial data F (x) = Aδ(x) (A single hump), the solution is a nonlinear diffusion wave,
A
similar to those of the heat equation ct = νcxx when R = 2ν 1.
5. Also, N-waves and periodic waves can be found in the solution with proper initial data.
Clearly, the addition of the second order derivative kxx to the LWR conservation law does two
things to it because of diffusion: 1) it smoothes the shocks of the LWR conservation law, thereby
providing a shock structure, and 2) it guarantees the uniqueness of solution for small ν, thereby
providing a way to pick solutions from the LWR conservation law. This latter property is often
exploited in numerical computations of shock wave solutions.
The diffusion corrected conservation law of (69), when f∗ (k) is not quadratic, can be approx-
imated by the Burger’s equation. In fact, the aforementioned properties of the Burger’s equation
(except Property #2, which must be modified) are shared by all known diffusion corrected or
viscous traffic flow models, including the following popular viscosity-corrected PW model:

k v k k 0 0 0 k
+ c2 (k) = v∗ −v + . (71)
v t k v v x τ 0 ν v xx
The viscosity-corrected PW model, however, can become unstable in certain ranges of traffic and
therefore has additional properties. These properties, including the collapse of homogeneous traffic
under local and global perturbations and the formation of vehicle-clusters in stop-and-go traffic, are
well documented in (Kerner &Konhaüser 1993, 1994; Kerner, Konhaüser & Schilke 1995; Kerner and
Rehborn 1999; Kühne & Beckschulte 1993). Interested readers are referred to the aforementioned
literature for detailed discussions of these properties.
4.2 Acceleration noise and a stochastic flow model

Same as section 5.3 in the original text.
5 Numerical approximations of continuum models

All of our continuum models are described by partial differential equations, some (i.e, LWR, PW
and Zhang) are hyperbolic while others (e.g., the viscous model of (69)) parabolic. Proper ini-
tial/boundary conditions must be prescribed to these equations to form a well posed problem. The
solutions of continuum models involving general initial/boundary conditions are tedious, if not
difficult, to obtain analytically. Numerical procedures are often employed to solve such problems.
Typically, finite difference methods are applied to solve hyperbolic traffic flow models while
finite difference or finite element methods are used to solve parabolic traffic flow models. Both
methods start with a discretization of the time-space domain (t ≥ 0, −L < x < L), with the
following grid mesh being the most common:
xi = ih, i = 0, ±1, ±2, · · · , L/h,

tj = jk, j = 0, 1, 2, · · · , T /k.
5-25
where h ≡ Δx and k ≡ Δt, and (xi , tj ) are the grid points of this mesh.
Let the values of U (x, t) on those grid points denoted by Uij (see Fig. 8). Then the time-space
derivatives in a continuum model can be approximated using values at these grid points, and we
obtain a set of finite difference equation(s) (FDE). For example, the space derivative Ux can be
approximated in a number of ways:
⎧ j j
⎪
⎪ Ui −Ui−1
⎪
⎪ , Forward Difference
⎨ j h j
[Ux ]ji = Ui+1 −Ui−1
, Center Difference (72)
⎪
⎪ 2h
⎪
⎪ j
Ui+1 −Uij
⎩ , Backward Difference
h
Figure 8: Time-space grid
However, the numerical approximation of continuum models is not a simple exercise of replacing
continuous variables and their derivatives with discrete ones and their differences. This is particu-
larly true for hyperbolic traffic models, because in these models discontinuities or shock waves can
develop spontaneously even from smooth initial data. The existence of shocks presents a major
challenge to the development of numerical approximations that are consistent with and convergent
to the model equations when the mesh size is further and further refined. A valid approximation
must meet the following three conditions:
1. it is consistent with the original PDE, i.e., the FDE converges to the PDE when h and κ
approach 0 (consistency),
2. numerical errors introduced by the FDE do not increase over time ( stability), and
3. its solutions converge to the right solutions of the original PDE when h and κ approach 0
(convergence).
In this section we present two consistent, stable and convergent finite difference approximations
of the system (31) with initial/boundary data of (73) and a finite element approximation of the
5-26
viscous model (33) with the same initial and boundary data. It should noted that the LWR,
PW and Zhang models can all be expressed in the form of (31). Therefore the finite difference
approximations presented here applies to all three models. We begin our presentation with finite
difference approximations.
5.1 Finite difference methods for solving inviscid models

In this section we present two finite difference schemes to solve the system of (31) with the following
initial/boundary data:
I.C.: U(x, 0) = U0 (x), −L ≤ x ≤ L

(73)
B.C.: U(−L, t) = U− (t), U (L, t) = U+ (t), t ≥ 0.
where U0 and U± are vector valued functions of space and time, respectively, and may contain a
countable number of jumps. This system of equations include the Kinematic wave model of LWR,
the higher-order models of Payne-Whitham and of Zhang.
It turns out that a particular form of hyperbolic PDEs, called the conservative form, is specially
suited for developing finite difference schemes that ensure the aforementioned three conditions. A
conservative form of (31), for example, is given by (44), i.e.,
Ut + F (U)x = R(U ).
(44) is called a conservative form because it arises from certain conservative phenomena in convec-
tive transport:
∂
U (x, t)dx + F (U )dx = R(U )dx. (74)
∂t L ∂L L
For example, the conservation of vehicles on a finite road segment [x1 , x2 ] leads to a specific case
of (74): x2
∂ x2
k(x, t)dx + (kv)(x2 , t) − (kv)(x1 , t) = r(k, v)dx. (75)
∂t x1 x1
Using the conservative form, we can develop a conservative finite difference approximation of (44).
The advantage of a conservative finite difference approximation is that it ensures the correct com-
putation of shock speeds. A finite difference approximation of (44) is conservative if it can be
written as
j
Uij+1 − Uij F̃ (Ui+1 , Uij ) − F̃ (Uij , Ui−1
j
) j j j
+ = R̃(Ui+1 , Ui , Ui−1 ), (76)
κ h
where F̃ in (76) is called numerical flux (explained later) 3 . When a finite difference scheme is
in conservative form, the condition for consistency is particularly simple. It requires that the
numerical flux function satisfies
F̃ (U, U) = F (U).
3
The argument list of the numerical flux function can involve more than two nodal points, depending on the
required accuracy of the finite difference approximation. (76) uses two nodal values to compute its numerical flux
and is first order accurate.
5-27
Examples of consistent conservative finite difference schemes in traffic flow are the finite difference
schemes of (Michalopoulos, Beskos & Lin (1984)), (Daganzo 1994), and (Lebacque, 1996) in the
scalar case (i.e., the LWR model), the finite difference scheme of Leo and Pretty (Leo and Pretty
1992) and Zhang (2000b) in the system case (e.g., the PW model).
A consistent, conservative finite difference approximation, if linear, is guaranteed to converge
to the correct solution if it meets a stability condition (LeVeque 1992). This condition, generally
referred to as the CFL (Courant-Friedrichs-Lewy) condition, says that the cell advance speed hκ
cannot be greater than the maximum absolute characteristic velocity, i.e.,

κ
max λi ≤ 1, i = 1, · · · , n.
h
Although this theorem has not been proven for most nonlinear systems, computational expe-
riences indicate that the CFL condition is sufficient to ensure convergence for a large number of
nonlinear systems. Thus we require the CFL condition in all of our finite difference approximations.
The remaining task in our finite difference approximations is to obtain the numerical flux
function that ensures the consistency, stability and convergence properties. Before explaining what
j
a numerical flux function is, we first make clear what Ui represents in our scheme of things. Suppose
that u(x, t) is a weak solution of the integral conservation law (74). We can write (74) as
x x tj+1 x
i+1/2 i+1/2 i+1/2
u(x, tj+1 )dx = u(x, tj )dx + R(u(x, t))dxdt (77)
xi−1/2 xi−1/2 tj xi−1/2
tj+1
tj+1
− F (u(xi+1/2 , t)dt − F (u(xi−1/2 , t)dt
tj tj
where i − 1/2 and i + 1/2 denotes the left and right boundary of cell i (see Fig. 8), respectively.
If we interpret Uij as the cell average
xi+1/2
1
Uij = u(x, tj )dx (79)
h xi−1/2
then tj+1
j 1
F̃ (Ui+1 , Uij ) = F (u(xi+1/2 , t)dt (80)
κ tj
is the average flux passing through the cell boundary xi+1/2 in the time interval (tj , tj+1 ), and
tj+1 x
1 i+1/2
hR̃ = R(u(x, t))dxdt
κ tj xi−1/2
is the average inflow into cell i from the source during time interval (tj , tj+1 ). With these definitions,
the integral conservation law (74) reduces to (76), and this is why (76) is called a conservative
approximation.
There are a number of ways to construct a numerical flux function, perhaps the most intuitive
and illustrative is the one obtained using Godunov’s finite difference method. The Godunov method
solves locally a Riemann problem at each cell boundary for the time interval (tj , tj+1 ), using the
5-28
cell averages Uij as initial data. It then pieces together these Riemann solutions at time tj+1 and
average them using (79) to obtain new initial data for tj+2 , and this process is repeated until T /k
is reached. Recall that the Riemann problem for the homogeneous equation
Ut + F (U)x = 0 (81)
of our continuum traffic models have been solved in Sections 2 and 3. We can then apply the
principle of superposition to solve the traffic models with source terms:
Ut + F (U)x = R(U ). (82)
Using the aforementioned procedure, we obtain a Godunov type of difference equation for (44):

∗j ∗j
Uij+1 − Uij F Ui+1/2 − F Ui−1/2
+ = R̃. (83)
k h
in which the numerical flux function reads

j j ∗j
F̃ (Ui+1 , Ui ) = F Ui+1/2 , (84)
and the source flux is computed by

Ui+1 + Ui−1
R̃ = R .
2
∗j
The variables Ui+1/2 , i = 0, ±1, ±2, · · · , ±L/h are obtained from solving a series of Riemann
problems to the homogeneous equation at cell boundaries. Owing to space limitations, we cannot
in this monograph fully describe how this is done for higher-order models (interested readers are
∗j
referred to Zhang 2000b). However, the computation of Ui+1/2 is particularly simple when the
equilibrium relation v = v∗ (k) is assumed. It leads to the following finite difference equation

∗j ∗j
kij+1 − kij f∗ ki+1/2 − f∗ ki−1/2
+ = 0, vij+1 = v∗ (kij+1 ), (85)
k h

∗j
where the cell boundary flow f∗ ki+1/2 is given by a simple formula (LeVeque 1992, Bui, et. al.
1992): ⎧ j j
⎨ min j
ki ≤k≤ρji+1 f∗ (k),
∗j
if ki < ki+1
f∗ ρi+1/2 = j j .
⎩ max j j f ∗ (k), if k > k
k ≥k≥ρ i
i i+1 i+1
When f∗ (k) is concave, this formula can be further streamlined through the introduction of a supply
and demand function for each cell (Daganzo 1994, 1995b; Lebacque 1996):

f∗ (kij ), if ji < k∗
Demand: D(i, j) = , (86)
f (k ∗), if ji ≥ k∗

f∗ (kij ), if kij > k∗
Supply: S(i, j) = , (87)
f (k∗ ), if kij ≤ k∗
5-29

where k∗ is the critical density at which f∗ (k∗ ) = 0. And we have

∗j
f∗ ki+1/2 = min{Di , Si+1 }, (88)
which also applies to boundary cells and bottlenecks.

Another difference approximation of (31) uses the idea of Lax-Friedrichs center differencing. It
leads to the following numerical flux:
j
j F (Ui+1 ) + F (Uij ) h Ui+1
j
+ Uij
F̃ (Ui+1 , Uij ) = − . (89)
2 κ 2
It is easy to check that this flux function meets the consistency requirement and the resulting finite
difference approximation
j j
U + Ui κ
Uij+1 = i+1 − j
F (Ui+1 j
) − F (Ui−1 ) + hκR̃
2 2h
is also conservative. This center difference scheme remained till recent years a popular choice
of approximation of the kinematic wave model (e.g., Michalopolous 1988, Michalopolous, Beskos
& Yamauchi 1984, Michalopolous, Kwon, & Khang 1991, Michalopolous, Lin, & Beskos 1987)
and being used lately to approximate higher-order models (Zhang 2000d). Zhang and Wu (1999)
investigated the convergence properties of this scheme and found that in comparison with Godunov-
type of schemes, the center difference scheme has faster convergence rate with respect to expansion
wave solutions and slower convergence rate with respect to shock solutions. The reason is that this
difference scheme has built-in numerical viscosity which smoothes shocks.
5.2 Finite element methods for solving viscous models

Apart from finite difference methods, the method of finite element is also employed to solve con-
tinuum traffic flow equations. In this section we show how the latter is used to solve the viscosity-
corrected PW equations.
First we introduce an auxiliary variable w : w = vx , and normalize all the state and time-space
variables in the following way:
k v w x t
k = v = w = x = t = . (90)
kjam vf vf vf τ τ
Then the unknown variables in the viscosity-corrected PW model can be expressed by a vector
⎛ ⎞
k
⎜ ⎟
η=⎝v ⎠ (91)
w
and the model itself by the following vector-valued quasi-linear partial differential equation:
Aηt + Bηx = C (92)
5-30
with (note that for convenience the ’s are dropped from the notations)
⎛ ⎞ ⎛ ⎞ ⎛ ⎞
1 0 0 v 0 0 −kw
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
A = ⎝0 1 0⎠ B = ⎝ 1k F1r 0 1
Re ⎠ C = ⎝ −vw +v∗ −v ⎠ , (93)
0 0 0 0 1 0 w
kinetic energy (inertia influence)

F r ≡ Froude number =
potential energy(pressure)
1 2 2
2 kvf vf
= 2 =
c0 k c0
length velocity vf2 τ
R ≡ Reynolds number = = . (94)
kinem. viscosity ν0
The problem also comes with possibly two initial conditions and six boundary conditions. Be-
cause of the hyperbolic nature of the viscosity-corrected PW model, however, only certain combi-
nations of these initial/boundary data are allowed.
In the finite element method, we replace the continuous functions
⎛ ⎞
k(x, t)
⎜ ⎟
η(x, t) = ⎝ u(x, t) ⎠ (95)
w(x, t)
by functions defined as a lattice:
η(x0 + iΔx, t0 + jΔt) ≡ ηi,j (96)
and all derivatives by center difference quotients:

1
ηx → (ηi+1,j+1 − ηi,j+1 + ηi+1,j − ηi,j ) (97)
2Δx
and the function values by the midpoint values:
1
ηt → (ηi+1,j+1 + ηi,j+1 − ηi+1,j − ηi,j ). (98)
2Δt
1
η → (ηi+1,j+1 + ηi,j+1 + ηi+1,j + ηi,j ) (99)
4
We then do step-wise integration, starting from time step j = 0 and ending at time step j = J, as
shown in Fig. (fig 5.16 from original text). To ensure the stability of the numerical procedure, an
implicit integration scheme is used to compute the unknown variables ηi,j , ηi+1,j (for notational
simplicity we’ll drop subscript j in the remaining text of this section). It turns out that the
Newtonian iteration procedure is perfectly suited for this purpose. In this procedure the variables
ηi , ηi+1 are replaced by an approximation η̃i , ηi+1
˜ and the deviations δηi , δηi+1 are computed by
linearizing the original equations. Denoting the deviation vector by
⎛ ⎞
δ ki
⎜ ⎟
δi ≡ ⎝ δ ui ⎠ (100)
δ wi
5-31
then the basic equations can be written in the form of
Ai δi+1 + Bi δi = Ri (101)
with
2 2 1 c2
α= ,β = ,κ = = 02 (102)
Δx Δt Fr vf
⎛ ⎞
β + αU + W Kx K
⎜ ⎟
Ai = ⎝ −Ue (K) − κ K
x 1
K 2 + ακ K β + W + 1 U − αν ⎠ (103)
0 α −1
⎛ ⎞
β − αU + W Kx K
⎜ ⎟
Bi = ⎝ −Ue (K) − κ K
1
K 2 − ακ K
x
β + W + 1 U + αν ⎠ (104)
0 α −1
⎛ ⎞
Kt + Kx U + K
⎜ ⎟
Ri = −4 ⎝ Ut + UW − Ue(K) + U + κ KKx − νWx ⎠ (105)
Ux − W
where abbreviations
1
K= (ki+1 + ki + ki+1,j + ki,j ) (106)
4
1
Kt = (ki+1 + ki − ki+1,j − ki,j ) (107)
2Δt
are used.
Starting with the initial condition as the lowest approximation
ηi = ηi,j=0 ηi−1 = 0 (108)
and using the left boundary condition

⎛ ⎞
0
⎜ ⎟
ki=0,j , Vi=0,j =⇒ δ0 = ⎝ 0 ⎠ (109)
δw0
the δi is computed recursively by

δi+1 = A−1
i (Ri − Bi δi ) (110)
as a function of δw0 , which in turn is determined by the right boundary condition

⎛ ⎞
δkI
⎜ ⎟
vi=I,j =⇒ δI = ⎝ 0 ⎠ . (111)
δwI
An alternative rearrangement of the deviations δi is possible in order to produce a tridiagonal

form which facilitates the fit of the boundary conditions (Kerner and Konhaüser, 1993).
5-32
5.3 Applications
5.3.1 Calibration of model parameters with field measurements
All the continuum models discussed in this monograph contain certain static relations and pa-
rameters. These relations include, for the LWR model, the fundamental diagram f∗ (k), and for
higher-order models, v∗ (k). Since f∗ (k) = kv∗ (k), knowing one would know the other. The pa-
rameters include, but not limited to, free flow speed vf , jam wave speed cj , sound speed c0 (< 0)
(PW model), jam density kjam , critical density kc , capacity qc, and relaxation time τ (PW model).
These relations and parameters capture certain fundamental characteristics of the local driving
environment and driver population, and have to be calibrated/obtained locally before application
of the corresponding models. The attainment of the parameters are achieved in two ways: direct
measurement and data fitting. The former, when its cost is acceptable, is always preferred if there’s
a a choice of the two.
The interpretations of the parameters, in most cases, are straightforward and intuitive, which
also suggest ways to measure them directly from field data. Among the various speeds, for example,
one can easily measure free flow speed and jam wave speed, but not traffic sound speed c0 . For those
parameters that can be directly measured, to obtain them is a simple matter of data gathering and
processing and we will not elaborate on them here. Rather, we focus on the calibration of those
parameters that have confusing interpretations in literature and are difficult to measure directly.
These include sound speed c0 and relaxation time τ .
Recall that the definition of traffic sound speed is the speed of sound waves minus the speed of

traffic that carries these sound waves. In the LWR model, the sound wave speeds are f∗ (k), and

the traffic speed is v∗ (k), therefore c0 = f∗ (k) − v∗ (k) = kv∗ (k) is variable. That is, sound speed in
the LWR (and Zhang’s model for that matter) is not a fundamental parameter. The PW model,
however, fixes the sound speed c0 as a fundamental parameter and assumes that it is a constant.
The latter assumption is questionable because it is unlikely that drivers respond to stimuli with
the same intensity under free-flow and jam traffic conditions. With that being said, we turn our
attention to the possible ways of measuring c0 . Recall that λ1,2 = v ± c0 in the PW model, which
suggests that if we can measure the speed of the slower wave λ1 and traffic speed v, we can then
compute the sound speed c0 . This can be done with instrumented vehicles on a single lane-highway
where one can measure the acceleration, speed and position of each vehicle in the traffic stream.
Clearly, this is a costly way of obtaining sound speed and is rarely done in practice. In reality, c0 ,
together with another parameter τ , is obtained through data fitting.
Before describing the data fitting procedure, we want to reexamine the interpretation of τ , such
that we have a sense of its range and would know roughly if the value obtained from our data fitting
exercise makes sense. Relaxation in traffic flow refers to the process where non-equilibrium traffic
approaches equilibrium traffic overtime4 . The pace of this process is controlled by relaxation time
τ . Clearly, τ takes the human reaction time as its lower limit, which is about 1 − 1.8 seconds. Its
4
one should not confuse the latter with free-flow traffic, although free-flow traffic is also equilibrium traffic, so is
jam traffic!
5-33
upper limit, in theory, can be infinite. In practice, one never observes congestion that lasts longer
than a day, not to mention infinite. The upper limit of relaxation, one speculates, would be in the
order of a few minutes, the period of a stop-start wave. This speculation tends to be supported by
existing calibration exercises (del Castillo and Benitez 1995, Cremer et al 1993, Kühne 1991, 1984,
Papageorgiou et al 1990) that reported τ values ranging from1.8s to 108s.
The calibration of model parameters through data fitting usually involves the following steps:
1. Data collection: collection of road data in the forms of number of lanes, locations of ramps,
and so forth, and traffic data, in the forms of time series data of flux, occupancy and spot
speeds, at various locations,
2. Numerical approximations of the traffic flow model involved,
3. Calibration: obtain the fundamental diagram for each location from measured data, which in
turn determines parameters such as free-flow speed, jam density, capacity flux, and jam wave
speed, and obtain other parameters in the model through data fitting.
The process of data fitting involves minimizing some pre-defined performance measures. One
common measure is the sum of square errors between model outputs and measured data:

2
P I = γ1 dt (vcal. (d, t) − vmeas.(d, t)) + γ2 dt (kcal. (d, t) − kmeas. (d, t))2 (112)
which is a function of model parameters to be calibrated, e.g.,
P I = P I(c0 , τ ) (113)
for the PW model.

Because of the hyperbolic nature of the continuum traffic flow models, it is crucial to make sure
that the finite difference or finite element approximations are correct and accurate. For this reason
it is advocated that another step be added to the calibration or validation procedure: the step of
checking the finite difference approximation (Zhang 2001). The best way to check the correctness
of a numerical approximation, apart from theoretical considerations, is to run through benchmark
problems, such as Riemann problems. In this way one can rid of transient and boundary conditions
that may hide the inadequacies of the approximation (Zhang 2001).
For the specific examples of calibrating the model parameters, the readers are referred to the
following literature:
• del Castillo and Benitez 1995 (A2 Amsterdam-Utrecht, the Netherlands).
• Kühne and Langbein-Euchner 1993 (A3 Fürth-Erlangen near Nuremberg, Germany)
• Papageorgiou et al 1990 (Boulevard Peripherique, Paris, France), and
• Sailer 1996 (Interstate 35W in Minneapolis, Minnesota, U.S.A.).
5-34
5.3.2 Multilane traffic flow dynamics
Numerical examples of a two-lane ring road to be provided by Panos?
5.3.3 Traffic flow on a ring road with a bottleneck5
In this section we use the finite difference approximations of the LWR and PW models developed
earlier to simulate traffic on a ring road. The length of the ring road is L = 800l = 22.4 km. The
simulation time is T = 500τ = 2500 s = 41.7 min. We partition the road [0, L] into N = 100 cells
and the time interval [0, T ] into K = 500 steps. Hence, the length of each cell is Δx = 0.224 km
and the length of each time step is Δt = 5 s. Since λ∗ ≤ vf = 5l/τ , we find the CFL condition
number
Δt
λ∗ ≤ 0.625 < 1.
Δx
Moreover, we adopt in this simulation the fundamental diagram used in (Kerner and Konhäuser,
1994; Herrman and Kerner, 1998) with the following parameters: the relaxation time τ = 5 s; the
unit length l = 0.028 km; the free flow speed vf = 5.0l/τ = 0.028 km/s = 100.8 km/h; the jam
density of a single lane ρj = 180 veh/km/lane; c0 = 2.48445l/τ = 0.014 km/s = 50.0865 km/h;
The equilibrium speed-density relationship is therefore
⎡ −1 ⎤
ρ
v∗(ρ, a(x)) = 5.0461 ⎣ 1 + exp{[ − 0.25]/0.06} − 3.72 × 10−6 ⎦ l/τ,
a(x)ρj
where a(x) is the number of lanes at location x. The equilibrium functions v∗ (ρ, a(x)) and f∗ (ρ, a(x))
are given in Figure 9.
The first simulation is about the homogeneous LWR model. Here we assume that the ring road
has single lane everywhere; i.e., a(x) = 1, for x ∈ [0, L] , and use a global perturbation as the initial
condition
ρ(x, 0) = ρh + Δρ0 sin 2πx L , x ∈ [0, L], (114)
v(x, 0) = v∗ (ρ(x, 0), 1), x ∈ [0, L],
with ρh = 28 veh/km and Δρ0 = 3 veh/km; and the corresponding initial condition (114) is depicted
in Figure 10.
The results are shown as contour plots in Figure 11, from which we observe that initially wave
interactions are strong but gradually the bulge sharpens from behind and expands from front to
form a so-called N -wave that travels around the ring with a nearly fixed profile.
In the second simulation we created a bottleneck on the ring road with the following lane
configuration:

1, x ∈ [320l, 400l),
a(x) = (115)
2, elsewhere .
5
The results in this section are from unpublished material of Jin and Zhang (2000a,b)
5-35
Figure 9: The Kerner-Konhäuser model of speed-density and flow-density relations
Figure 10: Initial condition (114) with ρh = 28 veh/km and Δρ0 = 3 veh/km
5-36
Figure 11: Solutions of the homogeneous LWR model with initial condition in Figure 10
5-37
As before, we also use a global perturbation as the initial condition
ρ(x, 0) = a(x)(ρh + Δρ0 sin 2πx

L ), x ∈ [0, L], (116)
v(x, 0) = v∗ (ρ(x, 0), a(x)), x ∈ [0, L],
with ρh = 28 veh/km/lane and Δρ0 = 3 veh/km/lane (the corresponding initial condition (116) is
depicted in Figure 12).
Figure 12: Initial condition (116) with ρh = 28 veh/km/lane and Δρ0 = 3 veh/km/lane
The results for this simulation are shown in Figure 13, and are more interesting. We observe
from this figure that at first flow increases in the bottleneck to make the bottleneck saturated, then
a queue forms upstream of the bottleneck, whose tail propagates upstream as a shock. In the same
time, traffic emerges from the bottleneck accelerates in an expansion wave. After a while, all the
commotion settles and an equilibrium state is reached, where a stationary queue forms upstream
of the bottleneck, whose in/out flow rate equals the capacity of the bottleneck.
The third and fourth simulation runs are for the PW model, where we used the same initial con-
ditions for density as in the first and second simulations, respectively, but different initial conditions
for traffic speeds. These initial conditions (I.C.) are
5-38
Figure 13: Solutions of the inhomogeneous LWR model with initial condition (116)
5-39
I.C. for the third simulation
a(x) = 1, x ∈ [0, L]
ρ(x, 0) = ρh + Δρ0 sin 2πxL , x ∈ [0, L], (117)
2πx
v(x, 0) = v∗ (ρh , 1) + Δv0 sin L , x ∈ [0, L].
I.C. for the fourth simulation

1, x ∈ [320l, 400l)
a(x) =
2, elsewhere
(118)
ρ(x, 0) = a(x)(ρh + Δρ0 sin 2πx
L ), x ∈ [0, L],
2πx
v(x, 0) = v∗(ρh , a(x)) + Δv0 sin L , x ∈ [0, L].
The parameters are ρh = 28 veh/km, Δρ0 = 3 veh/km and Δv0 = 0.002 km/s.
Again we use the same time step and cell size, which yields a CFL number of
Δt
λ2 ≤ 0.9375 < 1
Δx
that ensures numerical stability of our finite difference approximation. The results of these simu-
lation runs are shown in Figure 14 and Figure 15 respectively. Note that the PW model solution
for the homogeneous road is slightly different that the corresponding LWR solution due to non-
equilibrium initial speed, but the PW solution soon (about 10τ ) looks very much like the LWR
solution. This can be seen more clearly from time-slice plots of vehicle density, speed and flow rate
shown in Figure 16. As can been seen from that figure, the solutions are nearly indistinguishable
after t = 140τ . In contrast, the PW solution for the inhomogeneous road, although shows similar
patterns as the the corresponding LWR solution, does not converge to the LWR solution in long
time (see Figure 15 & Figure 17). In long time, both solutions predict the same location of the
tail anb head of the queue, but different discharge rate from the queue—traffic leaving the queue
at capacity flow rate in the LWR solution, but below capacity flow rate in the PW solution (see
Figure 17). This result highlights not only the differences between the two models, but also the
importance and need for careful experimental validations of these models6 .
6
The pikes in density that exceed jam density are possibly caused by traffic being unstable near the tail of the
queue.
5-40
Figure 14: Solutions of the PW model with initial condition (117)
5-41
Figure 15: Solutions of the PW model with initial condition (118)
5-42
Figure 16: Comparison of the LWR model and the PW model on a homogeneous ring road: Solid
line is used for the LWR model, and dashed line for the PW model
5-43
Figure 17: Comparison of the LWR model and the PW model on an inhomogeneous ring road:
Solid line is used for the LWR model, and dashed line for the PW model
5-44
References
Bui, D., Nelson, P. & Narasimhan, S. L. (1992). Computational realizations of the entropy condition
in modeling congested traffic flow, Technical Report, Texas Transportation Research Institute.
Courant, R. & Friedrichs, K. O. ( 1948) , Supersonic flow and shock waves, Interscience Publishers,
Inc., New York.
Courant, R. and Hilbert, D. (1962). Methods of mathematical physics, Interscience Publishers, Inc.,
New York.
Cremer, M., F. Meiner, and S. Schrieber (1993). On Predictive Schemes in Dynamic Rerouting
Strategies. In C. Daganzo (ed). Theory of Transportation and Traffic Flow, pp. 407-462.
Daganzo, C. F. (1994). The cell transmission model: a dynamic representation of highway traffic
consistent with hydrodynamic theory, Transpn. Res. B, Vol. 28B, No.4, 269-287.
Daganzo, C. F. (1995a). Requiem for second-order approximations of traffic flow. Transpn. Res.
-B, vol. 29B, No.4, 277-286.
Daganzo, C. F. (1995b). The cell transmission model, Part II: network traffic. Transpn. Res. -B,
vol. 29B, No. 2, 79-93.
Daganzo, C. F. (1995c) , ‘A finite difference approximation of the kinematic wave model of traffic
flow’, Transportation Research, B 29B(4), 261–276.
Daganzo, C. F. (1997). Fundamentals of Transportation and Traffic Operations. Pergamon, Oxford.
Daganzo, C. F. (1999a). A Behavioral Theory of Multi-Lane Traffic Flow Part I: Long Homogeneous
Freeway Sections, ITS Working Paper, UCB-ITS-RR-99-5.
Daganzo, C. F. (1999b). A Behavioral Theory of Multi-Lane Traffic Flow Part II: Merges and the
Onset of Congestion”, ITS Working Paper, UCB-ITS-RR-99-6.
del Castillo, J. M. Pintado, P. and F. G. Benitez (1994). The reaction time of drivers and the
stability of traffic flow. Transpn. Res.-B. Vol. 28B, No. 1, 35-60.
del Castillo, J. M. and F. G. Benitez (1995). On functional form of the speed-density relationship—I:
general theory, II: empirical investigation. Transp. Res. 29B, 373-406.
Dressler, R. F. (1949). Mathematical Solution of the Problem of Roll-Waves in Inclined Open

Channels. Commun. Pure and Appl. Math, 2, 149-194.
Edie, L. C. and E. Bavarez (1967). Generation and propagation of stop-start traffic waves. In L.
C. Edie, R. Herman and R. Rothery, eds., Vehicular Traffic Science, Elsevier, 26-37.
5-45
Gazis, D. C., R. Herman, and G. H. Weiss (1962). Density Oscillations Between Lanes of a Multilane
Highway. Oper. Res.,10, 658-667.
Ferrari, P. (1989). The Effect of Driver Behaviour on Motorway Reliability. Transportation Re-
search, 23B, 139-150.
Herrmann, M. and B.S. Kerner (1998). Local cluster effect in different traffic flow models, {em
Physica A 255, 163-198.
Jin, W. L. and H. M. Zhang (2000a). The inhomogeneous kinematic wave traffic flow model as a
resonant nonlinear system, submitted to Transpn. Sci..
Jin, W. L. and H. M. Zhang (2000b). Numerical studies on the PW model, Working Paper,
Institute of Transportation Studies, University of California at Davis.
Kerner, B. S. and Konhaüser, P. (1993). Cluster effect in initially homogeneous trafficflow, Physical
Review E 48(4), 2335–2338.
Kerner,B.S. and P. Konhäuser (1994). Structure and parameters of clusters in traffic flow, Physical
Review E Volume 50, Number 1, 54-83.
Kerner, B. S., Konhaüser, P. and M. Shilke (1996). A new approach to problems of traffic flow
theory. in: Proceedings of the 13th Int. Symp. on Transportation and Traffic Theory, J. B.
Lesort, editor. 79-102, Pergamon, Oxford, U.K..
Kerner, B. S. and H. Rehborn (1999). Theory of congested traffic flow: self organization without
bottlenecks. Proceedings of the 14th Int. Symp. on Transportation and Traffic Theory, Cedar,
A., editor. 147-171, Pergamon, New York, NY.
Koshi, M., Iwasaki, M. and I. Ohkura (1983). Some findings and an overview on vehicular flow
characteristics. Proceedings of the 8th Int. Symp. on Transportation and Traffic Theory, Hurdle,
V., Hauer, E. and G. Stuart, editors. 403-451, University of Toronto Press, Toronto, Canada.
Kuhne, R. D. (1984). Macroscopic freeway model for dense traffic— stop-start waves and incident
detection, Ninth International Symposium on Transportation and Traffic Theory, VNU Science
Press, 20-42.
Kuhne, R. D. (1987). Freeway Speed Distribution and Acceleration Noise in N. H. Gartner, N. H.

M. Wilson (eds.). Proceedings of the Tenth International Symposium on Transportation and
Traffic Theory, Elsevier, 119-137.
Kuhne, R. D. (1989). Freeway control and incident detection using a stochastic continuum theory of
traffic flow. In Proc. 1st int. conf. on applied advanced technology in transportation engineering,
pages 287–292, San Diego, CA.
5-46
Kühne, R. D. and Beckschulte, R. (1993). Non-linearity stochastics of unstable traffic flow, in C. F.

Daganzo, ed., ‘Transportation and Traffic Theory’, Elsevier Science Publishers., pp. 367–386.
Kühne, R. D. and K. Langbein-Euchner (1995). Parameter Validation. Report as an Order

of Daimler-Benz Research, Stuttgart. Available by Steierwald Schnharting und Partner,
Stuttgart.
Lax, P. D. (1972). Hyperbolic systems of conservations laws and the mathematical theory of shock
waves, Society for Industrial and Applied Mathematics, Philadelphia, Pennsylvania.
Lebacque, J. P. (1996). The Godunov scheme and what it means for first order traffic flow models.
Proceedings of the 13th Int. Symp. on Transportation and Traffic Theory, J-P Lesort, editor.
647-677, Pergamon, New York, NY.
Lebacque, J. P. (1999). Macroscopic traffic flow models: a question of order. Proceedings of the
14th Int. Symp. on Transportation and Traffic Theory, Cedar, A., editor. 147-171, Pergamon,
New York, NY.
Leo, C. J., and R. L. Pretty (1992). Numerical simulations of macroscopic continuum traffic models.
Transpn. Res.-B. 26B, No.3, 207-220.
Leutzbach, W. (1985). Introduction to the Theory of Traffic Flow. Springer Pub., 184-193.
Leutzbach, W. (1991) Measurements of Mean Speed Time Series Autobahn A5 near Karlsruhe,
Germany. Institute of Transport Studies, University of Karlsruhe, Germany.
LeVeque, R. (1992) Numerical methods for conservation laws, Birkhäuser Verlag.
Lighthill, M. J. and Whitham, G. B. (1955). On kinematic waves: II. a theory of traffic flow on
long crowded roads. In Proc. Royal Society, volume 229(1178) of A, 317–345.
Michalopoulos, P. G., Beskos, D. E. & Lin, J. K. (1984) , ‘Analysis of interrupted traffic flow by
finite difference methods’, Transportation Research, B 18B, 377–396.
Michalopoulos, P. G., D. E. Beskos, and Y. Yamauchi, (1984). Multilane Traffic Flow Dynamics:
Some Macroscopic Considerations. Transportation Research, 18B, No. 4/5, pp. 377-395.
Michalopoulos, P. G., Jaw Kuan Lin, and D. E. Beskos, (1987). Integrated Modelling and Numerical
Treatment of Freeway Flow. Appl. Mathem. Model, Vol. 11, No. 401, 447-458.
Michalopoulos, P. G., E. Kwon, and E. G. Khang, (1991). Enhancements and Field Testing of a
Dynamic Simulation Program. Transportation Research Record, 1320, pp. 203-215.
Newell, G. F. (1961). Nonlinear Effects in the Dynamics of Car Following, Operations Research,
Vol 9, pp.209-229.
5-47
Newell, G. F. (1965). Instability in dense highway traffic, a review. In Almond, P., editor, Proceed-
ings of the Second International Symposium on the Theory of Traffic Flow, pages 73–85.
Newell, G. F. (1993). A simplified theory of kinematic waves in highway traffic, I general theory, II
queuing at freeway bottlenecks, III multi-destination flows. Transpn. Res. B, Vol. 27, 281-313.
Papageorgiou, M., Blosseville, J., and Hadj-Salem, H.(1990) Modeling and Real-Time Control
of Traffic Flow on The Southern Part of Boulevard Peripherique in Paris: Part I: Modeling,
Transportation Research, A, 24 (5) 345-359.
Papageorgiou, M. (1998). Some remarks on macroscopic flow modeling. Transportation Research,

A, 32 (5) 323-329.
Payne, H. J. (1971). Models of freeway traffic and control. In Bekey, G. A., editor, Mathematical
Models of Public Systems, volume 1 of Simulation Councils Proc. Ser., 51–60.
Payne, H. J. (1979). FREFLO: A Macroscopic Simulation Model of Freeway Traffic. Transportation

Research Record 722, pp. 68-77.
Richards, P. I. (1956). Shock waves on the highway. Operations Research, 4:42–51.
Sailer, H. (1996). Fuid Dynamical Modelling of Traffic Flow on Highways–Physical Basis and
Numerical Examples; Dissertation. (submitted for a Diploma) at the Institute for Theoretical
Physics, University Innsbruck.
Treiterer, J. and J. A. Myers (1974). The hysteresis phenomena in traffic flow. Proceedings of the
sixth Int. Symp. on Transportation and Traffic Theory, D. J. Buckley, editor. 13-38.
Verweij, H. D. (1985). Filewaarschuwing en Verkeer-safwikkeling. Ministry of Traffic and Water-

Ways, Delft, The Netherlands.
Whitham, G. B. (1974). Linear and nonlinear waves. John Wiley & Sons, New York.
Zhang, H. M. (1998). A theory of nonequilibrium traffic flow. Transportation Research, B,

32(7):485–498.
Zhang, H. M. (1999). Analyses of the stability and wave properties of a new continuum traffic
theory. Transportation Research, B, 33 (6) 399-415.
Zhang, H. M. and T. Wu (1999) Numerical simulation and analysis of traffic flow. Transportation
Zhang, H. M. (2000a). Structural properties of solutions arising from a nonequilibrium traffic flow
theory. Transportation Research, B. 34, 583-603 .
Zhang, H. M. (2000b). A finite difference approximation of a non-equilibrium traffic flow model.

Transportation Research, B. (in press).
5-48
Zhang, H. M. (2000c). Anisotropic property revisited—when is it violated in traffic flow? Working

paper, Institute of Transportation Studies, University of California at Davis (accepted for
publication in Trans. Res., -B).
Zhang, H. M. (2000d). Phase transitions in a non-equilibrium traffic theory, ” ASCE Journal of

Transportation Engineering. Vol. 126, No. 1, 1-12.
Zhang, H. M. and T. Kim (2000). Effects of relaxation and anticipation on Riemann solutions of
the PW model-a numerical investigation. Transportation Research Record (in press).
Zhang, H. M. (2001). New perspectives on continuum traffic flow models. Special issue on traffic
flow theory, H. M. Zhang, ed., Journal of Networks and Spatial Economics, Vol. 1, Issue 1.
2001.
5-49

MACROSCOPIC FLOW MODELS
BY JAMES C. WILLIAMS9
9
Associate Professor, Department of Civil Engineering, University of Texas at Arlington, Box 19308,
Arlington, TX 76019-0308.
Note to reader: The symbols used in Chapter 6 are the same as those used in the original sources. Therefore, the reader is cautioned that
the same symbol may be used for different quantities in different sections of this chapter. The symbol definitions below include the sections
in which the symbols are used if the particular symbol definition changes within the chapter or is a definition particular to this chapter.
In each case, the symbols are defined as they are introduced within the text of the chapter. Symbol units are given only where they help
define the quantity; in most cases, the units may be in either English or metric units as necessary to be consistent with other units in a
relation.
A = area of town (Section 6.2.1)

c = capacity (vehicles per unit time per unit width of road) (Section 6.2.1)
D = delay per intersection (Section 6.2.2)
f = fraction of area devoted to major roads (Section 6.1.1)
f = fraction of area devoted to roads (Section 6.2.1)
f = number of signalized intersections per mile (Section 6.2.2)
fr = fraction of moving vehicles in a designated network (Section 6.3)
fs = fraction of stopped vehicles in a designated network (Section 6.3)
fs,min = minimum fraction of vehicles stopped in a network (Section 6.4)
I = total distance traveled per unit area, or traffic intensity (pcu/hour/km) (Sections 6.1.1 and 6.2.3)
J = fraction of roadways used for traffic movement (Section 6.2.1)
K = average network concentration (ratio of the number of vehicles in a network and the network length, Section 6.4)
Kj = jam network concentration (Section 6.4)
N = number of vehicles per unit time that can enter the CBD (Section 6.2.1)
n = quality of traffic indicator (two-fluid model parameter, Section 6.3)
Q = capacity (pcu/hr) (Section 6.2.2)
Q = average network flow, weighted average over all links in a designated network (Section 6.4)
q = average flow (pcu/hr)
R = road density, i.e., length or area of roads per unit area (Section 6.2.3)
r = distance from CBD
T = average travel time per unit distance, averaged over all vehicles in a designated network (Section 6.3)
Tm = average minimum trip time per unit distance (two-fluid model parameter, Section 6.3)
Tr = average moving (running) time per unit distance, averaged over all vehicles in a designated network (Section 6.3)
Ts = average stopped time per unit distance, averaged over all vehicles in a designated network (Section 6.3)
V = average network speed, averaged over all vehicles in a designated network (Section 6.4)
Vf = network free flow speed (Section 6.4)
Vm = average maximum running speed (Section 6.2.3)
Vr = average speed of moving (running) vehicles, averaged over all in a designated network (Section 6.3)
v = average speed
v = weighted space mean speed (Section 6.2.3)
vr = average running speed, i.e., average speed while moving (Section 6.2.2)
w = average street width
= Zahavi’s network parameter (Section 6.2.3)
= g/c time, i.e. ration of effective green to cycle length
6.
MACROSCOPIC FLOW MODELS
Mobility within an urban area is a major component of that area's level provides this measurement in terms of the three basic
quality of life and an important issue facing many cities as they variables of traffic flow: speed, flow (or volume), and
grow and their transportation facilities become congested. There concentration. These three variables, appropriately defined, can
is no shortage of techniques to improve traffic flow, ranging also be used to describe traffic at the network level. This
from traffic signal timing optimization (with elaborate, description must be one that can overcome the intractabilities of
computer-based routines as well as simpler, manual, heuristic existing flow theories when network component interactions are
methods) to minor physical changes, such as adding a lane by the taken into account.
elimination of parking. However, the difficulty lies in evaluating
the effectiveness of these techniques. A number of methods The work in this chapter views traffic in a network from a
currently in use, reflecting progress in traffic flow theory and macroscopic point of view. Microscopic analyses run into two
practice of the last thirty years, can effectively evaluate changes major difficulties when applied to a street network:
in the performance of an intersection or an arterial. But a
dilemma is created when these individual components, 1) Each street block (link) and intersection are modeled
connected to form the traffic network, are dealt with collectively. individually. A proper accounting of the interactions
between adjacent network components (particularly in the
The need, then, is for a consistent, reliable means to evaluate case of closely spaced traffic signals) quickly leads to
traffic performance in a network under various traffic and intractable problems.
geometric configurations. The development of such
performance models extends traffic flow theory into the network 2) Since the analysis is performed for each network
level and provides traffic engineers with a means to evaluate component, it is difficult to summarize the results in a
system-wide control strategies in urban areas. In addition, the meaningful fashion so that the overall network performance
quality of service provided to the motorists could be monitored can be evaluated.
to evaluate a city's ability to manage growth. For instance,
network performance models could also be used by a state Simulation can be used to resolve the first difficulty, but the
agency to compare traffic conditions between cities in order to second remains; traffic simulation is discussed in Chapter 10.
more equitably allocate funds for transportation system
improvements. The Highway Capacity Manual (Transportation Research Board
1994) is the basic reference used to evaluate the quality of traffic
The performance of a traffic system is the response of that service, yet does not address the problem at the network level.
system to given travel demand levels. The traffic system consists While some material is devoted to assessing the level of service
of the network topology (street width and configuration) and the on arterials, it is largely a summation of effects at individual
traffic control system (e.g., traffic signals, designation of one- intersections. Several travel time models, beginning with the
and two-way streets, and lane configuration). The number of travel time contour map, are briefly reviewed in the next section,
trips between origin and destination points, along with the followed by a description of general network models in Section
desired arrival and/or departure times comprise the travel 6.2. The two-fluid model of town traffic, also a general network
demand levels. The system response, i.e., the resulting flow model, is discussed separately in Section 6.3 due to the extent of
pattern, can be measured in terms of the level of service the model's development through analytical, field, and simulation
provided to the motorists. Traffic flow theory at the intersection studies. Extensions of the two-fluid model into general network
and arterial models are examined in Section 6.4, and the chapter references
are in the final section.
6.1 Travel Time Models

Travel time contour maps provide an overview of how well a dispatched away from a specified location in the network, and
street network is operating at a specific time. Vehicles can be each vehicle's time and position noted at desired intervals.

$&526&23,& /2: 2'(/6
Contours of equal travel time can be established, providing case, general model forms providing the best fit to the data were
information on the average travel times and mean speeds over selected. Traffic intensity (I, defined as the total distance
the network. However, the information is limited in that the traveled per unit area, with units of pcu/hour/km) tends to
travel times are related to a single point, and the study would decrease with increasing distance from the CBD,
likely have to be repeated for other locations. Also, substantial
resources are required to establish statistical significance. Most
importantly, though, is that it is difficult to capture network I
A exp r/a (6.1)
performance with only one variable (travel time or speed in this
case), as the network can be offering quite different levels of
service at the same speed. where r is the distance from the CBD, and A and a are
parameters. Each of the four cities had unique values of A and
This type of model has be generalized by several authors to a, while A was also found to vary between peak and off-peak
estimate average network travel times (per unit distance) or periods. The data from the four cities is shown in Figure 6.1.
speeds as a function of the distance from the central business A similar relation was found between the fraction of the area
district (CBD) of a city, unlike travel time contour maps which which is major road (f) and the distance from the CBD,
consider only travel times away from a specific point.
f
B exp r/b , (6.2)
6.1.1 General Traffic Characteristics
as a Function of the Distance
where b and B are parameters for each town. Traffic intensity
from the CBD and fraction of area which is major road were found to be
linearly related, as was average speed and distance from the
Vaughan, Ioannou, and Phylactou (1972) hypothesized several CBD. Since only traffic on major streets is considered, these
general models using data from four cities in England. In each
Figure 6.1
Total Vehicle Distance Traveled Per Unit Area on Major Roads as a
Function of the Distance from the Town Center (Vaughan et al. 1972).

$&526&23,& /2: 2'(/6
results are somewhat arbitrary, depending on the streets selected had been fitted to data from a single city (Angel and Hyman
as major. 1970). The negative exponential asymptotically approaches
some maximum average speed.
6.1.2 Average Speed as a Function of The fifth function, suggested by Lyman and Everall (1971),
Distance from the CBD 1b 2r 2
v
(6.7)
acb 2r 2
Branston (1974) investigated five functions relating average
speed (v) to the distance from the CBD (r) using data collected
by the Road Research Laboratory (RRL) in 1963 for six cities in
also suggested a finite maximum average speed at the city
England. The data was fitted to each function using least-
outskirts. It had originally be applied to data for radial and ring
squares regression for each city separately and for the aggregated
roads separately, but was used for all roads here.
data from all six cities combined. City centers were defined as
the point where the radial streets intersected, and the journey
Two of the functions were quickly discarded: The linear model
speed in the CBD was that found within 0.3 km of the selected
(Equation 6.5) overestimated the average speed in the CBDs by
center. Average speed for each route section was found by
3 to 4 km/h, reflecting an inability to predict the rapid rise in
dividing the section length by the actual travel time
average speed with increasing distance from the city center. The
(miles/minute). The five selected functions are described below,
modified power curve (Equation 6.4) estimated negative speeds
where a, b, and c are constants estimated for the data. A power
in the city centers for two of the cities, and a zero speed for the
curve,
aggregated data. While obtaining the second smallest sum of
v
arb (6.3) squares (negative exponential, Equation 6.6, had the smallest),
the original aim of using this model (to avoid the estimation of
a zero journey speed in the city center) was not achieved.
was drawn from Wardrop's work (1969), but predicts a zero
The fitted curves for the remaining three functions (negative
speed in the city center (at r = 0). Accordingly, Branston also
exponential, Equation 6.6; power curve, Equation 6.3; and
fitted a more general form,
Lyman and Everall, Equation 6.7) are shown for the data from
v
c ar b , (6.4) Nottingham in Figure 6.2. All three functions realistically
predict a leveling off of average speed at the city outskirts, but
only the Lyman-Everall function indicates a leveling off in the
CBD. However, the power curve showed an overall better fit
where c represents the speed at the city center.
than the Lyman-Everall model, and was preferred.
Earlier work by Beimborn (1970) suggested a strictly linear
While the negative exponential function showed a somewhat
form, up to some maximum speed at the city edge, which was
better fit than the power curve, it was also rejected because of its
defined as the point where the average speed reached its
greater complexity in estimation (a feature shared with the
maximum (i.e., stopped increasing with increasing distance from
Lyman-Everall function). Truncating the power function at
the center). None of the cities in Branston's data set had a clear
measured downtown speeds was suggested to overcome its
maximum limit to average speed, so a strict linear function alone
drawback of estimating zero speeds in the city center. The
was tested:
complete data set for Nottingham is shown in Figure 6.3,
v
a br . (6.5) showing the fitted power function and the truncation at r = 0.3
km.
A negative exponential function,

v
a b e cr , (6.6)

$&526&23,& /2: 2'(/6
Figure 6.2
Grouped Data for Nottingham Showing Fitted a) Power Curve,
b) Negative Exponential Curve, and c) Lyman-Everall Curve
(Branston 1974, Portions of Figures 1A, 1B, and 1C).
Figure 6.3
Complete Data Plot for Nottingham; Power Curve
Fitted to the Grouped Data (Branston 1974, Figure 3).

$&526&23,& /2: 2'(/6
If the data is broken down by individual radial routes, as shown Hutchinson (1974) used RRL data collected in 1967 from eight
in Figure 6.4, the relation between speed and distance from the cities in England to reexamine Equations 6.3 and 6.6 (power
city center is stronger than when the aggregated data is curve and negative exponential) with an eye towards simplifying
examined. them.
Figure 6.4
Data from Individual Radial Routes in Nottingham,
Best Fit Curve for Each Route is Shown (Branston 1974, Figure 4).

$&526&23,& /2: 2'(/6
The exponents of the power functions fitted by Branston (1974) city. Assuming that any speed between 50 and 75 km/h would
fell in the range 0.27 to 0.36, suggesting the following make little difference, Hutchinson selected 60 km/h, and
simplification
v
60 a e r/R . (6.9)
v
kr 1/3
(6.8)
Hutchinson found that this model raised the sum of squares by

When fitted to Branston's data, there was an average of 18 30 percent (on the average) over the general form used by
percent increase in the sum of squares. The other parameter, k, Branston. R was found to be strongly correlated with the city
was found to be significantly correlated with the city population, population, as well as showing different averages with peak and
with different values for peak and off-peak conditions. The off-peak conditions, while a was correlated with neither the city
parameter k was found to increase with increasing population, population nor the peak vs. off-peak conditions. The difference
and was 9 percent smaller in the peak than in the off-peak. in the Rs between peak and off-peak conditions (30 percent
higher during peaks) implies that low speeds spread out over
In considering the negative exponential model (Equation 6.6), more of the network during the peak, but that conditions in the
Hutchinson reasoned that average speed becomes less city center are not significantly different. Hutchinson (1974)
characteristic of a city with increasing r, and, as such, it would used RRL data collected in 1967 from eight cities in England to
be reasonable to select a single maximum limit for v for every reexamine Equations 6.3 and 6.6 (power curve and negative
exponential) with an eye towards simplifying them.
6.2 General Network Models

A number of models incorporating performance measures other where is a constant. General relationships between f and
than speed have been proposed. Early work by Wardrop and (N/cA) for three general network types (Smeed 1965) are
Smeed (Wardrop 1952; Smeed 1968) dealt largely with the shown in Figure 6.5. Smeed estimated a value of c (capacity per
development of macroscopic models for arterials, which were unit width of road) by using one of Wardrop's speed-flow
later extended to general network models. equations for central London (Smeed and Wardrop 1964),
q
2440 0.220 v 3 , (6.11)
6.2.1 Network Capacity
Smeed (1966) considered the number of vehicles which can where v is the speed in kilometers/hour, and q the average flow
"usefully" enter the central area of a city, and defined N as the in pcus/hour, and divided by the average road width, 12.6
number of vehicles per unit time that can enter the city center. meters,
In general, N depends on the general design of the road network,
c
58.2 0.00524 v 3 . (6.12)
width of roads, type of intersection control, distribution of
destinations, and vehicle mix. The principle variables for towns
with similar networks, shapes, types of control, and vehicles are:
A, the area of the town; f, the fraction of area devoted to roads; A different speed-flow relation which provided a better fit for
and c, the capacity, expressed in vehicles per unit time per unit speeds below 16 km/h resulted in c = 68 -0.13 v2 (Smeed 1963).
width of road (assumed to be the same for all roads). These are
related as follows: Equation 6.12 is shown in Figure 6.6 for radial-arc, radial, and
ring type networks for speeds of 16 and 32 km/h. Data from
N
f c A, (6.10) several cities, also plotted in Figure 6.6, suggests that c=30,

0$&526&23,& )/2: 02'(/6
Note: (e=excluding area of ring road, I=including area of ring road)
Figure 6.5
Theoretical Capacity of Urban Street Systems (Smeed 1966, Figure 2).
Figure 6.6
Vehicles Entering the CBDs of Towns Compared with the Corresponding
Theoretical Capacities of the Road Systems (Smeed 1966, Figure 4).

0$&526&23,& )/2: 02'(/6
and using the peak period speed of 16 km/h in central London, 6.2.2 Speed and Flow Relations
Equation 6.10 becomes
Thomson (1967b) used data from central London to develop a
N
33 0.003 v 3 f A , (6.13)
linear speed-flow model. The data had been collected once
every two years over a 14-year period by the RRL and the
Greater London Council. The data consisted of a network-wide
where v is in miles/hour and A in square feet. It should be noted average speed and flow each year it was collected. The average
that f represents the fraction of total area usefully devoted to speed was found by vehicles circulating through central London
roads. An alternate formulation (Smeed 1968) is on predetermined routes. Average flows were found by first
converting measured link flows into equivalent passenger
N
33 0.003 v 3 J f A (6.14)
carunits, then averaging the link flows weighted by their
respective link lengths. Two data points (each consisting of an
average speed and flow) were found for each of the eight years
where f is the fraction of area actually devoted to roads, while J the data was collected: peak and off-peak.
is the fraction of roadways used for traffic movement. J was
found to range between 0.22 and 0.46 in several cities in Plotting the two points for each year, Figure 6.7, resulted in a
England. The large fraction of unused roadway is mostly due to series of negatively sloped trends. Also, the speed-flow capacity
the uneven distribution of traffic on all streets. The number of (defined as the flow that can be moved at a given speed)
vehicles which can circulate in a town depends strongly on their gradually increased over the years, likely due to geometric and
average speed, and is directly proportional to the area of usable traffic control improvements and "more efficient vehicles." This
roadway. For a given area devoted to roads, the larger the indicated that the speed-flow curve had been gradually changing,
central city, the smaller the number of vehicles which can indicating that each year's speed and flow fell on different curves.
circulate in the network, suggesting that a widely dispersed town Two data points were inadequate to determine the shape of the
is not necessarily the most economical design. curve, so all sixteen data points were used by accounting for the
Figure 6.7
Speeds and Flows in Central London, 1952-1966,
Peak and Off-Peak (Thomson 1967b, Figure 11)

0$&526&23,& )/2: 02'(/6
changing capacity of the network, and scaling each year's flow The equation implies a free-flow speed of about 48.3 km/h
measurement to a selected base year. Using linear regression, however, there were no flows less than 2200 pcu/hour in the
the following equation was found: historical data.
v
30.2 0.0086 q (6.15)
Thomson used data collected on several subsequent Sundays
(Thomson 1967a) to get low flow data points. These are
reflected in the trend shown in Figure 6.9.Also shown is a curve
where v is the average speed in kilometers/hour and q is the developed by Smeed and Wardrop using data from a single year
average flow in pcu/hour. This relation is plotted in Figure 6.8. only.
Note: Scaled to 1964 equivalent flows.

Figure 6.8
Speeds and Scaled Flows, 1952-1966 (Thomson 1967b, Figure 2).
Figure 6.9
Estimated Speed-Flow Relations in Central London
(Main Road Network) (Thomson 1967b, Figure 4).

$&526&23,& /2: 2'(/6
The selected area of central London could be broken into inner

1
and outer zones, distinguished principally by traffic signal
1 fd , (6.18)
densities, respectively 7.5 and 3.6 traffic signals per route-mile. v vr
Speed and flow conditions were found to be significantly
different between the zones, as shown in Figure 6.10, and for the
inner zone, where v is the average speed in mi/h, vr, the running speed in
mi/h, d the delay per intersection in hours, and f the number of
v
24.3 0.0075 q , (6.16)
signalized intersections per mile. Assuming vr = a(1-q/Q) and
d = b/(1-q/s), where q is the flow in pcu/hr, Q is the capacity
in pcu/hr, is the g/c time, and s is the saturation flow in pcu/hr,
and for the outer zone, and combining into Equation 6.18,
v
34.0 0.0092 q . (6.17) 1 1

fb (6.19)
v a (1 q/Q) 1 q/s
Wardrop (1968) directly incorporated average street width and

average signal spacing into a relation between average speed and Using an expression for running speed found for central London
flow, where the average speed includes the stopped time. In (Smeed and Wardrop 1964; RRL 1965),
order to obtain average speeds, the delay at signalized
0.70q 430
intersections must be considered along with the running speed vr
31 (6.20)
between the controlled intersections, where running speed is 3w
defined as the average speed while moving. Since speed is the
inverse of travel time, this relation can be expressed as:
Figure 6.10
Speed-Flow Relations in Inner and Outer Zones of Central Area
(Thomson 1967a, Figure 5).

$&526&23,& /2: 2'(/6
where w is the average roadway width in feet, and an average For the delay term, five controlled intersections per mile and a
street width of 42 feet (in central London), Equation 6.20 g/c of 0.45 were found for central London. Additionally, the
becomes vr = 28 - 0.0056 q, or 24 mi/h, whichever is less. The intersection capacity was assumed to be proportional to the
coefficient of q was modified to 0.0058 to better fit the observed average stop line width, given that it is more than 5 meters wide
running speed. (RRL 1965), which was assumed to be proportional to the
roadway width. The general form for the delay equation (second
Using observed values of 0.038 hours/mile stopped time, 2180 term of Equation 6.21) is
pcu/hr flow, and 2610 pcu/hr capacity, the numerator of the
second term of Equation 6.19 (fb) was found to be 0.0057.
fb
Substituting the observed values into Equation 6.19, fd
(6.23)
1 q /k w
1 1 0.0057

.
v 28 0.0058 q q
1
2610 where k is a constant. For central London, w = 42, = 0.45,
and kw = Q = 2770, thus k = 147, yielding
Simplifying,
fb
fd
(6.24)
1 q /147 w
1 1 1

(6.21)
v 28 0.0058 q 197 0.0775 q
Given that f = 5 signals/mile and fb = 0.00507 for central
London, b = 0.00101, yielding
Revising the capacity to 2770 pcu/hour (to reflect 1966 data),
f
thus changing the coefficient of q in the second term of Equation fd
. (6.25)
6.21 to 0.071, this equation provided a better fit than Thomson's 1000 6.8 q/ w
linear relation (Thomson 1967b) and recognizes the known
information on the ultimate capacity of the intersections.
Combining, then, for the general equation for average speed:
Generalizing this equation for urban areas other than London,
and knowing that the average street width in central London was
1 1 f
12.6 meters, the running speed can be written
.
v 140 q
31 0.0244 1000 6.8 q (6.26)
w w w
430
vr
31 aq
3w w
140 aq

31 . The sensitivity of Equation 6.26 to flow, average street width,
w w number of signalized intersections per mile, and the fraction of
green time are shown in Figures 6.11, 6.12, and 6.13. By
calibrating this relation on geometric and traffic control features
Since a/w = 0.0058 when w = 42 by Equation 6.21, a=0.0244, in the network, Wardrop extended the usefulness of earlier speed
then flow relations. While fitting nicely for central London, the
applicability of this relation to other cities in its generalized
140
vr
31 0.0244 q . (6.22) format (Equation 6.26) is not shown, due to a lack of available
w w data.

$&526&23,& /2: 2'(/6
Figure 6.11
Effect of Roadway Width on Relation Between Average (Journey)
Speed and Flow in Typical Case (Wardrop 1968, Figure 5).
Figure 6.12
Effect of Number of Intersections Per Mile on Relation Between
Average (Journey) Speed and Flow in Typical Case (Wardrop 1968, Figure 6).

$&526&23,& /2: 2'(/6
Figure 6.13
Effect of Capacity of Intersections on Relation Between
Average (Journey) Speed and Flow in Typical Case (Wardrop 1968, Figure 7).
Godfrey (1969) examined the relations between the average -relationship, below, and the two-fluid theory of town traffic.
speed and the concentration (defined as the number of vehicles The two-fluid theory has been developed and applied to a greater
in the network), shown in Figure 6.14, and between average extent than the other models discussed in this section, and is
speed and the vehicle miles traveled in the network in one hour, described in Section 6.3.
shown in Figure 6.15. Floating vehicles on circuits within the
network were used to estimate average speed and aerial Zahavi (1972a; 1972b) selected three principal variables, I, the
photographs were used to estimate concentration. traffic intensity (here defined as the distance traveled per unit
area), R, the road density (the length or area of roads per unit
There is a certain concentration that results in the maximum flow area), and v, the weighted space mean speed. Using data from
(or the maximum number of miles traveled, see Figure 6.15), England and the United States, values of I, v, and R were found
which occurs around 10 miles/hour. As traffic builds up past for different regions in different cities. In investigating various
this optimum, average speeds show little deterioration, but there relationships between I and v/R, a linear fit was found between
is excessive queuing to get into the network (either from car the logarithms of the variables:
parking lots within the network or on streets leading into the
designated network). Godfrey also notes that expanding an
intersection to accommodate more traffic will move the queue to I
( v/R)m , (6.27)
another location within the network, unless the bottlenecks
downstream are cleared. where and m are parameters. Trends for London and
Pittsburgh are shown in Figure 6.16. The slope (m) was found
to be close to -1 for all six cities examined, reducing Equation
6.2.3 General Network Models 6.27 to
Incorporating Network Parameters I
R/v , (6.28)
Some models have defined specific parameters which intend to

quantify the quality of traffic service provided to the users in the where is different for each city. Relative values of the
network. Two principal models are discussed in this chapter, the variables were calculated by finding the ratio between observed

$&526&23,& /2: 2'(/6
Figure 6.14
Relationship Between Average (Journey)
Speed and Number of Vehicles on Town
Center Network (Godfrey 1969, Figure 1).
Figure 6.15
Relationship Between Average (Journey) Speed of Vehicles
and Total Vehicle Mileage on Network (Godfrey 1969, Figure 2).

$&526&23,& /2: 2'(/6
Figure 6.16
The -Relationship for the Arterial Networks of London and Pittsburgh,
in Absolute Values (Zahavi 1972a, Figure 1).
values of I and v/R for each sector and the average value for the The two-fluid model also uses parameters to evaluate the level
entire city. The relationship between the relative values is of service in a network and is described in Section 6.3.
shown in Figure 6.17, where the observations for London and
Pittsburgh fall along the same line.
6.2.4 Continuum Models
The physical characteristics of the road network, such as street
widths and intersection density, were found to have a strong Models have been developed which assume an arbitrarily fine
effect on the value of for each zone in a city. Thus, may grid of streets, i.e., infinitely many streets, to circumvent the
serve as a measure of the combined effects of the network errors created on the relatively sparse networks typically used
characteristics and traffic performance, and can possibly be used during the trip or network assignment phase in transportation
as an indicator for the level of service. The map of London is planning (Newell 1980). A basic street pattern is superimposed
shown in Figure 6.18. Zones are shown by the dashed lines, over this continuum of streets to restrict travel to appropriate
with dotted circles indicating zone centroids. Values of were directions. Thus, if a square grid were used, travel on the street
calculated for each zone and contour lines of equal were network would be limited to the two available directions (the x
drawn, showing areas of (relatively) good and poor traffic flow and y directions in a Cartesian plot), but origins and destinations
conditions. (The quality of traffic service improves with could be located anywhere in the network.
increasing .)
Individual street characteristics do not have to be specifically
Unfortunately, Buckley and Wardrop (1980) have shown that modeled, but network-wide travel time averages and capacities
is strongly related to the space mean speed, and Ardekani (per unit area) must be used for traffic on the local streets. Other
(1984), through the use of aerial photographs, has shown that street patterns include radial-ring and other grids (triangular, for
has a high positive correlation with the network concentration. example).

$&526&23,& /2: 2'(/6
Figure 6.17
The -Relationship for the Arterial Networks of London and Pittsburgh,
in Relative Values (Zahavi 1972a, Figure 2).
Figure 6.18
The -Map for London, in Relative Values (Zahavi 1972b, Figure 1).

0$&526&23,& )/2: 02'(/6
While the continuum comprises the local streets, the major (within the constraints provided by the superimposed grid) to the
streets (such as arterials and freeways) are modeled directly. network of major streets.
Thus, the continuum of local streets provides direct access
6.3 Two-Fluid Theory

An important result from Prigogine and Herman's (1971) kinetic
theory of traffic flow is that two distinct flow regimes can be Vr
Vm f r n , (6.29)
shown. These are individual and collective flows and are a
where Vm and n are parameters. Vm is the average maximum
function of the vehicle concentration. When the concentration
running speed, and n is an indicator of the quality of traffic
rises so that the traffic is in the collective flow regime, the flow
service in the network; both are discussed below. The average
pattern becomes largely independent of the will of individual
speed, V, can be defined as Vr fr , and combining with Equation
drivers.
6.29,
Because the kinetic theory deals with multi-lane traffic, the two- V
Vm f rn1 . (6.30)
fluid theory of town traffic was proposed by Herman and
Prigogine (Herman and Prigogine 1979; Herman and Ardekani Since f r + fs = 1, where fs is the fraction of vehicles stopped,
1984) as a description of traffic in the collective flow regime in Equation 6.30 can be rewritten
an urban street network. Vehicles in the traffic stream are
divided into two classes (thus, two fluid): moving and stopped V
Vm (1 f s) n 1 . (6.31)
vehicles. Those in the latter class include vehicles stopped in the Boundary conditions are satisfied with this relation: when fs=0,
traffic stream, i.e., stopped for traffic signals and stop signs,
V=Vm , and when fs=1, V=0.
stopped for vehicles loading and unloading which are blocking
a moving lane, stopped for normal congestion, etc., but excludes
This relation can also be expressed in average travel times rather
those out of the traffic stream (e.g., parked cars).
than average speeds. Note that T represents the average travel
time, Tr the running (moving) time, and Ts the stop time, all per
The two-fluid model provides a macroscopic measure of the
unit distance, and that T=1/V, Tr=1/Vr , and Tm=1/Vm , where Tm
quality of traffic service in a street network which is independent
is the average minimum trip time per unit distance.
of concentration. The model is based on two assumptions:
The second assumption of the two-fluid model relates the
(1) The average running speed in a street network is
fraction of time a test vehicle circulating in a network is stopped
proportional to the fraction of vehicles that are moving,
to the average fraction of vehicles stopped during the same
and
period, or
(2) The fractional stop time of a test vehicle circulating in
a network is equal to the average fraction of the Ts
vehicles stopped during the same period. fs
. (6.32)
T
The variables used in the two-fluid model represent network- This relation has been proven analytically (Ardekani and
wide averages taken over a given period of time. Herman 1987), and represents the ergodic principle embedded
in the model, i.e., that the network conditions can be represented
The first assumption of the two-fluid theory relates the average by a single vehicle appropriately sampling the network.
speed of the moving (running) vehicles, Vr , to the fraction of
moving vehicles, fr , in the following manner: Restating Equation 6.31 in terms of travel time,

0$&526&23,& )/2: 02'(/6
observations of stopped and moving times gathered in each

T
Tm ( 1 f s) (n1) . (6.33) network. The log transform of Equation 6.35,
Incorporating Equation 6.32, 1 n
ln Tr
ln Tm ln T (6.37)
(n1) n1 n1
T
Tm 1 ( Ts /T ) , (6.34)
provides a linear expression for the use of least squares analysis.
realizing that T = Tr + Ts , and solving for Tr ,
1 n Empirical information has been collected with chase cars
Tr
Tm n1 T n1 . (6.35) following randomly selected cars in designated networks. Runs
have been broken into one- or two-mile trips, and the running
The formal two-fluid model formulation, then, is time (Tr) and total trip time (T) for each one- or two-mile trip
1 n from the observations for the parameter estimation. Results tend
Ts
T Tm n1 T n1 . (6.36) to form a nearly linear relationship when trip time is plotted
against stop time (Equation 6.36) as shown in Figure 6.19 for
A number of field studies have borne out the two-fluid model data collected in Austin, Texas. The value of Tm is reflected by
(Herman and Ardekani 1984; Ardekani and Herman 1987; the y-intercept (i.e., T at Ts=0), and n by the slope of the curve.
Ardekani et al. 1985); and have indicated that urban street Data points representing higher concentration levels lie higher
networks can be characterized by the two model parameters, n along the curve.
and Tm . These parameters have been estimated using
Note: Each point represents one test run approximately 1 or 2 miles long.
Figure 6.19
Trip Time vs. Stop Time for the Non-Freeway Street Network of the Austin CBD
(Herman and Ardekani 1984, Figure 3).

0$&526&23,& )/2: 02'(/6
6.3.1 Two-Fluid Parameters congestion, and when congestion is high, vehicles when moving,
travel at a lower speed (or higher running time per unit distance)
The parameter Tm is the average minimum trip time per unit than they do when congestion is low. In fact, field studies have
distance, and it represents the trip time that might be shown that n varies from 0.8 to 3.0, with a smaller value
experienced by an individual vehicle alone in the network with typically indicating better operating conditions in the network.
no stops. This parameter is unlikely to be measured directly, In other words, n is a measure of the resistance of the network to
since a lone vehicle driving though the network very late at night degraded operation with increased demand. Higher values of n
is likely to have to stop at a red traffic signal or a stop sign. indicate networks that degrade faster as demand increases.
Tm , then, is a measure of the uncongested speed, and a higher Because the two-fluid parameters reflect how the network
value would indicate a lower speed, typically resulting in poorer responds to changes in demand, they must be measured and
operation. Tm has been found to range from 1.5 to 3.0 evaluated in a network over the entire range of demand
minutes/mile, with smaller values typically representing better conditions.
operating conditions in the network.
While lower n and Tm values represent, in general, better traffic
As stop time per unit distance ( Ts ) increases for a single value operations in a network, often there is a tradeoff. For example,
of n, the total trip time also increases. Because T=Tr+Ts , the two-fluid trends for four cities are shown in Figure 6.20. In
total trip time must increase at least as fast as the stop time. If comparing Houston (Tm=2.70 min/mile, n=0.80) and Austin
n=0, Tr is constant (by Equation 6.35), and trip time would (Tm=1.78 min/mile, n=1.65), one finds that traffic in Austin
increase at the same rate as the stop time. If n>0, trip time moves at significantly higher average speeds during off-peak
increases at a faster rate than the stop time, meaning that running conditions (lower concentration); at higher concentrations, the
time is also increasing. Intuitively, n must be greater than zero, curves essentially overlap, indicating similar operating
since the usual cause for increased stop time is increased conditions. Thus, despite a higher value of n, traffic conditions
Note: Trip Time vs. Stop Time Two-Fluid Model Trends for CBD Data From the Cities of Austin, Houston, and San Antonio,
Texas, and Matamoros, Mexico.
Figure 6.20
Trip Time vs. Stop Time Two-Fluid Model Trends

$&526&23,& /2: 2'(/6
are better in Austin than Houston, at least at lower being chased imitating the other driver's actions so as to reflect,
concentrations. Different values of the two-fluid parameters are as closely as possible, the fraction of time the other driver spends
found for different city street networks, as was shown above and stopped. The objective is to sample the behavior of the drivers
in Figure 6.21. The identification of specific features which have in the network as well as the commonly used routes in the street
the greatest effect on these parameters has been approached network. The chase car's trip history is then broken into one-
through extensive field studies and computer simulation. mile (typically) segments, and Tr and T calculated for each mile.
The (Tr ,T) observations are then used in the estimation of the
two-fluid parameters.
6.3.2 Two-Fluid Parameters: Influence
of Driver Behavior One important aspect of the chase car study is driver behavior,
both that of the test car driver and the drivers sampled in the
Data for the estimation of the two-fluid parameters is collected network. One study addressed the question of extreme driver
through chase car studies, where the driver is instructed to follow behaviors, and found that a test car driver instructed to drive
a randomly selected vehicle until it either parks or leaves the aggressively established a significantly different two-fluid trend
designated network, after which a nearby vehicle is selected and than one instructed to drive conservatively in the same network
followed. The chase car driver is instructed to follow the vehicle at the same time (Herman et al. 1988).
Note: Trip Time vs. Stop Time Two-Fluid Model Trends for Dallas and Houston, Texas, compared to the trends in Milwaukee,
Wisconsin, and in London and Brussels.
Figure 6.21
Trip Time vs. Stop Time Two-Fluid Model Trends Comparison

$&526&23,& /2: 2'(/6
The two-fluid trends resulting from the these studies in two cities at lower network concentrations, the aggressive driver can take
are shown in Figure 6.22. In both cases, the normal trend was advantage of the less crowded streets and significantly lower his
found through a standard chase car study, conducted at the same trip times.
time as the aggressive and conservative test drivers were in the
network. In both cases, the two-fluid trends established by the As shown in Figure 6.22b, aggressive driving behavior more
aggressive and conservative driver are significantly different. In closely reflects normal driving habits in Austin, suggesting more
Roanoke (Figure 6.22a), the normal trend lies between the aggressive driving overall. Also, all three trends converge at
aggressive and conservative trends, as expected. However, the high demand (concentration) levels, indicating that, perhaps, the
aggressive trend approaches the normal trend at high demand Austin network would suffer congestion to a greater extent than
levels, reflecting the inability of the aggressive driver to reduce Roanoke, reducing all drivers to conservative behavior (at least
his trip and stop times during peak periods. On the other hand, as represented in the two-fluid parameters).
Note: The two-fluid trends for aggressive, normal, and conservative drivers in (a) Roanoke, Virginia, and (b) Austin, Texas
Figure 6.22
Two-Fluid Trends for Aggressive, Normal, and Conservative Drivers
(Herman et al. 1988, Figures 5 and 8).

$&526&23,& /2: 2'(/6
The results of this study reveal the importance of the behavior of approaches with signal progression. Of these, only two features
the chase car driver in standard two-fluid studies. While the (average block length and intersection density) can be
effects on the two-fluid parameters of using two different chase considered fixed, and, as such, not useful in formulating network
car drivers in the same network at the same time has not been improvements. In addition, one feature (average number of
investigated, there is thought to be little difference between two lanes per street), also used in the previous study (Ayadh 1986),
well-trained drivers. To the extent possible, however, the same can typically be increased only by eliminating parking (if
driver has been used in different studies that are directly present), yielding only limited opportunities for improvement of
compared. traffic flow. Data was collected in ten cities; in seven of the
cities, more than one study was conducted as major geometric
changes or revised signal timings were implemented, yielding
6.3.3 Two-Fluid Parameters: Influence nineteen networks for this study. As before, the two-fluid
parameters in each network were estimated from chase car data
of Network Features (Field Studies)
and the network features were determined from maps, field
studies, and local traffic engineers. Regression analysis yielded
Geometric and traffic control features of a street network also
the following models:
play an important role in the quality of service provided by a
network. If relationships between specific features and the two-
fluid parameters can be established, the information could be Tm
3.93 0.0035 X5 0.047 X6 0.433 X10 (6.39)
used to identify specific measures to improve traffic flow and and n
1.73 1.124 X2 0.180 X3 0.0042 X5 0.271 X9
provide a means to compare the relative improvements.
Ayadh (1986) selected seven network features: lane miles per

where X2 is the fraction of one-way streets, X3 the average
square mile, number of intersections per square mile, fraction of
number of lanes per street, X5 the signal density, X6 the average
one-way streets, average signal cycle length, average block
speed limit, X9 the fraction of actuated signals, and X10 the
length, average number of lanes per street, and average block
fraction of approaches with good progression. The R2 for these
length to block width ratio. The area of the street network under
equations, 0.72 and 0.75 (respectively), are lower than those for
consideration is used with the first two variables to allow a direct
Equation 6.38 (both very close to 1), reflecting the larger data
comparison between cities. Data for the seven variables were
size. The only feature in common with the previous model
collected for four cities from maps and in the field. Through a
(Equation 6.38) is the appearance of the fraction of one-way
regression analysis, the following models were selected:
streets in the model for n. Since all features selected can be
Tm
3.59 0.54 C6 and changed through operational practices (signal density can be
n
0.21 2.97 C3 0.22 C7 (6.38) changed by placing signals on flash), the models have potential
practical application. Computer simulation has also been used
to investigate these relationships, and is discussed in Section
6.3.4.
where C3 is the fraction of one-way streets, C6 the average
number of lanes per street, and C7 the average block length to
block width ratio. Of these network features, only one (the
fraction of one-way streets) is relatively inexpensive to 6.3.4 Two-Fluid Parameters: Estimation
implement. One feature, the block length to block width ratio, by Computer Simulation
is a topological feature which would be considered fixed for any
established street network. Computer simulation has many advantages over field data in the
study of network models. Conditions not found in the field can
Ardekani et al. (1992), selected ten network features: average be evaluated and new control strategies can be easily tested. In
block length, fraction of one-way streets, average number of the case of the two-fluid model, the entire vehicle population in
lanes per street, intersection density, signal density, average the network can be used in the estimation of the model
speed limit, average cycle length, fraction of curb miles with parameters, rather than the small sample used in the chase car
parking allowed, fraction of signals actuated, and fraction of studies. TRAF-NETSIM (Mahmassani et al. 1984), a

$&526&23,& /2: 2'(/6
microscopic traffic simulation model, has been used successfully can be simulated by recording the trip history of a single vehicle
with the two-fluid model. for one mile, then randomly selecting another vehicle in the
network. Because the two-fluid model is non-linear
Most of the simulation work to-date has used a generic grid (specifically, Equation 6.35, the log transform of which is used
network in order to isolate the effects of specific network to estimate the parameters), estimations performed at the
features on the two-fluid parameters (FHWA 1993). Typically, network level and at the individual vehicle level result in
the simulated network has been a 5 x 5 intersection grid made up different values of the parameters, and are not directly
entire of two-way streets. Traffic signals are at each intersection comparable. The sampling strategy, which was found to provide
and uniform turning movements are applied throughout. The the best parameter estimates, required a single vehicle
network is closed, i.e., vehicles are not allowed to leave the circulating in the network for at least 15 minutes. However, due
network, thus maintaining constant concentration during the to the wide variance of the estimate (due to the possibility of a
simulation run. The trip histories of all the vehicles circulating relatively small number of "chased" cars dominating the sample
in the network are aggregated to form a single (Tr , T) estimation), the estimate using a single vehicle was often far
observation for use in the two-fluid parameter estimation. A from the parameter estimated at the network level. On the other
series of five to ten runs over a range of network concentrations hand, using 20 vehicles to sample the network resulted in
(nearly zero to 60 or 80 vehicles/lane-mile) are required to estimates much closer to those at the network level. The much
estimate the two-fluid parameters. smaller variance of the estimates made with twenty vehicles,
however, resulted in the estimate being significantly different
Initial simulation runs in the test network showed both T and Ts from the network-level estimate. The implication of this study
increasing with concentration, but Tr remaining nearly constant, is that, while estimates at the network and individual vehicle
indicating a very low value of n (Mahmassani et al. 1984). In levels can not be directly compared, as long as the same
its default condition, NETSIM generates few of the vehicle sampling strategy is used, the resulting two-fluid parameters,
interaction of the type found in most urban street networks, although biased from the "true" value, can be used in making
resulting in flow which is much more idealized than in the field. direct comparisons.
The short-term event feature of NETSIM was used to increase
the inter-vehicular interaction (Williams et al. 1985). With this
feature, NETSIM blocks the right lane of the specified link at 6.3.5 Two-Fluid Parameters:
mid-block; the user specifies the average time for each blockage Influence of Network Features
and the number of blockages per hour, which are stochastically
applied by NETSIM. In effect, this represents a vehicle stopping (Simulation Studies)
for a short time (e.g., a commercial vehicle unloading goods),
blocking the right lane, and requiring vehicles to change lanes to The question in Section 6.3.3, above, regarding the influence of
go around it. The two-fluid parameters (and n in particular) geometric and control features of a network on the two-fluid
were very sensitive to the duration and frequency of the short- parameters was revisited with an extensive simulation study
term events. For example, using an average 45-second event (Bhat 1994). The network features selected were: average
every two minutes, n rose from 0.076 to 0.845 and Tm fell from block length, fraction of one-way streets, average number of
2.238 to 2.135. With the use of the short-term events, the values lanes per street, signals per intersection, average speed limit,
of both parameters were within the ranges found in the field average signal cycle length, fraction of curb miles with parking,
studies. Further simulation studies found both block length and fraction of signalized approaches in progression. A
(here, distance between signalized intersections) and the use of uniform-precision central composite design was selected as the
progression to have significant effects on the two-fluid experimental design, resulting in 164 combination of the eight
parameters (Williams et al. 1985). network variables. The simulated network was increased to 11
by 11 intersections; again, vehicles were not allowed to leave the
Simulation has also provided the means to investigate the use of network, but traffic data was collected only on the interior 9 by
the chase car technique in estimating the two-fluid parameters 9 intersection grid, thus eliminating the edge effects caused by
(Williams et al. 1995). The network-wide averages in a the necessarily different turning movements at the boundaries.
simulation model can be directly computed; and chase car data Ten simulation runs were made for each combination of

$&526&23,& /2: 2'(/6
variables over a range of concentrations from near zero to about NETSIM reflected traffic conditions in San Antonio, NETSIM
35 vehicles/lane-mile. was calibrated with the two-fluid model.
Regression analysis yielded the following models: Turning movement counts used in the development of the new
signal timing plans were available for coding NETSIM.
Tm
1.049 1.453 X2 0.684 X3 0.024 X6 and
(6.40) Simulation runs were made for 31 periods throughout the day,
n
4.468 1.391 X3 0.048 X5 0.042 X6
and the two-fluid parameters were estimated and compared with
those found in the field. By a trial and error process, NETSIM
was calibrated by
where X2 is the fraction of one-way streets, X 3the number of
lanes per street, X5 average speed limit, and X 6 average cycle Increasing the sluggishness of drivers, by increasing
length. The R2 (0.26 and 0.16 for Equation 6.40) was headways during queue discharge at traffic signals
considerably lower than that for the models estimated with data and reducing maximum acceleration,
from field studies (Equation 6.39). Additionally, the only Adding vehicle/driver types to increase the range of
variable in common between Equations 6.39 and 6.40 is the sluggishness represented in the network, and
number of lanes per street in the equation for n. Additional work Reducing the desired speed on all links to 32.2 km/h
is required to clarify these relationships. during peaks and 40.25 km/h otherwise (Denney
1993).
Three measures of effectiveness (MOEs) were used in the

6.3.6 Two-Fluid Model: evaluation: total delay, number of stops, and fuel consumption.
A Practical Application The changes noted for all three MOEs were greater between
calibrated and uncalibrated NETSIM results than between
When the traffic signals in downtown San Antonio were retimed, before and after results. Reported relative improvements were
TRAF-NETSIM was selected to quantify the improvements in also affected. The errors in the reported improvements without
the network. In order to assure that the results reported by calibration ranged from 16 percent to 132 percent (Denney
1994).
6.4 Two-Fluid Model and Traffic Network Flow Models

Computer simulation provides an opportunity to investigate fluid model, to the concentration. In addition, using values of
network-level relationships between the three fundamental flow, speed, and concentration independently computed from the
variables of traffic flow, speed (V), flow (Q), and concentration simulations, the network-level version of the fundamental
(K), defined as average quantities taken over all vehicles in the relation Q=KV was numerically verified (Mahmassani et al.
network over some observation period (Mahmassani et al. 1984; Williams et al. 1987).
1984). While the existence of "nice" relations between these
variables could not be expected, given the complexity of network Three model systems were derived and tested against simulation
interconnections, simulation results indicate relationships similar results (Williams et al. 1987; Mahmassani et al. 1987); each
to those developed for arterials may be appropriate (Mahmassani model system assumed Q=KV and the two-fluid model, and
et al. 1984; Williams et al. 1985). A series of simulation runs, consisted of three relations:
as described in Section 6.3.4, above, was made at concentration
levels between 10 and 100 vehicles/lane-mile. The results are
shown in Figure 6.23, and bear a close resemblance to their
V
f (K) , (6.41)
counterparts for individual road sections. The fourth plot shows
the relation of fs , the fraction of vehicles stopped from the two-

$&526&23,& /2: 2'(/6
Figure 6.23
Simulation Results in a Closed CBD-Type Street Network.
(Williams et al. 1987, Figures 1-4).
A model system is defined by specifying one of the above relationships; the other two can then be analytically derived. (A
relation between Q and V could also be derived.)
Q
g (K) , and (6.42) Model System 1 is based on a postulated relationship between
the average fraction of vehicles stopped and the network
f s
h (K) . (6.43) concentration from the two-fluid theory (Herman and Prigogine

$&526&23,& /2: 2'(/6
1979), later modified to reflect that the minimum fs > 0 then by using Q=KV,
(Ardekani and Herman 1987):
f s
fs,min ( 1 fs,min ) ( K/Kj ) , Q
Vf ( K K 2 /Kj ) . (6.49)
(6.44)
Equations 6.47 through 6.49 were fitted to the simulation data

where fs,min is the minimum fraction of vehicles stopped in a and are shown in Figure 6.25. The difference between the
network, Kj is the jam concentration (at which the network is Method 1 and Method 2 curves in the fs-K plot (Figure 6.25) is
effectively saturated), and is a parameter which reflects the described above. Model System 3 uses a non-linear bell-shaped
quality of service in a network. The other two relations can be function for the V-K model, originally proposed by Drake, et al.,
readily found, first by substituting fs from Equation 6.44 into for arterials (Gerlough and Huber 1975):
Equation 6.31:
V
Vf exp[ ( K/Km )d ] , (6.50)
n1
V
Vm (1 fs,min ) [1 (K/Kj )]n1 , (6.45)
where Km is the concentration at maximum flow, and and d are

then by using Q=KV, parameters. The fs-K and Q-K relations can be derived as shown
for Model System 2:
Q
K Vm (1 fs,min )n1 [1 (K/Kj )]n1 . (6.46)
f s
1 { (Vf /Vm ) exp [ (K/Km )d ] }1/(n1) and (6.51)
Equations 6.44 through 6.46 were fitted to the simulated data Q

K Vf exp[ ( K/Km )d ] . (6.52)
and are shown in Figure 6.24. Because the point representing
the highest concentration (about 100 vehicles/lane-mile) did not
lie in the same linear lnTr - lnT trend as the other points, the two-
fluid parameters n and Tm were estimated with and without the
highest concentration point, resulting in the Method 1 and Equations 6.50 through 6.52 were fitted to the simulation data
Method 2 curves, respectively, in the V-K and Q-K curves in and are shown in Figure 6.26.
Figure 6.24.
Two important conclusions can be drawn from this work. First,
Model System 2 adopts Greenshields' linear speed-concentration that relatively simple macroscopic relations between network-
relationship (Gerlough and Huber 1975), level variables appear to work. Further, two of the models
shown are similar to those established at the individual facility
level. Second, the two-fluid model serves well as the theoretical
V
Vf (1 K/Kj ) , (6.47) link between the postulated and derived functions, providing
another demonstration of the model's validity. In the second and
third model systems particularly, the derived fs-K function
where Vf is the free flow speed (and is distinct from Vm ; performed remarkable well against the simulated data, even
Vf Vm always, and typically Vf < Vm ). The fs-K relation can be though it was not directly calibrated using that data.
found by substituting Equation 6.47 into Equation 6.31 and
solving for fs :
f s
1 [(Vf /Vm) (1 K/Kj ) ]1/(n1) , (6.48)

0$&526&23,& )/2: 02'(/6
<
Figure 6.24
Comparison of Model System 1 with Observed Simulation Results
(Williams et al. 1987, Figure 5, 7, and 8).

0$&526&23,& )/2: 02'(/6
Figure 6.25

0$&526&23,& )/2: 02'(/6
Figure 6.26

$&526&23,& /2: 2'(/6
6.5 Concluding Remarks

As the scope of traffic control possibilities widens with the optimization of the control system) becomes clear. While the
development of ITS (Intelligent Transportation Systems) models discussed in this chapter are not ready for easy
applications, the need for a comprehensive, network-wide implementation, they do have promise, as in the application of
evaluation tool (as well as one that would assist in the the two-fluid model in San Antonio (Denney et al. 1993; 1994).
References
Denney, Jr., R. W., J. C. Williams, S. C. S. Bhat, and
Angel, S. and G. M. Hyman (1970). Urban Velocity Fields. S. A. Ardekani, (1993). Calibrating NETSIM for a CBD
Environment and Planning, Vol. 2. Using the Two-Fluid Model. Large Urban Systems
Ardekani, S. A. (1984). The Two-Fluid Characterization of (Proceedings of the Advanced Traffic Management
Urban Traffic: Theory, Observation, and Experiment. Conference), S. Yagar, A. Santiago, editors, Federal
Ph.D. Dissertation, University of Texas at Austin. Highway Administration, U.S. Department of
Ardekani, S. A., V. Torres-Verdin, and R. Herman (1985). Transportation.
The Two-Fluid Model and Traffic Quality in Mexico City Denney, Jr., R. W., J. C. Williams, and S. C. S. Bhat, (1994).
(El Modelo Bifluido y la Calidad del Tránsito en la Ciudad Calibrating NETSIM Using the Two-Fluid Model.
de México). Revista Ingeniería Civil. Compendium of Technical Papers (Proceedings of the 64th
Ardekani, S. A. and R. Herman, (1987). Urban Network-Wide ITE Annual Meeting, Dallas), Institute of Transportation
Variables and Their Relations. Transportation Science, Engineers.
Vol. 21, No. 1. Federal Highway Administration, U. S. Department of
Ardekani, S. A., J. C. Williams,, and S. Bhat, (1992). Influence Transportation (1993). TRAF User Reference Guide,
of Urban Network Features on Quality of Traffic Service. Version 4.0.
Transportation Research Record 1358, Transportation Gerlough, D. L. and M. J. Huber, (1975). Traffic Flow
Research Board. Theory: A Monograph, Special Report 165, Transportation
Ayadh, M. T. (1986). Influence of the City Geometric Research Board.
Features on the Two Fluid Model Parameters, M.S. Godfrey, J. W. (1969). The Mechanism of a Road Network.
Thesis, Virginia Polytechnic Institute and State University. Traffic Engineering and Control, Vol. 11, No. 7.
Beimborn, E. A. (1970). A Grid Travel Time Model. Herman, R. and S. A. Ardekani, (1984).
Transportation Science, Vol. 4. Characterizing Traffic Conditions in Urban Areas.
Bhat, S .C. S. (1994). Effects of Geometric and Control Transportation Science, Vol. 18, No. 2.
Features on Network Traffic: A Simulation Study. Ph.D. Herman, R. and I. Prigogine, (1979). A Two-Fluid Approach to
Dissertation, University of Texas at Arlington. Town Traffic. Science, Vol. 204, pp. 148-151.
Branston, D. M. (1974). Urban Traffic Speeds—I: A Hutchinson, T. P. (1974). Urban Traffic Speeds—II: Relation
Comparison of Proposed Expressions Relating Journey of the Parameters of Two Simpler Models to Size of the
Speed to Distance from a Town Center. Transportation City and Time of Day. Transportation Science, Vol. 8,
Science, Vol. 8, No. 1. No. 1.
Buckley, D. G. and J. G. Wardrop, (1980). Some General Lyman, D. A. and P. F. Everall, (1971). Car Journey Times in
Properties of a Traffic Network. Australian Road London—1970. RRL Report LR 416, Road Research
Research, Vol. 10, No. 1. Laboratory.
Mahmassani, H., J. C. Williams, and R. Herman, (1984).
Investigation of Network-Level Traffic Flow Relationships:
Some Simulation Results. Transportation Research Record
971, Transportation Research Board.

$&526&23,& /2: 2'(/6
Mahmassani, H. S., J. C. Williams, and R. Herman, (1987). Transportation Research Board (1994). Highway Capacity
Performance of Urban Traffic Networks. Transportation Manual, Special Report 209, (with revisions).
and Traffic Theory (Proceedings of the Tenth International Vaughan, R., A. Ioannou, and R. Phylactou (1972). Traffic
on Transportation and Traffic Theory Symposium, Characteristics as a Function of the Distance to the Town
Cambridge, Massachusetts), N.H. Gartner, N.H.M. Wilson, Centre. Traffic Engineering and Control, Vol. 14, No. 5.
editors, Elsevier. Wardrop, J. G. (1952). Some Theoretical Aspects of Road
Newell, G. F. (1980). Traffic Flow on Transportation Traffic Research. Proceedings of the Institution of Civil
Networks, Chapter 4, MIT Press. Engineers, Vol. 1, Part 2.
Prigogine, I. and R. Herman, (1971). Kinetic Theory of Wardrop, J. G. (1968). Journey Speed and Flow in Central
Vehicular Traffic, American Elsevier. Urban Areas. Traffic Engineering and Control, Vol. 9, No.
Road Research Laboratory (1965). Research on Road Traffic. 11.
Smeed, R. J. (1963). International Road Safety and Wardrop, J. G. (1969). Minimum Cost Paths in Urban Areas.
TrafficReview, Vol. 11, No. 1. Beiträge zur Theorie des Verkehrsflusses, Universitat (TH)
Smeed, R. J. and J. G. Wardrop, (1964). An Exploratory Karlsruhe.
Comparison of the Advantages of Cars and Buses for Williams, J. C., H. S. Mahmassani, and R. Herman (1995).
Travel in Urban Areas. Journal of the Institute of Sampling Strategies for Two-Fluid Model Parameter
Transportation, Vol. 30, No. 9. Estimation in Urban Networks. Transportation Research -
Smeed, R. J. (1965). Journal of the Institute of Mathematics Part A, Vol. 29A, No. 3.
and its Applications, Vol. 1. Williams, J. C., H. S. Mahmassani, and R. Herman, (1985).
Smeed, R. J. (1966). Road Capacity of City Centers. Traffic Analysis of Traffic Network Flow Relations and Two-Fluid
Engineering and Control, Vol. 8, No. 7. Model Parameter Sensitivity. Transportation Research
Smeed, R. J. (1968). Traffic Studies and Urban Congestion. Record 1005, Transportation Research Board.
Journal of Transport Economics and Policy, Vol. 2, Williams, J. C., H. S. Mahmassani, and R. Herman, (1987).
No. 1. Urban Traffic Network Flow Models. Transportation
Thomson, J. M. (1967a). Speeds and Flows of Traffic in Research Record 1112, Transportation Research Board.
Central London: 1. Sunday Traffic Survey. Traffic Zahavi, Y. (1972a). Traffic Performance Evaluation of Road
Engineering and Control, Vol. 8, No. 11. Networks by the -Relationship. Traffic Engineering and
Thomson, J. M. (1967b). Speeds and Flows of Traffic Control, Vol. 14, No. 5.
in Central London: 2. Speed-Flow Relations. Traffic Zahavi, Y. (1972b). Traffic Performance Evaluation of Road
Engineering and Control, Vol. 8, No. 12. Networks by the -Relationship, Part 2. Traffic
Engineering and Control, Vol. 14, No. 6.

TRAFFIC IMPACT MODELS
BY SIA ARDEKANI10
EZRA HAUER11
BAHRAM JAMEI12
10
Associate Professor, Civil Engineering Department, University of Texas at Arlington, Box 19308,
Arlington, TX 76019-0308
11
Professor, Department of Civil Engineering, University of Toronto, Toronto, Ontario Cananda M5S 1A4
12
Transportation Planning Engineer, Virginia Department of Transportation, Fairfax, Va.
Chapter 7 - Frequently used Symbols
= fuel consumption per unit distance

% = parameter
a = instantaneous acceleration rate (a>0, km/hr/sec)
A = acceleration rate
AADT = average annual daily traffic (vehicles/day)
ACCT = acceleration time
ADT = average daily traffic (vehicles/day)
C1 = average peak hour concentration
C8 = annual maximum 8-hour concentration
CLT = Deceleration time
ds = average stopped delay per vehicle (secs)
E = Engine size (cm3)
EFI = idle emission factor
EFL = intersection link composite emission factor
f = instantaneous fuel consumption (mL/sec)
F = fuel consumed (lit/km) [Watson, et al model]
F = average fuel consumption per roadway section (mL) [Akcelik model]
f1 = fuel consumption rate while cruising (mL/km)
f2 = fuel consumption rate while idling (mL/sec)
f3 = excess fuel consumption per vehicle stop (mL)
fc = steady-state fuel consumption rate at cruising speed (mL/km)
h = average number of stops per vehicle
K1 = parameter representing idle flow rate (mL/sec)
K2 = parameter representing fuel consumption to overcome rolling resistance
K4,K5 = parameters related to fuel consumption due to positive acceleration
L = payload (Kg)
LQU = queue length
m = expected number of single-vehicle accidents per unit time
NDLA = vehicles delayed per cycle per lane
P = probability of a single-vehicle accident
PKE = positive kinetic energy
q = flow (vph)
qi = flow rate of type i vehicles (vph)
S = speed
SPD = cruise speed
T = average travel time per unit distance
u = model parameter related to driving conditions
V = average speed (km/hr) [Elemental model]
V = instantaneous speed (km/hr) [Akcelik and Bayley model]
Vc = steady-state cruising speed (km/hr)
Vf = final speed (km/hr)
Vi = initial speed (km/hr)
VMT = vehicle miles of travel
Vs = space mean speed (km/hr)
VSP = at-rest vehicle spacing
7.
TRAFFIC IMPACT MODELS
7.1 Traffic and Safety

7.1.1 Introduction
This section ought to be about how traffic flow, speed and the pertains to the same period of time which the accident frequency
like are related to accident frequency and severity. However, represents. Thus, eg., if the ordinate shows the expected number
due to limitation of space, only the relationship between accident of fatal accidents/year in 1972-1976 for a certain road section,
frequency and traffic flow will be discussed. The terminology then the AADT is the average for the period 1972-1976.
that pertains to characteristics of the traffic stream has already
been established and only a few definitions need to be added. Naturally, mi can be a function of more than one traffic flow.
The ‘safety’ of an entity is defined as ‘the number of accidents Thus, eg., head-on collisions may depend on the two opposing
by type, expected to occur on the entity in a certain period, flows; collisions between pedestrians and left-turning traffic
per unit of time’. In this definition, ‘accident types’ are depend on the flow of pedestrians, the flow of straight-through
categories such as rear-end, sideswipe, single-vehicle, multi- vehicles, and the flow of left-turning vehicles etc.. In short, the
vehicle, injury, property damage only, etc. The word ‘expected’ arguments of the safety performance function can be several
is as in probability theory: what would be the average-in-the- flows.
long-run if it was possible to freeze all the relevant
circumstances of the period at their average, and then repeat it In practice it is common to use the term ‘accident rate’. The
over and over again. The word ‘entity’ may mean a specific road accident rate is proportional to the slope of the line joining the
section or intersection, a group of horizontal curves with the origin and a point of the safety performance function. Thus, at
same radius, the set of all signalized intersections in point A of Figure 7.1, where AADT is 3000 vehicles per day
Philadelphia, etc. Since the safety of every entity changes in and where the expected number of accidents for this road section
time, one must be specific about the period. Furthermore, to is 1.05 accidents per year, the accident rate is
facilitate communication, safety is usually expressed as a 1.05/(3000×365)=0.96×10-6 accidents/vehicle. At point B the
frequency. Thus, eg., one might speak about the expected accident rate is 1.2/(4000×365)=0.82×10-6 accidents/vehicle. If
number of fatal accidents/year in 1972-1976 for a certain road the road section was, say, 1.7 km long, the same accident rates
section. To standardize further one often divides by the section could be written as 1.05/(3000×365×1.7)=0.56×10-6
length; now the units may be, say, accidents/(year × km). accidents/vehicle-km and 1.2/(4000×365×1.7)=0.48×10-6
accidents/vehicle-km.
As defined, the safety of an entity is a string of expected
frequencies, m1, m2, . . . ,mi, . . . , one for each accident type The safety performance function of an entity is seldom a straight
chosen. However, for the purpose of this discussion it will line. If so, the accident rate is not constant but varies with traffic
suffice to speak about one (unspecified) accident type, the flow. As a consequence, if one wishes to compare the safety of
expected accident frequency of which is mi. two or more entities serving different flows, one can not use the
accident rate for this purpose. The widespread habit of using
accident rates to judge the relative safety of different entities or
7.1.2 Flow and Safety to assess changes in safety of the same entity is inappropriate and
often harmful. To illustrate, suppose that the AADT on the road
The functional relationship between mi and the traffic flow which section in Figure 7.1 increased from 3000 ‘before pavement
the entity serves, is a ‘safety performance function’. A safety resurfacing’ to 4000 ‘after pavement resurfacing’ and that the
performance function is depicted schematically in Figure 7.1. average accident frequency increased from 1.05 ‘before’ to 1.3
For the moment its shape is immaterial. It tells how for some ‘after’. Note that 1.2 accidents/year would be expected at
entity the expected frequency of accidents of some type would be AADT=4000 had the road surface remained unchanged (see
changing if traffic flow on the entity changed while all other Figure 7.1). Since 1.3 > 1.2 one must conclude that following
conditions affecting accident occurrence remained fixed. While resurfacing there was a deterioration of 0.1 accidents/year. But
the flow may be in any units, it is usually understood that it

75$)),& ,03$&7 02'(/6
Figure 7.1
Safety Performance Function and Accident Rate.
less than the accident rate ‘before resurfacing’ (1.05/3000×365= 7.1.3 Logical Considerations
0.96×10-6) which erroneously suggests that there has been an
improvement. Similar arguments against the use of accident It stands to reason that there is some kind of relationship
rates can be found in Pfundt (1969), Hakkert et al. (1976), between traffic flow and safety. For one, without traffic there are
Mahalel (1986), Brundell-Freij & Ekman (1991), Andreassen no traffic accidents. So, the safety performance function must go
(1991). through the origin. Also, the three interrelated characteristics of
the traffic stream - flow, speed and density - all influence the
To avoid such errors, the simple rule is that safety comparisons three interrelated aspects of safety - the frequency of
are legitimate only when the entities or periods are compared as opportunities for accidents to occur, the chance of accident
if they served the same amount of traffic. To accomplish occurrence given an opportunity, and the severity of the outcome
such equalization, one needs to know the safety performance given an accident. However, while a relationship may be
function. Only in the special case when the performance presumed to exist, it is rather difficult to learn much about its
function happens to be a straight line, may one divide by traffic mathematical form by purely deductive reasoning.
flow and then compare accident rates. However, to judge
whether the safety performance function is a straight line, one Using logic only, one could argue as follows: "If, as in
must know its shape, and when the shape of the safety probability theory, the passage of a vehicle through a road
performance function is known, the computation of an accident section or an intersection is a ‘trial’ the ‘outcome’ of which can
rate is superfluous. It is therefore best not to make use of be ‘accident’ or ‘no-accident’ with some fixed probability.
accident rates. For this reason, the rest of the discussion is about Assume further that vehicle passages are so infrequent that this
expected accident frequencies, not rates. probability is not influenced by the frequency at which the
‘trials’ occur. Under such conditions the expected number of
Knowledge of safety performance functions is an important single-vehicle accidents in a fixed time period must be
element of rational road safety management. The nature and proportional to the number of trials in that time period - that is
shape of this function is subject to some logical considerations. to flow." In symbols, msingle-vehicle=qp, where q is flow and p is the
However, much of the inquiry must be empirical. probability of a single-vehicle accident in one passage of a

75$)),& ,03$&7 02'(/6
vehicle. In this, p is a constant that does not depend on q. Thus, Most theoretical inquiry into the relationship between flow and
one may argue, that the number of single-vehicle accidents ought safety seems to lack detail. Thus, eg., most researchers try to
to be proportional to flow, but only at very low flows. relate the frequency of right-angle collisions at signalized
intersections to the two conflicting flows. However, on
As flow and density increase to a point where a driver can see reflection, the second and subsequent vehicles of a platoon may
the vehicle ahead, the correspondence between the mental have a much lesser chance to be involved in such a collision than
picture of independent trials and between reality becomes the first vehicle. Therefore it might make only a slight difference
strained. The probability of a ‘trial’ to result in a single-vehicle whether 2 or 20 vehicles have to stop for the same red signal.
accident now depends on how close other vehicles are; that is For this reason, the total flow is likely to be only weakly and
p = p(q). Should p(q) be an increasing function of q, then circuitously related to the number of situations which generate
msingle-vehicle would increase more than in proportion with flow. right angle collisions at signalized intersections. There seems to
Conversely, if an increase in flow diminishes the probability that be scope and promise for more detailed, elaborate and realistic
a vehicle passing the road section will be in a single-vehicle theorizing. In addition, most theorizing to date attempted to
accident, then msingle-vehicle would increase less than in proportion relate safety to flow only. However, since flow, speed and
to traffic flow; indeed, msingle-vehicle= qp(q) can even decrease as density are connected, safety models could be richer if they
traffic flow increases beyond a certain point. Thus, by logical contained all relevant characteristics of the traffic stream. Thus,
reasoning one can only conclude that near the origin, the safety eg., a close correspondence has been established between the
performance function for single-vehicle accidents ought to be a number of potential overtakings derived from flow and speed
straight line. distribution and accident involvement as a function of speed
(Hauer 1971). Welbourne (1979) extends the ideas to crossing
If the safety performance function depends on two conflicting traffic and collisions with fixed objects. Ceder (1982) attempts
flows (car-train collisions at rail-highway grade crossings, car- to link accidents to headway distributions (that are a function of
truck collisions on roads, car-pedestrian collision at intersections flow) through probabilistic modeling.
etc.) then, near the origin, mi should be proportional to the
product of the two flows. One could also use the paradigm of There is an additional aspect of the safety performance function
probability theory to speculate that (at very low flows) the which may benefit from logical examination. The claim was that
expected number of collisions with vehicles parked on shoulders it is reasonable to postulate the existence of a relationship
is proportional to the square of the flows: in the language of between the traffic triad ‘flow, speed and density’ and between
‘trials’, ‘outcomes’, the number of vehicles parked on the the safety triad ‘frequency of opportunities, chance of accident
shoulder ought to be proportional to the passing flow and the given opportunity, and severity of the outcome given accident’.
number of vehicles colliding with the parked cars ought to be However, if there is a cause-effect relationship, it must be
proportional to the same flow. From here there is only a small between accidents and the traffic characteristics near the time of
step to argue that, say, the number of rear-end collisions should their occurrence. One must ask whether there still is some
also be proportional to q2. Again, this reasoning applies only to meaningful safety performance function between accidents and
very low flows. How m depends on q when speed choice, traffic flow when flow is averaged over, say, a year. Whether the
alertness and other aspects of behavior are also a function of habit of relating accidents to AADTs (that is, averages over a
flow, cannot be anticipated by speculation alone. year) materially distorts the estimated safety performance
function is at present unclear. In a study by Quaye et al. (1993)
This is as far as logical reasoning seems to go at present. It only three separate models were estimated from 15 minute flows,
tells us what the shape of the safety performance function should which then were aggregated into 1 hour flows and then into 7
be near the origin. Further from the origin, when p changes with hour flows. The three models differed but little. Persaud and
q, not much is gained thinking of m as the product qp(q). Since Dzbik (1993) call models that relate hourly flows to accidents
the familiar paradigm of ‘trials’ and ‘outcomes’ ceased to fit "microscopic" and models that relate AADT to yearly accident
reality, and the notion of ‘opportunity to have an accident’ is counts "macroscopic".
vague, it might be better to focus directly on the function m =
m(q) instead of its decomposition into the product qp(q). As will become evident shortly, empirical inquiries about safety
performance functions display a disconcerting variety of results.

75$)),& ,03$&7 02'(/6
A part of this variety could be explained by the fact that the most whether the result of a cross section study enables one to
ubiquitous data for such inquiries consist of flow estimates anticipate how safety of a certain facility would change as a
which are averages pertaining to periods of one year or longer. result of a change in traffic flow.
Less common are studies that relate different traffic flows on the
7.1.4 Empirical Studies same facility to the corresponding accident counts. In this case
we have a ’time-sequence’ study. In such a study one obtains
Empirical studies about the association of traffic flow and the flow that prevailed at the time of each accident and the
accident frequency seldom involve experimentation; their nature number of hours in the study period when various flows
is that of fitting functions to data. What is known about safety prevailed. The number of accidents in a flow group divided by
performance functions comes from studies in which the the number of hours when such flows prevailed is the accident
researcher assembles a set of data of traffic flows and accident frequency (see, eg., Leutzbach et al. 1970, Ceder and Livneh
counts, chooses a certain function that is thought to represent the 1978, Hall and Pendelton 1991). This approach might obviate
relationship between the two, and then uses statistical techniques some of the problems that beset the cross-section study.
to estimate the parameters of the chosen function. Accordingly, However, the time-sequence study comes with its own
discussion here can be divided into sections dealing with: (1) the difficulties. If data points are AADTs and annual accident
kinds of study and data, (2) functional forms or models, (3) counts over a period of many years, then the range of the AADTs
parameter estimates. is usually too small to say much about any relationship. In
addition, over the many years, driver demography, norms of
behavior, vehicle fleet, weather, and many other factors also
7.1.4.1 Kinds Of Study And Data change. It is therefore difficult to distinguish between what is
due to changes in traffic flow and what is due to the many other
Data for the empirical investigations are accident counts and factors that have also changed. If the data points are traffic flows
traffic flow estimates. A number-pair consisting of the accident and accidents over a day, different difficulties arise. For one, the
count for a certain period and the estimated flow for the same count of accidents (on one road and when traffic flow is in a
period is a ‘data point’ in a Cartesian representation such as specified range) is bound to be small. Also, low traffic flows
Figure 7.1. To examine the relationship between traffic flow and occur mostly during the night, and can not be used to estimate
accident frequency, many such points covering an adequate the safety performance function for the day. Also, peak hour
range of traffic flows are required. drivers are safer en-route to work than on their way home in the
afternoon, and off-peak drivers tend to be a different lot
There are two study prototypes (for a discussion see eg., Jovanis altogether.
and Chang 1987). The most common way to obtain data points
for a range of flows is to choose many road sections or
intersections that are similar, except that they serve different 7.1.4.2 Models
flows. In this case we have a ‘cross-section’ study. In such a
study the accident counts will reflect not only the influence of The first step of an empirical study of the relationship between
traffic flow but also of all else that goes with traffic flow. In traffic flow and accident frequency is to assemble, plot and
particular, facilities which carry larger flows tend to be built and examine the data. The next step is to select the candidate model
maintained to higher standards and tend to have better markings equation(s) which might fit the data and serve as the safety
and traffic control features. This introduces a systematic bias performance function. Satterthwaite (1981, section 3) reviews
into ‘cross-section’ models. If a road that is built and maintained the most commonly used models. Only those that are plausible
to a higher standard is safer, then accident counts on high-flow and depend on traffic flow only are listed below. Traffic flow,
roads will tend to be smaller than had the roads been built, while important, is but one of the many variables on which the
maintained and operated in the same way as the lower flow expected accident frequency depends. However, since the
roads. Thus, in a cross-section study, it is difficult to separate monograph is devoted to traffic flow, the dependence on other
what is due to traffic flow, and what is due to all other factors variables will not be pursued here. The Greek letters are the
which depend on traffic flow. It is therefore questionable unknown parameters to be estimated from data.

75$)),& ,03$&7 02'(/6
When only one traffic stream is relevant the power function and The common feature of the models most often used is that they
polynomial models have been used: are linear or can be made so by logarithmization. This simplifies
statistical parameter estimation. The shapes of these functions
m = q (7.1) are shown in Figure 7.2.
m = q + q2 +... (7.2)
The power function (Equation 7.1) is simple and can well satisfy
At times the more complex power form the logical requirements near the origin (namely, that when q=0
m=0, and that =1 when one flow is involved or =2 when two
m = q+log(q) (7.1a) flows are involved). However, its simplicity is also its downfall.
If logic dictates, eg., that near the origin =1 (say, for single-
is used which is also akin to the polynomial model 7.2 when vehicle accidents), then the safety performance function has to be
written in logarithms a straight line even for the higher flows where p(q) is not
constant any more. Similarly, if logic says that =2, the
log(m) = log() + log(q) + [log(q)] 2 (7.2a) quadratic growth applies for all q. In short, if is selected to
meet requirements of logic, the model may not fit the data further
When two or more traffic streams or kinds of vehicles are from the origin. Conversely, if is selected to fit the data best,
relevant, the product of power functions seems common: the logical requirements will not be met.
m = q1q2... (7.3)
Figure 7.2
Shapes of Selected Model Equations.

75$)),& ,03$&7 02'(/6
The popularity of the power function in empirical research where k = 1 or 2 in accord with the near-origin requirements.
derives less from its suitability than from it being ‘lazy user When <0 the function has a maximum at q = -k/. Its form
friendly’; most software for statistical parameter estimation can when k=1 and k=2 is shown in Figure 7.3. The advantage of this
accommodate the power function with little effort. The model is that it can meet the near-origin requirements and still
polynomial model (Equation 7.2) never genuinely satisfies the can follow the shape of the data.
near-origin requirements. Its advantage is that by using more
terms (and more parameters) the curve can be bent and shaped A word of caution is in order. In the present context the focus is
almost at will. This is achieved at the expense of parsimony in on how accident frequency depends on traffic flow. Accordingly,
parameters. the models were written with flow (q) as the principal
independent variable. However, traffic flow is but one of the
If the data suggest that as flow increases beyond a certain level, many causal factors which affect accident frequency. Road
the slope of m(q) is diminishing, perhaps even grows negative, geometry, time of day, vehicle fleet, norms of behaviour and the
an additional model that is parsimonious in parameters might like all play a part. Therefore, what is at times lumped into a
deserve consideration: single parameter ( in equation 7.1, 7.1a, 7.3 and 7.4) really
represents a complex multivariate expression. In short, the
m=qkeq (7.4) modeling of accident frequency is multivariate in nature.
Figure 7.3
Two Forms of the Model in Equation 7.4.

75$)),& ,03$&7 02'(/6
7.1.4.3 Parameter Estimates in model 7.1 is 0.58; for rear-end accidents the exponent of
AADT is 1.43. In a time-sequence study, Hall and Pendelton
With the data in hand and the functional form selected, the next (1990) use ten mile 2 and 4-lane road segments surrounding
step is to estimate the parameters (, ,...). In earlier work, permanent counting stations in New Mexico and provide a
estimation was often by minimization of squared deviations. wealth of information about accident rates in relations to
This practice now seems deficient. Recognizing the discrete hourly flows and time of day. In an extensive cross-section
nature of accident counts, the fact that their variance increases study, Zegeer et al. (1986) find that the exponent of ADT is
with their mean, and the possible existence of over-dispersion, 0.88 for the total number of accidents on rural two-lane
it now seems that more appropriate statistical techniques are roads. Ng and Hauer (1989) use the same data as Zegeer
called for (see eg., Hauer 1992, Miaou and Lum 1993). and show that the parameters differ from state to state and
also by lane width. For non-intersection accidents on rural
Results of past research are diverse. Part of the diversity is due two lane roads in New York State, Hauer et al. (1994) found
to the problems which beset both the cross-section and the time that when m is measured in [accidents/(mile-year)] and
sequence of studies; another part is due to the use of AADT and AADT is used for q, then in model 1, in 13 years varies
similar long-period averages that have a less than direct tie to from 0.0024-0.0028 and =0.78. Persaud (1992) using data
accident occurrence; some of the diversity may come from on rural roads in Ontario finds the exponent of AADT to vary
various methodological shortcomings (focus on accident rates, between 0.73 and 0.89, depending on lane and shoulder
choice of simplistic model equations, use of inappropriate width. For urban two-lane roads in Ontario the exponent is
statistical techniques); and much is due to the diversity between 0.72 For urban multi-lane roads (divided or undivided)
jurisdictions in what counts as a reportable accident and in the =1.14, for rural multi-lane divided roads it is 0.62 but for
proportion of reportable accidents that get reported. Hauer and undivided roads it is again 1.13.
Persaud (1996) provide a comprehensive review of safety
performance functions and their parameter values (for two-lane For California freeways Lundy (1965) shows that the
roads, multi-lane roads without access control, freeways, accidents per million vehicle miles increase roughly linearly
intersections and interchanges) based on North American data. with ADT. This implies the quadratic relationship of model
A brief summary of this information and some international 2. Based on the figures in Slatterly and Cleveland (1969),
results are given below. with m measured in [accidents/day], m = (5.8×10-7)ADT +
(2.4×10-11)ADT2 for four-lane freeways, m = (6.6×10-7)ADT
A. Road Sections. In a cross-section study of Danish rural + (.94 × 10-11)ADT2 for six-lane freeways and m = (5.4 ×
roads Thorson (1967) estimated the exponent of ADT to be 10 -7 ) × ADT + (.78 × 10-11) × ADT2 for eight-lane
0.7. In a similar study of German rural roads Pfundt (1968) freeways. Leutzbach (1970) examines daytime accidents on
estimated the exponent of ADT to be 0.85. Kihlberg and a stretch of an autobahn. Fitting a power function to his
Tharp (1968) conducted an extensive cross-section study Figure (1c) and with m measured in accidents per day,
using data from several states. For sections that are 0.5 miles m =(3×10-11)×(hourly flow)3. However, there is an indication
long, they estimate the parameters for a series of road types in this and other data that as flow increases, the accident rate
and geometric features. The model used is an elaborated initially diminishes and then increases again. If so a third
power function m=(ADT)(ADT)log(ADT). The report degree polynomial might be a better choice. Jovanis and
contains a rich set of results but creates little order in the Chang (1987) fit model 7.3 to the Indiana Toll Road and find
otherwise bewildering variety. Ceder and Livneh (1982) the exponents to be 0.25 and 0.23 for car and truck-miles.
used both cross-sectional and time-sequence data for Persaud and Dzbik (1993) find that when yearly accidents
interurban road sections in Israel, using the simple power are related to AADT, m=0.147×(AADT/1000)1.135 for 4-lane
function model (Equation. 1). The diverse results are difficult freeways but, when hourly flows are related to
to summarize. Cleveland et al. (1985) divide low-volume accidents/hour, m = 0.00145 × (hourly flow/1000)0.717.
rural two-lane roads into ‘bundles’ by geometry and find the Huang et al. (1992) report for California that Number of
ADT exponents to range from 0.49 to 0.93 for off-road accidents = 0.65 + 0.666 × million-vehicle-miles.
accidents Recent studies in the UK show that on urban road
sections, for single-vehicle accidents the exponent of AADT

75$)),& ,03$&7 02'(/6
B. Intersections. Tanner (1953) finds for rural T intersections turning cars)1.32 × (hourly pedestrian flow)0.34; when the left
in the UK the exponents to be 0.56 and 0.62 for left turning turning vehicles have to find gaps in the opposing traffic,
traffic and main road traffic respectively. Roosmark (1966) m = 1.29 × 10-7 (hourly flow of left-turning cars)0.36 ×
finds for similar intersections in Sweden the corresponding (hourly pedestrian flow) 0.86.
exponents to be 0.42 and 0.71. For intersections on divided
highways McDonald (1953) gives m in accidents per year as
m = 0.000783 × (major road ADT)0.455 × (cross road 7.1.5 Closure
ADT)0.633. For signalized intersections in California, Webb
(1955) gives m in accidents per year, as m = 0.00019 × Many aspects of the traffic stream are related to the frequency
(major road ADT)0.55 × (cross road ADT)0.55 in urban areas and severity of accidents; only the relationship with flow has
with speeds below 40 km/h ; = 0.0056 × (major road been discussed here. How safety depends on flow is important
ADT) 0.45 × (cross road ADT) 0.38 in semi-urban areas with to know. The relationship of traffic flow to accident frequency
speeds in the 40-70 km/h range; and = 0.007 × (major is called the ‘safety performance function’. Only when the safety
road ADT)0.51 × (cross road ADT) 0.29in rural areas with performance function is known, can one judge whether one kind
speeds above 70 km/h. For rural stop-controlled intersections of design is safer than another, or whether an intervention has
in Minnesota, Bonneson and McCoy (1993) give m [in affected the safety of a facility. Simple division by flow to
accidents/year] = 0.692 × (major road ADT/1000)0.256 × compute accident rates is insufficient because the typical safety
(cross road ADT/1000)0.831. Recent studies done in the UK performance function is not linear.
show that at signalized intersections for single vehicle
accidents the exponent of AADT in model 1 is 0.89; for right Past research about safety performance functions has led to
angle accidents (model 7.3) the exponents of AADT are 0.36 diverse results. This is partly due to the use of flow data which
and 0.60 and for accident to left-turning traffic (their right are an average over a long time period (such as AADT), partly
turn) the exponents are 0.57 and 0.46. Using data from due to the difficulties which are inherent in the cross-sectional
Quebec, Belangér finds that the expected annual and the time-sequence studies, and partly because accident
number of accidents at unsignalized rural intersections is reporting and roadway definitions vary among jurisdictions.
0.002×(Major road ADT)0.42×(Cross road ADT)0.51. However, a large part of the diversity is due to the fact that
accident frequency depends on many factors in addition to traffic
C. Pedestrians. Studies in the UK show that for nearside flow and that the dependence is complex.Today, some of the past
pedestrian accidents on urban road sections the exponents for difficulties can be overcome. Better information about traffic
vehicle and pedestrian AADTs in model 7.3 are 0.73 and flows is now available (eg. from freeway-traffic-management-
0.42. With m measured in [pedestrian accidents/year] systems, permanent counting stations, or weight-in-motion
Brüde and Larsson (1993) find that, at intersections, devices); also better methods for the multivariate analysis of
m = (7.3 × 10-6)(incoming traffic/day)0.50 (crossing accident counts now exist. However, in addition to progress in
pedestrians/day)0.7 . With m measured in (pedestrian statistical modeling, significant advances seem possible through
accidents/hour), Quaye et al. (1993) find that if left-turning the infusion of detailed theoretical modeling which makes use of
vehicles at signalized intersections do not face opposing all relevant characteristics of the traffic stream such as speed,
vehicular traffic then m = 1.82 × 10-8(hourly flow of left- flow, density, headways and shock waves.
7.2 Fuel Consumption Models

Substantial energy savings can be achieved through urban traffic saved each day. It is further estimated that 45 percent of the total
management strategies aimed at improving mobility and energy consumption in the U.S. is by vehicles on roads. This
reducing delay. It is conservatively estimated, for example, that amounts to some 240 million liters (63 million gallons) of
if all the nearly 250,000 traffic signals in the U.S. were optimally petroleum per day, of which nearly one-half is used by vehicles
timed, over 19 million liters (5 million gallons) of fuel would be under urban driving conditions.

75$)),& ,03$&7 02'(/6
Fuel consumption and emissions have thus become increasingly pavement roughness, grades, wind, altitude, etc. is hoped to be
important measures of effectiveness in evaluating traffic small due to collective effects if data points represent aggregate
management strategies. Substantial research on vehicular energy values over sufficiently long observation periods. Alternatively,
consumption has been conducted since the 1970's, resulting in model estimates may be adjusted for known effects of roadway
an array of fuel consumption models. In this section, a number grade, ambient temperature, altitude, wind conditions, payload,
of such models which have been widely adopted are reviewed. etc.
Given a fixed set of vehicle and driver characteristics and

7.2.1 Factors Influencing Vehicular environmental conditions, the influence of traffic-related factors
on fuel consumption can be modeled. A number of studies in
Fuel Consumption
Great Britain (Everall 1968), Australia (Pelensky et al. 1968),
and the United States (Chang et al. 1976; Evans and Herman
Many factors affect the rate of fuel consumption. These factors
1978; Evans et al. 1976) all indicated that the fuel consumption
can be broadly categorized into four groups: vehicle,
per unit distance in urban driving can be approximated by a
environment, driver, and traffic conditions. The main variables
linear function of the reciprocal of the average speed. One such
in the traffic category include speed, number of stops, speed
model was proposed by Evans, Herman, and Lam (1976), who
noise and acceleration noise. Speed noise and acceleration noise
studied the effect of sixteen traffic variables on fuel consumption.
measure the amount of variability in speed and acceleration in
They concluded that speed alone accounts for over 70 percent of
terms of the variance of these variables. The degree of driver
the variability in fuel consumption for a given vehicle.
aggressiveness also manifests itself in speed and acceleration
Furthermore, they showed that at speeds greater than about 55
rates and influences the fuel consumption rate.
km/h, fuel consumption rate is progressively influenced by the
aerodynamic conditions. They classified traffic conditions as
Factors related to the driving environment which could affect
urban (V < 55 km/h) versus highway (V > 55 km/h) traffic
fuel consumption include roadway gradient, wind conditions,
showing that unlike the highway regime, in urban driving fuel
ambient temperature, altitude, and pavement type (for example,
efficiency improves with higher average speeds (Figure 7.4).
AC/PCC/gravel) and surface conditions (roughness, wet/dry).
Vehicle characteristics influencing energy consumption include
total vehicle mass, engine size, engine type (for example,
gasoline, diesel, electric, CNG), transmission type, tire type and 7.2.3 Urban Fuel Consumption Models
size, tire pressure, wheel alignment, the status of brake and
carburetion systems, engine temperature, oil viscosity, gasoline Based on the aforementioned observations, Herman and co-
type (regular, unleaded, etc.), vehicle shape, and the degree of workers (Chang and Herman 1981; Chang et al. 1976; Evans
use of auxiliary electric devices such as air-conditioning, radio, and Herman 1978; Evans et al. 1976) proposed a simple
wipers, etc. A discussion of the degree of influence of most of theoretically-based model expressing fuel consumption in urban
the above variables on vehicle fuel efficiency is documented by conditions as a linear function of the average trip time per unit
the Ontario Ministry of Transportation and Communications distance (reciprocal of average speed). This model, known as
(TEMP 1982). the Elemental Model, is expressed as:
7.2.2 Model Specifications =K 1 + K2 T, V < 55 km/hr (7.5)

where,
Fuel consumption models are generally used to evaluate the : fuel consumption per unit distance
effectiveness and impact of various traffic management T : average travel time per unit distance
strategies. As such these models are developed using data and
collected under a given set of vehicle fleet and performance V(=1/T) : average speed
characteristics such as weight, engine size, transmission type,
tire size and pressure, engine tune-up and temperature K1 and K2 are the model parameters. K1 (in mL/km) represents
conditions, etc. The variation in fuel consumption due to other fuel used to overcome the rolling friction and is closely related
factors such as driver characteristics, ambient temperature,

75$)),& ,03$&7 02'(/6
Figure 7.4
Fuel Consumption Data for a Ford Fairmont (6-Cyl.)
Data Points represent both City and Highway Conditions.
to the vehicle mass (Figure 7.5). K2 (in mL/sec) is a function In F = average fuel consumption per roadway section
an effort to improve the accuracy of the Elemental Model, other (mL)
researchers have considered additional independent variables. Xs = total section distance (km)
Among them, Akcelik and co-workers (Akcelik 1981; ds = average stopped delay per vehicle (secs)
Richardson and Akcelik1983) proposed a model which h = average number of stops per vehicle
separately estimates the fuel consumed in each of the three f1 = fuel consumption rate while cruising (mL/km)
portions of an urban driving cycle, namely, during cruising, f2 = fuel consumption rate while idling (mL/sec)
idling, and deceleration-acceleration cycle. Hence, the fuel f3 = excess fuel consumption per vehicle stop (mL)
consumed along an urban roadway section is estimated as:
The model is similar to that used in the TRANSYT-7F
F = f1 Xs + f2 ds + f3 h, (7.11) simulation package (Wallace 1984).
where, Herman and Ardekani (1985), through extensive field studies,

have shown that delay and number of stops should not be used

75$)),& ,03$&7 02'(/6
Figure 7.5
Fuel Consumption
Versus Trip Time
per Unit Distance for a Number of Passenger Car Models.
together in the same model as estimator variables. This is due where,

to the tendency for the number of stops per unit distance to be Vf = final speed (km/hr)
highly correlated with delay per unit distance under urban traffic Vi = initial speed (km/hr)
conditions. They propose an extension of the elemental model Xs = total section length (km)
of Equation 7.5 in which a correction is applied to the fuel
consumption estimate based on the elemental model depending A number of other urban fuel consumption models are discussed
on whether the number of stops made is more or less than the by other researchers, among which the work by Hooker et al.
expected number of stops for a given average speed (Herman (1983), Fisk (1989), Pitt et al. (1987), and Biggs and Akcelik
and Ardekani 1985). (1986) should be mentioned.
A yet more elaborate urban fuel consumption model has been set
forth by Watson et al. (1980). The model incorporates the 7.2.4 Highway Models
changes in the positive kinetic energy during acceleration as a
predictor variable, namely, Highway driving corresponds to driving conditions under which
average speeds are high enough so that the aero-dynamic effects
F = K1 + K2/Vs + K3 Vs + K4 PKE, (7.12) on fuel consumption become significant. This occurs at average
speeds over about 55 km/h (Evans et al. 1976). Two highway
where, models based on constant cruising speed are those by Vincent et
F = fuel consumed (Lit/km) al. (1980) and Post et al. (1981). The two models are valid at
Vs = space mean speed (km/hr) any speed range, so long as a relatively constant cruise speed can
be maintained (steady state speeds). The steady-state speed
The term PKE represents the sum of the positive kinetic energy requirement is, of course, more easily achievable under highway
changes during acceleration in m/sec2, and is calculated as driving conditions.
follows:
PKE =
(Vf 2 - Vi2)/(12,960 Xs) (7.13)

75$)),& ,03$&7 02'(/6
480
360 1974 Standard -Sized car
240
120
1976 Subcompact car
0 75 150 225 300 375
Average Trip Time Unit Distance, T (Secs/ km)
Figure 7.6
Fuel Consumption Data and the Elemental Model Fit
for Two Types of Passenger Cars (Evans and Herman 1978).
The model by Vincent, Mitchell, and Robertson is used in the Calibration of this model for a Melbourne University test car
TRANSYT-8 computer package (Vincent et al. 1980) and is in (Ford Cortina Wagon, 6-Cyl, 4.1L, automatic transmission)
the form: yields (Akcelik 1983) the following parameter values (Also see
Figure 7.7):
fc = a + b Vc + c Vc2,
(7.14) b1 = 15.9 mL/km
b2 = 2,520 mL/hr
where, b3 = 0.00792 mL-hr2/km3.
Vc = steady-state cruising speed (km/hr) Instantaneous fuel consumption models may also be used to
fc = steady-state fuel consumption rate at cruising estimate fuel consumption for non-steady-state speed conditions
speed (mL/km), calibration of this model for a under both urban or highway traffic regimes. These models are
mid-size passenger car yields used in a number of microscopic traffic simulation packages
such as NETSIM (Lieberman et al. 1979) and MULTSIM
a = 170 mL/km, (Gipps and Wilson 1980) to estimate fuel consumption based on
b = -4.55 mL-hr/km2; and instantaneous speeds and accelerations of individual vehicles.
c = 0.049 mL-hr2/km3 (Akcelik 1983).
A second steady-state fuel model formulated by Post et al. By examining a comprehensive form of the instantaneous model,
(1981) adds a V2 term to the elemental model of Equation 7.5 to Akcelik and Bayley (1983) find the following simpler form of
account for the aero-dynamic effects, namely, the function to be adequate, namely,
fc = b1 + b2 /Vc + b3 Vc2. (7.15) f = K1 + K2 V + K3 V3 + K4 aV + K5 a2V

(7.16)

75$)),& ,03$&7 02'(/6
300
250
200
150
100
50
0 20 40 60 80 100
Constant Cruise Speed, Vc (km/h)
Figure 7.7
Constant-Speed Fuel Consumption per Unit Distance
for the Melbourne University Test Car (Akcelik 1983).
where, 7.2.5 Discussion

f = instantaneous fuel consumption (mL/sec)
V = instantaneous speed (km/hr) During the past two decades, the U.S. national concerns over
a = instantaneous acceleration rate dependence on foreign oil and air quality have renewed
(a > 0)(km/hr/sec) interest in vehicle fuel efficiency and use of alternative fuels.
K1 = parameter representing idle flow rate Automotive engineers have made major advances in vehicular
(mL/sec) fuel efficiency (Hickman and Waters 1991; Greene and Duleep
K2 = parameter representing fuel consumption 1993; Komor et al. 1993; Greene and Liu 1988).
to overcome rolling resistance
K3 = parameter representing fuel consumption Such major changes do not however invalidate the models
to overcome air resistance presented, since the underlying physical laws of energy
K4, K5 = parameters related to fuel consumption due consumption remain unchanged. These include the relation
to positive acceleration between energy consumption rate and the vehicle mass, engine
size, speed, and speed noise. What does change is the need to
The above model has been used by Kenworthy et al. (1986) to recalibrate the model parameters for the newer mix of vehicles.
assess the impact of speed limits on fuel consumption. This argument is equally applicable to alternative fuel vehicles
(DeLuchi et al. 1989), with the exception thatthere may also be
a need to redefine the variable units, for example, from mL/sec
or Lit/km to KwH/sec or KwH/km, respectively.

75$)),& ,03$&7 02'(/6
7.3 Air Quality Models

7.3.1 Introduction Carbon monoxide which is one of the main pollutants is a
colorless, and poisonous gas which is produced by the
Transportation affects the quality of our daily lives. It influences incomplete burning of carbon in fuels. The NAAQS standard
our economic conditions, safety, accessibility, and capability to for ambient CO specifies upper limits for both one-hour and
reach people and places. Efficient and safe transportation eight-hour averages that should not be exceeded more than once
satisfies us all but in contrast, the inefficient and safe use of our per year. The level for one-hour standard is 35 parts per million
transportation system and facilities which result in traffic (ppm), and for the eight-hour standard is 9 ppm. Most
congestions and polluted air produces personal frustration and information and trends focus on the 8-hour average because it is
great economic loss. the more restrictive limit (EPA 1990).
The hazardous air pollutants come from both mobile and

stationary sources. Mobile sources include passenger cars, light 7.3.2 Air Quality Impacts of Transportation
and heavy trucks, buses, motorcycles, boats, and aircraft. Control Measures
Stationary sources range from oil refineries to dry cleaners and
iron and steel plants to gas stations. Some of the measures available for reducing traffic congestion
and improving mobility and air quality is documented in a report
This section concentrates on mobile source air pollutants which prepared by Institute of Transportation Engineers (ITE) in 1989.
comprise more than half of the U.S. air quality problems. This "toolbox" cites, as a primary cause of traffic congestion, the
Transportation and tailpipe control measure programs in increasing number of individuals commuting by automobile in
addition to highway air quality models are discussed in the metropolitan areas, to and from locations dispersed throughout
section. a wide region, and through areas where adequate highway
capacity does not exist. The specific actions that can be taken to
Under the 1970 U.S. Clean Air Act, each state must prepare a improve the situation are categorized under five components as
State Implementation Plan (SIP) describing how it will controll follows:
emissions from mobile and stationary sources to meet the
requirements of the National Ambient Air Quality Standards 1) Getting the most out of the existing highway system
(NAAQS) for six pollutants: (1) particulate matter (formerly - Intelligent Transportation Systems (ITS)
known as total suspended particulate (TSP) and now as PM10 - urban freeways (ramp metering, HOV's)
which emphasizes the smaller particles), (2) sulfur dioxide - arterial and local streets (super streets, parking
(SO2), (3) carbon monoxide (CO), (4) nitrogen dioxide (NO2), management)
(5) ozone (O3), and (6) lead (Pb). - enforcement
2) Building new capacity (new highway, reconstruction)
The 1990 U.S. Clean Air Act requires tighter pollution standards 3) Providing transit service (paratransit service, encouraging
especially for cars and trucks and would empower the transit use)
Environmental Protection Agency (EPA) to withhold highway 4) Managing transportation demand
funds from states which fail to meet the standards for major air - strategic approaches to avoiding congestion (road
pollutants (USDOT 1990). The 1970 and 1990 federal emission pricing)
standards for motor vehicles are shown in Table 7.1. - mitigating existing congestion (ridesharing)
5) Funding and institutional measures
This section concentrates on mobile source air pollutants which - funding (fuel taxes, toll roads)
comprise more than half of the U.S. air quality problems. - institutional measures (transportation management
Exhaust from these sources contain carbon monoxide, volatile associations)
organic compounds (VOCs), nitrogen oxides, particulates, and
lead. The VOCs along with nitrogen oxides are the major
elements contributing to the formation of "smog".

75$)),& ,03$&7 02'(/6
Table 7.1
Federal Emission Standards
1970 Standards 1990 Standards

(grams/km) (grams/km)
Light Duty Vehicles1 (0-3, 340 Kgs)
Carbon Monoxide (CO) 2.11 2.11
Hydrocarbons (HC) 0.25 0.16
Oxides of Nitrogen (NOx) 0.62 0.252
Particulates 0.12 0.05
Light Duty Vehicles1 (1,700-2,600 Kgs)
1970 Standards 1990 Standards

(grams/km) (grams/km)
Light Duty Trucks (over 2,600 Kgs GVWR)

1
Light duty vehicles include light duty trucks.
2
The new emission standards specified in this table are for useful life of 5 years or 80,000 Kms whichever first occurs.
7.3.3 Tailpipe Control Measures The following set of emission control measures, if implemented,
has the potential to substantially decrease exhaust emissions
It is clear that in order to achieve the air quality standards and to from major air pollutants:
reduce the pollution from the motor vehicle emissions,
substantial additional emission reduction measures are essential. limiting gasoline volatility to 62.0 KPa (9.0 psi) RVP;
According to the U.S. Office of Technology Assessment (OTA), adopting "onboard" refueling emissions controls;
the mobile source emissions are even higher in most sources - "enhanced" inspection and maintenance programs;
responsible for 48 percent of VOCs in non-attainment U.S. cities requiring "onboard diagnostics" for emission control
in 1985, compared to other individual source categories (Walsh systems;
1989). adopting full useful life (160,000 kms) requirements;

75$)),& ,03$&7 02'(/6
requiring alternative fuel usage; - humidity

utilizing oxygenated fuels/reformulated gasoline - wind and temperature fluctuations
measures; traffic data
adopting California standards (Tier I and II). - traffic volume
- vehicle speed
Some of the measures recommended in California standards - vehicle length or type
include: improved inspection and maintenance such as site types
"centralized" inspection and maintenance programs, heavy duty - at-grade sites
vehicle smoke enforcement, establishing new diesel fuel quality - elevated sites
standards, new methanol-fueled buses, urban bus system - cut sites
electrification, and use of radial tires on light duty vehicles. period of measurement
In the case of diesel-fueled LDT's (0-3,750 lvw) and light-duty Some of these models which estimate the pollutant emissions
vehicles, before the model year 2004, the applicable standards from highway vehicles are discussed in more detail in the
for NOx shall be 0.62 grams/km for a useful life as defined following sections.
above.
7.3.4.1 UMTA Model

7.3.4 Highway Air Quality Models
The most simple of these air quality models is the one which
As discussed earlier, federal, state, and local environmental relates vehicular speeds and emission levels (USDOT 1985).
regulations require that the air quality impacts of transportation - The procedure is not elaborate but is a quick-response technique
related projects be analyzed and be quantified. For this purpose, for comparison purposes. This UMTA (now Federal Transit
the Federal Highway Administration (FHWA) has issued Administration) model contains vehicular emission factors
guidelines to ensure that air quality effects are considered during related to speed of travel for freeways and surfaced arterials.
planning, design, and construction of highway improvements, so
that these plans are consistent with State Implementation Plans The model uses a combination of free flow and restrained (peak
(SIPs) for achieving and maintaining air quality standards . period) speeds. It assumes that one-third of daily travel would
occur in peak hours of flow reflecting restrained (congested)
For example, the level of CO associated with a given project is speeds, while two-thirds would reflect free-flow speed
a highway - related air quality impact that requires evaluation. characteristics. For a complete table of composite emission
In general, it must be determined whether the ambient standards factors categorized by autos and trucks for two calendar years of
for CO (35 ppm for 1 hour and 9 ppm for 8 hours, not to be 1987 and 1995 refer to Characteristics of Urban Transportation
exceeded more than once per year) will be satisfied or exceeded System, U.S. Department of Transportation, Urban Mass
as a result of highway improvements. This requirement calls for Transportation Administration, October 1985.
estimating CO concentrations on both local and areawide scales.
7.3.4.2 CALINE-4 Dispersion Model
A number of methods of varying sophistication and complexity
are used to estimate air pollutant levels. These techniques
This line source air quality model has been developed by the
include simple line-source-oriented Gaussian models as well as
California Department of Transportation (FHWA 1984). It is
more elaborate numerical models (TRB 1981). The databases
based on the Gaussian diffusion equation and employs a mixing
used in most models could be divided into the following
zone concept to characterize pollutant dispersion over the
categories:
roadway.
meteorological data
The model assesses air quality impacts near transportation
- wind speed and direction
facilities given source strength, meteorology, and site geometry.
- temperature
CALINE-4 can predict pollution concentrations for receptors

75$)),& ,03$&7 02'(/6
located within 500 meters of the roadway. It also has special distance from link endpoints to stopline;
options for modeling air quality near intersections, street acceleration and deceleration times (ACCT, DCLT);
canyons, and parking facilities. idle times at front and end of queue;
cruise speed (SPD); and
CALINE-4 uses a composite vehicle emission factor in grams idle emission rate (EFI).
per vehicle-mile and converts it to a modal emission factor. The
Environmental Protection Agency (EPA) has developed a series The following computed variables are determined for each link
of computer programs, the latest of which is called Mobile4.1 from the input variables:
(EPA 1991), for estimating composite mobile emission factors
given average route speed, percent cold and hot-starts, ambient acceleration rate;
temperature, vehicle mix, and prediction year. These emission deceleration rate;
factors are based on vehicle distribution weighted by type, age, acceleration length;
and operation mode, and were developed from certification and deceleration length;
surveillance data, mandated future emission standards, and acceleration-speed product;
special emission studies. FTP-75 (BAG2) time rate emission factor;
acceleration emission factor;
Composite emission factors represent the average emission rate cruise emission factor;
over a driving cycle. The cycle might include acceleration, deceleration emission factor; and
deceleration, cruise, and idle modes of operation. Emission rates queue length (LQU=NDLA*VSP), where VSP is the "at
specific to each of these modes are called modal emission rest" vehicle spacing.
factors. The speed correction factors used in composite emission
factor models, such as MOBILE4, are derived from variable The cumulative emission profiles (CEP) for acceleration,
driving cycles representative of typical urban trips. The Federal deceleration, cruise, and idle modes form the basis for
Test Procedures (FTP) for driving cycle are the basis for most of distributing the emissions. These profiles are constructed for
these data. each intersection link, and represent the cumulative emissions
per cycle per lane for the dominant movement. The CEP is
Typical input variables for the CALINE-4 model are shown in developed by determining the time in mode for each vehicle
Table 7.2. In case of an intersection, the following assumptions during an average cycle/lane event multiplied by the modal
are made for determining emission factors: emission time rate and summed over the number of vehicles.
CALINE4 can predict concentrations of relatively inert
uniform vehicle arrival rate; pollutants such as carbon monoxide (CO), and other pollutants
constant acceleration and deceleration rates; constant time like nitrogen dioxide (NO2) and suspended particles.
rate of emissions over duration of each mode;
deceleration time rate of emissions equals 1.5 times the idle
rate; 7.3.4.3 Mobie Source Emission Factor Model
an "at rest" vehicle spacing of 7 meters; and
all delayed vehicles come to a full stop. MOBILE4.1 is the latest version of mobile source emission
factor model developed by the U.S. Environmental Protection
In addition to composite emission factor at 26 km/hr (EFL), the Agency (EPA). It is a computer program that estimates
following variables must be quantified for each intersection link: hydrocarbon (HC), carbon monoxide (CO), and oxides of
nitrogen (NOx) emission factors for gasoline-fueled and diesel-
arrival volume in vehicles per hour; fueled highway motor vehicles.
departure volume in vehicles per hour;
average number of vehicles per cycle per lane for the MOBILE4.1 calculates emission factors for eight vehicle types
dominant; movement in two regions (low- and high-altitude). Its emission estimates
average number or vehicles delayed per cycle per lane for depend on various conditions such as ambient temperature,
the dominant movement (NDLA); speed, and mileage accrual rates. MOBILE4.1 will estimate
emission factors for any calendar year between 1960 and 2020.

75$)),& ,03$&7 02'(/6
Table 7.2
Standard Input Values for the CALINE4
I. Site Variables
Temperature ©
Wind Speed (m/s)
Wind Direction (deg)
Directional Variability (deg)
Atmospheric Stability (F)
Mixing Height (m)
Surface Roughness (cm)
Settling Velocity (m/s)
Deposition Velocity (m/s)
Ambient Temperature ©
II. Link Variables
Traffic Volume (veh/hr)
Emission Factor (grams/veh-mile)
Height (m)
Width (m)
Link Coordinates (m)
III. Receptor Locations (m)
The 25 most recent model years are considered to be in Speed correction factors are used by the model to correct exhaust
operation in each calendar year. It is to be used by the states in emissions for average speeds other than that of the FTP (32
the preparation of the highway mobile source portion of the 1990 km/hr). MOBILE4.1 uses three speed correction models: low
base year emission inventories required by the Clean Air Act speeds (4-32 km/hr), moderate speeds (32-77 km/hr), and high
Amendments of 1990. speeds (77-105 km/hr). The pattern of emissions as a function
of vehicle speed is similar for all pollutants, technologies, and
MOBILE4.1 calculates emission factors for gasoline-fueled model year groups. Emissions are greatest at the minimum
light-duty vehicles (LDVs), light-duty trucks (LDTs), heavy-duty speed of 4 km/hr, decline relatively rapidly as speeds increase
vehicles (HDVs), and motorcycles, and for diesel LDVs, LDTs, from 4 to 32 km/hr, decline more slowly as speeds increase from
and HDVs. It also includes provisions for modeling the effects 32 to 77 km/hr, and then increase with increasing speed to the
of oxygenated fuels (gasoline-alcohol and gasoline- ether blends) maximum speed of 105 km/hr.
on exhaust CO emissions. Some of the primary input variables
and their ranges are discussed below.

75$)),& ,03$&7 02'(/6
The vehicle miles traveled (VMT) mix is used to specify the based on interpolation of the calendar year 1990 and 1991
fraction of total highway VMT that is accumulated by each of the MOBILE4.1 emission factors.
eight regulated vehicle types. The VMT mix is used only to
calculate the composite emission factor for a given scenario on One important determinant of emissions performance is the
the basis of the eight vehicle class-specific emission factors. mode of operation. The EPA's emission factors are based on
Considering the dependence of the calculated VMT mix on the testing over the FTP cycle, which is divided into three segments
annual mileage accumulation rates and registration distributions or operating modes: cold start, stabilized, and hot start.
by age, EPA expects that states develop their own estimates of Emissions generally are highest when a vehicle is in the cold-
VMT by vehicle type for specific highway facility, sub-zones, start mode: the vehicle, engine, and emission control equipment
time of day, and so on. are all at ambient temperature and thus not performing at
optimum levels. Emissions are generally somewhat lower in hot
Many areas of the country have implemented inspection and start mode, when the vehicle is not yet completely warmed up
maintenance (I/M) programs as a means of further reducing but has not been sitting idle for sufficient time to have cooled
mobile source air pollution. MOBILE4.1 has the capability of completely to ambient temperature. Finally, emissions generally
modeling the impact of an operating I/M program on the are lowest when the vehicle is operating in stabilized mode, and
calculated emission factors, based on user specification of has been in continuous operation long enough for all systems to
certain parameters describing the program to be modeled. Some have attained relatively stable, fully "warmed-up" operating
of the parameters include: temperatures.
program start year and stringency level; The EPA has determined through its running loss emission test
first and last model years of vehicles subject to programs that the level of running loss emissions depends on
program; several variables: the average speed of the travel, the ambient
program type (centralized or decentralized); temperature, the volatility (RVP) of the fuel, and the length of
frequency of inspection (annual or biennial); and the trip. "Trip length" as used in MOBILE4.1 refers to the
test types. duration of the trip (how long the vehicle has been traveling), not
on the distance traveled in the trip (how far the vehicle has been
MOBILE4.1 (EPA 1991) has the ability to model uncontrolled driven). Test data show that for any given set of conditions
levels of refueling emissions as well as the impacts of the (average speed, ambient temperature, and fuel volatility),
implementation of either or both of the major types of vehicle running loss emissions are zero to negligible at first, but increase
recovery systems. These include the "Stage II" (at the pump) significantly as the duration of the trip is extended and the fuel
control of vehicle refueling emissions or the "onboard" (on the tank, fuel lines, and engine become heated.
vehicle) vapor recovery systems (VRS).
The minimum and maximum daily temperatures are used in 7.3.4.4 MICRO2
MOBILE4.1 in the calculation of the diurnal portion of
evaporative HC emissions, and in estimating the temperature of MICRO2 is an air quality model which computes the air
dispensed fuel for use in the calculation of refueling emissions. pollution emissions near an intersection. The concentration of
The minimum temperature must be between -18 C to 38 C (0 F the pollutants in the air around the intersection is not computed.
and 100 F), and the maximum temperature must be between -12 In order to determine the pollution concentration, a dispersion
C to 49 C (10 F and 120 F) inclusive. model which takes weather conditions such as wind, speed, and
direction into account should be used (Richards 1983).
The value used for calendar year in MOBILE4.1 defines the year
for which emission factors are to be calculated. The model has MICRO2 bases its emissions on typical values of the FTP
the ability to model emission factors for the year 1960 through performed for Denver, Colorado in the early 1980s. They are:
2020 inclusive. The base year (1990) inventories are based on
a typical day in the pollutant season, most commonly Summer FTP (1) HC 6.2 grams/veh/km
for ozone and Winter for CO. The base year HC inventories are FTP (2) CO 62.2 grams/veh/km

75$)),& ,03$&7 02'(/6
FTP (3) NOx 1.2 grams/veh/km Mean Speed NOx HC

(km/hr) (ppm) (ppm)
For lower than Denver altitudes or years beyond 1980's, 20 0.035 0.205
emission rates may be lower and should change from these initial
30 0.050 0.240
values.
40 0.070 0.260
The emission formulas as a function of acceleration and speeds 50 0.085 0.280
are as follow: 60 0.105 0.290
70 0.120 0.305
HC Emission (gram/sec) =
0.018 + 5.668*10-3 (A*S) + 2.165*10-4 (A*S2) (7.17a)
A quick but less accurate estimate of the annual maximum 8
hour CO concentration can also be obtained from the average
CO Emission (gram/sec) =
peak hour CO concentration estimate, as follows:
0.182 - 8.587*10-2 (A*S) + 1.279*10-2 (A*S2) (7.17b)
C8 = 1.85 C1 + 1.19 (7.19)
NOx Emission (gram/sec) =
3.86*10-3 + 8.767*10-3(A*S) (for A*S>0) (7.17c)
where C8 is the annual maximum 8 hour concentration, and C1
is the average peak hour concentration.
NOx Emission (gram/sec) =
1.43*10-3 - 1.830*10-4(A*S) (for A*S<0) (7.17d) A graphical screening test is introduced by which any properties
likely to experience an air pollution problem are identified. The
where, procedure first reduces the network to a system of long roads and
A = acceleration (meters/sec2) and roundabouts (if any). Then from a graph, the concentration of
S = speed (meters/sec). carbon monoxide for standard traffic conditions for locations at
any distance from each network element may be determined.
Factors are then applied to adjust for the traffic conditions at the
7.3.4.5 The TRRL Model site and the sum of the contributions from each element gives an
estimate of the likely average peak hour concentration. An
This model has been developed by the British Transport and example of the graphical screening test results is shown in Table
Road Research Laboratory (TRRL) and it predicts air pollution 7.3.
from road traffic (Hickman and Waterfield 1984). The
estimations of air pollution are in the form of hourly average
concentrations of carbon monoxide at selected locations around
a network of roads. The input data required are the
7.3.5 Other Mobile Source
configuration of the road network, the location of the receptor, Air Quality Models
traffic volumes and speeds, wind speed, and wind direction.
There are many other mobile source models which estimate the
The concentration of carbon monoxide may be used as to pollutant emission rates and concentrations near highway and
approximate the likely levels of other pollutants using the arterial streets. Most of these models relate vehicle speeds and
following relations: other variables such as vehicle year model, ambient temperature,
and traffic conditions to emission rates. A common example of
HC (ppm) = 1.8 CO (ppm) * R + 4.0 (7.18a) this type of relation could be found in Technical Advisory
#T6640.10 of EPA report, "Mobile Source Emission Factor
NOx (ppm) = CO (ppm) * R + 0.1 (7.18b) Tables for MOBILE3." Other popular models include HIWAY2
and CAL3QHC. HIWAY2 model has been developed by U.S.
where R is the ratio of pollutant emission rate to that of carbon EPA to estimate hourly concentrations of non-reactive
monoxide for a given mean vehicle speed, pollutants, like CO, downwind of roadways. It is usually used

75$)),& ,03$&7 02'(/6
Table 7.3
Graphical Screening Test Results for Existing Network
Distance from Centerline (m) 13 45 53 103
CO for 1000 vehs/hr at 100 km/h (ppm) 1.08 0.49 0.40 0.11
Traffic Flow (vehs/hr) 2,600 1,400 800 400
Speed (km/hr) 40 20 20 50
Speed Correction Factor 2.07 3.59 3.59 1.73
CO for Actual Traffic Conditions (ppm) 5.81 2.46 1.15 0.08
Total 1-Hour CO = 9.50(ppm)

Equivalent Annual Maximum 8-Hour = 18.76(ppm)
for analyzing at- grade highways and arterials in uniform wind is especially designed to handle near-saturated and/or over-
conditions at level terrain as well as at depressed sections (cuts) capacity traffic conditions and complex intersections where
of roadways. The model cannot be used if large obstructions major roadways interact through ramps and elevated highways.
such as buildings or large trees hinder the flow of air. The The model combines the CALINE3 line source dispersion model
simple terrain requirement makes this model less accurate for with an algorithm that internally estimates the length of the
urban conditions than CALINE4 type of models. queues formed by idling vehicles at signalized intersections. The
inputs to the model includeinformation and data commonly
The CAL3QHC model has the ability to account for the required by transportation models such as roadway geometries,
emissions generated by vehicles traveling near roadway receptor locations, vehicular emissions, and meteorological
intersections. Because idling emissions account for a substantial conditions. Emission factors used in the model should be
portion of the total emissions at an intersection, this capability obtained from mobile source emission factor models such as
represents a significant improvement in the prediction of MOBILE4.
pollutant concentrations over previous models. This EPA model
References
Akcelik, R. (1981). Fuel Efficiency and Other Objectives in Andreassen, D. (1991). Population and Registered Vehicle
Traffic System Management. Traffic Engineering & Data Vs. Road Deaths. Accident Analysis & Prev., Vol. 23,
Control, Vol. 22, pp. 54-65. No. 5, pp. 343-351.
Akcelik, R. And C. Bayley (1983). Some Results on Fuel Belmont, D. M. (1953). Effect of Average Speed And Volume
Consumption Models. Appeared in Progress in Fuel On Motor-vehicle Accidents On Two-lane Tangents. Proc.
Consumption Modelling for Urban Traffic Management. Highway Research Board, 32, pp. 383-395.
Research Report ARR No. 124, Austrailian Road Research
Board, pp. 51-56.

75$)),& ,03$&7 02'(/6
Biggs, D. C. and R. Akcelik (1986). An Energy-Related Federal Highway Administration (1984). CALINE4 - A
Model for Instantaneous Fuel Consumption. Traffic Dispersion Model for Predicting Air Pollutant
Engineering & Control, Vol.27, No.6, pp. 320-325. Concentrations Near Roadways. Report FHWA/CA/TL -
Bonneson, J. A., and P. T. McCoy (1993). Estimation 84/15.
of Safety At Two-way Stop-controlled Intersections On Fisk, C. S. (1989). The Australian Road Research Board
Rural Highways. Transportation Research Record 1401, pp. Instantaneous Model of Fuel Consumption. Transportation
83-89, Washington DC. Research, Vol. 23B, No. 5, pp. 373-385.
Brundell-Freij, K. and L. Ekman (1991). Flow and Safety. Gipps, P. G. and B. G. Wilson (1980). MULTISIM: A
Presented at the 70th Annual Meeting of the Transportation Computer Package for Simulating Multi-Lane Traffic
Research Board, Washington, DC. Flows. Proceedings of the 4th Biennial Conference of
Brüde, U. and J. Larsson (1993). Models For Predicting Simulation Society of Australia.
Accidents At Junctions Where Pedestrians And Cyclists Are Greene, D. L. and J. T. Liu (1988). Automotive Fuel Economy
Involved. How Well Do They Fit? Accident Analysis & Improvements and Consumers' Surplus. Transportation
Prev. Vol. 25, No. 6, pp. 499-509. Research, Vol. 22A, No. 3.
Ceder, A. and M. Livneh (1978). Further Evaluation Of The Greene, D. L. and K. G. Duleep (1993). Costs and Benefits of
Relationships Between Road Accidents and Average Daily Automotive Fuel Economy Improvement: A Partial
Traffic. Accident Analysis & Prev., Vol 10, pp. 95-109. Analysis. Transportation Research, Vol. 27A, No.3.
Ceder, A. (1982). Relationship Between Road Accidents And Hauer, E. (1971). Accident Overtaking and Speed Control.
Hourly Traffic Flow-II. Accident Analysis & Prev., Accident Analysis and Prevention, Vol. 3, No. 1, pp. 1-13.
Vol 14, No. 1, pp. 35-44. Hauer, E. (1992). Empirical Bayes Approach To The
Ceder, A. and M. Livneh (1982). Relationship Between Estimation Of "Unsafety"; The Multivariate Regression
Road Accidents And Hourly Traffic Flow-I. Accident Method. Accident Analysis & Prev., Vol. 24, No. 5,
Analysis & Prev., Vol. 14, No. 1. pp. 457-477.
Chang, M. F., L. Evans, R. Herman, and P. Wasielewski Hauer, E., D. Terry, and M. S. Griffith (1994). The Effect Of
(1976). Gasoline Consumption in Urban Traffic. Resurfacing On The Safety Of Rural Roads In New York
Transportation Research Record 599, Transportation State. Paper 940541 presented at the 73rd Annual Meeting
Research Board, pp. 25-30. of the Transportation Research Board, Washington, DC.
Chang, M. F. and R. Herman (1981). Trip Time Versus Stop Hall, J. W. And O. J. Pendleton (1990). Rural Accident Rate
Time and Fuel Consumption Characteristics in Cities. Variation with Traffic Volume. Transportation Research
Transportation Science, Vol. 15, No. 3, pp. 183-209. Record 1281, TRB, NRC, Washington, DC, pp. 62-70.
Cleveland, D. E., L. P. Kostyniuk, and K.-L. Ting (1985). Herman, R. and S. Ardekani (1985). The Influence of Stops on
Design and Safety On Moderate Volume Two-lane Roads. Vehicle Fuel Consumption in Urban Traffic. Transportation
Transportation Research Record, 1026, pp. 51-61. Science, Vol. 19, No. 1, pp. 1-12.
DeLuchi, M., Q. Wang, and D. Sperling (1989). Electric Hickman, A. J. and V. H. Waterfield (1984). A User's
Vehicles: Performance, Life-Cycle Costs, Emissions, and Guide to the Computer Programs for Predicting Air
Recharging Requirements. Transportation Research, Pollution from Road Traffic. TRRL Supplementary Report
Vol. 23A, No. 3. 806.
Evans, L., R. Herman, and T. Lam (1976). Multivariate Hickman, A. J. and M. H. L. Waters (1991). Improving
Analysis of Traffic Factors Related to Fuel Consumption in Automobile Fuel Economy. Traffic Engineering & Control,
Urban Driving. Transportation Science, Vol. 10, No. 2, pp. Vol. 23, No. 11.
205-215. Hooker, J. N., A. B. Rose, and G. F. Roberts (1983). Optimal
Evans, L. and R. Herman (1978). Automobile Fuel Economy Control of Automobiles for Fuel Economy. Transportation
on Fixed Urban Driving Schedules. Transportation Science, Science, Vol. 17, No. 2, pp. 146-167.
Vol. 12, No. 2, pp. 137-152. Institute of Transportation Engineers (1989). A Toolbox for
Everall, P. F. (1968). The Effect of Road and Traffic Alleviating Traffic Congestion. Publication No. IR.054A.
Conditions on Fuel Consumption. RRL Report LR 226,
Road Research Laboratory, Crowthorne, England.

75$)),& ,03$&7 02'(/6
Jovanis, P. P. and H-L. Chang (1987). Modelling The Ontario Ministry of Transportation and
Relationship Of Accidents To Miles Travelled. Communications (1982). A Technical Background
Transportation Research Record 1068, Washington, DC, Document For Automotive Fuel Economy. The
pp. 42-51. Transportation Energy Management Program (TEMP)
Kenworthy, J. R., H. Rainford, P. W. G. Newman, and T. J. Report No. DRS-82-01, Ontario, Canada.
Lyons (1986). Fuel Consumption, Time Saving and Pelensky, E., W. R. Blunden, and R. D. Munro (1968).
Freeway Speed Limits. Traffic Engineering & Control, Vol. Operating Costs of Cars in Urban Areas. Proceedings,
27, No. 9, pp. 455-459. Fourth Conference of the Australian Road Research Board,
Kihlberg, J. K. and K. J. Tharp (1968). Accident Rates Vol. 4, part 1, pp. 475-504.
As Related To Design Elements Of Rural Highways. Persaud B. and Dzbik, L. (1993). Accident Prediction Models
National Cooperative Highway Research Program Report, For Freeways. Transportation Research Record 1401, TRB,
47, Highway Research Board. NRC, Washington, DC, pp. 55-60.
Komor, P., S. F. Baldwin, and J. Dunkereley (1993). Pfundt, K. (1968). Comparative Studies Of Accidents
Technologies for Improving Transportation Energy On Rural Roads. Strassenbau und Strassenverkehrstechnik,
Efficiency in the Developing World. Transportation Vol. 82, Federal Ministry of Transport, Bonn.
Research, Vol 27A, No. 5. Pfundt, K. (1969). Three Difficulties In The Comparison Of
Lam, T. N. (July 1985). Estimating Fuel Consumption Accident Rates. Accident Analysis & Prevention, Vol. 1, pp.
From Engine Size. Journal of Transportation Engineering, 253-259.
Vol. 111, No. 4, pp. 339-357. Pitt, D. R., T. J. Lyons, P. W. G. Newman, and
Leutzbach, W., W. Siegener, and R. Wiedemann (1970). J. R. Kenworthy (1987). Fuel Consumption Models: An
Über den Zusammenhang Zwischen Verkehrsunfällen Und Evaluation Based on a Study of Perth's Traffic Patterns .
Verkehrsbelastung Auf Eienem deutschen Traffic Engineering & Control, Vol. 27, No. 2, pp. 60-68.
Autobahnabschnitt (About the Relationship between Traffic Post, K., J. Tomlin, D. Pitt, N. Carruthers, A. Maunder,
Accidents and Traffic Flow on a Section of a German J. H. Kent, and R. W. Bilger (1981). Fuel Economy and
Freeway). Accident Analysis & Prev. Emissions Research Annual Report. Charles Kolling
Vol. 2, pp. 93-102. Research Laboratory Technical Note: ER36, University of
Lieberman, E., R. D. Worrall, D. Wicks, and J. Woo (1979). Sydney.
NETSIM Model (5 Vols.). Federal Highway Administration, Quaye, K., L. Leden, and E. Hauer (1993). Pedestrian
Report No. FHWA-RD-77-41 to 77-45, Washington, DC. Accidents And Left-turning Traffic At Signalized
Lundy, R. A. (1965). Effect of Traffic Volumes And Number Intersections. AAA foundation for Traffic Safety,
Of Lanes On Freeway Accident Rates. Highway Research Washington, DC.
Record 99, HRB, National Research Council, Washington, Richards, G. G. (1983). MICRO2 - An Air Quality
DC, pp. 138-147. Intersection Model. Colorado Department of Highways.
Mahalel, D. (1986). A Note on Accident Risk. Transportation Richardson, A. J. and R. Akcelik (1983). Fuel Consumption
Research Record, 1068, National Research Council, and Data Needs for the Design and Evaluation of Urban
Washington, DC, pp. 85-89. Traffic Systems. Research Report ARR No. 124, Austrailian
McDonald, J. W. (1953). Relationship between Number of Road Research Board, pp. 51-56.
Accidents and Traffic Volume at Divided-Highway Roosmark, P.O. (1966). Trafikolyckor i trevagskorsningar.
Intersections. Bulletin No. 74, pp. 7-17. Statens Vaginstitut, Preliminary Report 22, Stockholm.
Miaou, S-P. and H. Lum (1993). Modeling Vehicle Accidents Satterthwaite, S. P. (1981). A Survey into Relationships
And Highway Geometric Design Relationships. Accident between Traffic Accidents and Traffic Volumes. Transport
Analysis & Prev. Vol. 25, No. 6, pp. 689-709. and Road Research Laboratory Supplementary Report, SR
Ng, J. C. N. and E. Hauer (1989). Accidents on Rural 692, pp. 41.
Two-Lane Roads: Differences Between Seven States. Slatterly, G. T. and D. E. Cleveland (1969). Traffic Volume.
Transportation Research Record, 1238, Traffic Control And Roadway Elements. Chapter 2,
pp. 1-9. Automotive Safety Foundation, pp. 8.

75$)),& ,03$&7 02'(/6
Tanner, J. C. (1953). Accidents At Rural Three-way Wallace, C. E., K. G. Courage, D. P. Reaves, G. W. Schoene,
Junctions. J. Instn. of Highway Engrs. II(11), pp. 56-67. and G. W. Euler (1984). TRANSY-7F User's Manual. U.S.
Thorson, O. (1967). Traffic Accidents and Road Layout. Dept. of Transportation, Office of Traffic Operations.
Copenhagen Technical University of Denmark. Walsh, M. P. (1989). Statement Before the U.S. Senate
Transportation Research Board (1981). Methodology for Environment and Public Works Committee.
Evaluating Highway Air Pollution Dispersion Models. Watson, H. C., E. E. Milkins, and G. A. Marshal (1980). A
NCHRP Report No. 245, National Research Council. Simplified Method for Quantifying Fuel Consumption of
U.S. DOT (1990). A Statement of National Transportation Vehicles in Urban Traffic. SAE-Australia, Vol. 40, No. 1,
Policy, Strategies for Action, Moving America - New pp. 6-13.
Directions, New Opportunities. Webb, G. M. (1955). The Relationship Between Accidents
Urban Mass Transportation Administration (1985). and Traffic Volumes at Signalized Intersections. Institute of
Characteristics of Urban Transportation System. U.S. Traffic Engineers. Proceedings.
Department of Transportation. Welbourne, E. R. (1979). Exposure Effects on the Rate
U. S. EPA (1990). National Air Quality and Emissions and Severity of Highway Collisions at Different Travel
Trends Report. Office of Air Quality. Speeds. Technical Memorandum TMVS 7903, Vehicle
U. S. EPA (1991). MOBILE4.1 User's Guide - Mobile Systems, Transport Canada.
Source Emission Factor Model. Report EPA-AA-TEB-91- Zegeer, C. V., J. Hummer, D. Reinfurt, L. Herf, and W.
01. Hunter (1986). Safety Effects of Cross-Section Design for
Vincent, R. A., A. I. Mitchell, and D. I. Robertson (1980). Two-Lane Roads. Volume I, Final Report, FHWA-RD-
User Guide to TRANSYT Version 8. Transport and Road 87/008, Federal Highway Administration, Transportation
Research Lab Report No. LR888. Research Board.

UNSIGNALIZED INTERSECTION THEORY
BY ROD J. TROUTBECK13
WERNER BRILON14
13
Professor, Head of the School, Civil Engineering, Queensland University of Technology, 2 George Street,
Brisbane 4000 Australia.
14
Professor, Institute for Transportation, Faculty of Civil Engineering, Ruhr University, D 44780 Bochum,
Germany.
bi = proportion of volume of movement i of the total volume on the shared lane

Cw = coefficient of variation of service times
D = total delay of minor street vehicles
Dq = average delay of vehicles in the queue at higher positions than the first
E(h) = mean headway
E(tc ) = the mean of the critical gap, tc
f(t) = density function for the distribution of gaps in the major stream
g(t) = number of minor stream vehicles which can enter into a major stream gap of size, t
L = logarithm
m = number of movements on the shared lane
n = number of vehicles
c = increment, which tends to 0, when Var(tc) approaches 0
f = increment, which tends to 0, when Var(tf ) approaches 0
q = flow in veh/sec
qs = capacity of the shared lane in veh/h
qm,i = capacity of movement i, if it operates on a separate lane in veh/h
qm = the entry capacity
qm = maximum traffic volume departing from the stop line in the minor stream in veh/sec
qp = major stream volume in veh/sec
t = time
tc = critical gap time
tf = follow-up times
tm = the shift in the curve
Var(tc) = variance of critical gaps
Var(tf ) = variance of follow-up-times
Var (W) = variance of service times
W = average service time. It is the average time a minor street vehicle spends in the first position of the queue near the intersection
W1 = service time for vehicles entering the empty system, i.e no vehicle is queuing on the vehicle's arrival
W2 = service time for vehicles joining the queue when other vehicles are already queuing
8.
UNSIGNALIZED INTERSECTION THEORY
8.1 Introduction
Unsignalized intersections are the most common intersection driver and the pattern of the inter-arrival times are
type. Although their capacities may be lower than other important.
intersection types, they do play an important part in the control
of traffic in a network. A poorly operating unsignalized This chapter describes both of these aspects when there are two
intersection may affect a signalized network or the operation of streams. The theory is then extended to intersections with more
an Intelligent Transportation System. than two streams.
The theory of the operation of unsignalized intersections is

fundamental to many elements of the theory used for other 8.1.2 Interaction of Streams at
intersections. For instance, queuing theory in traffic engineering
Unsignalized Intersections
used to analyze unsignalized intersections is also used to analyze
other intersection types.
A third requirement at unsignalized intersections is that the
interaction between streams be recognized and respected. At all
unsignalized intersections there is a hierarchy of streams. Some
8.1.1 The Attributes of a Gap streams have absolute priority, while others have to yield to
Acceptance Analysis Procedure higher order streams. In some cases, streams have to yield to
some streams which in turn have to yield to others. It is useful
Unsignalized intersections give no positive indication or control to consider the streams as having different levels of priority or
to the driver. He or she is not told when to leave the intersection. ranking. For instance:
The driver alone must decide when it is safe to enter the
intersection. The driver looks for a safe opportunity or "gap" in Rank 1 stream - has absolute priority and does not need to
the traffic to enter the intersection. This technique has been yield right of way to another stream,
described as gap acceptance. Gaps are measured in time and are
equal to headways. At unsignalized intersections a driver must Rank 2 stream - has to yield to a rank 1 stream,
also respect the priority of other drivers. There may be other
vehicles that will have priority over the driver trying to enter the Rank 3 stream - has to yield to a rank 2 stream and in turn to
traffic stream and the driver must yield to these drivers. a rank 1 stream, and
All analysis procedures have relied on gap acceptance theory to Rank 4 stream - has to yield to a rank 3 stream and in turn to
some extent or they have understood that the theory is the basis a rank 2 stream and to a rank 1 stream.
for the operation even if they have not used the theory explicitly.
Although gap acceptance is generally well understood, it is 8.1.3 Chapter Outline

useful to consider the gap acceptance process as one that has two
basic elements. Sections 8.2 discusses gap acceptance theory and this leads to
Section 8.3 which discusses some of the common headway
First is the extent drivers find the gaps or opportunities distributions used in the theory of unsignalized intersections.
of a particular size useful when attempting to enter the
intersection. Most unsignalized intersections have more than two interacting
Second is the manner in which gaps of a particular size are streams. Roundabouts and some merges are the only examples
made available to the driver. Consequently, the proportion of two interacting streams. Nevertheless, an understanding of
of gaps of a particular size that are offered to the entering the operation of two streams provides a basis to extend the
knowledge to intersections with more than two streams. Section
8-1
8.4 discusses the performance of intersections with two The theory described in this chapter is influenced by the human
interacting streams. factors and characteristics as described in Chapter 3, and in
particular, Sections 3.13 and 3.15. The reader will also note
Section 8.5 to 8.8 discuss the operation of more complex similarities between the material in this chapter and Chapter 9
intersections. Section 8.9 covers other theoretical treatments of dealing with signalized intersections. Finally, unsignalized
unsignalized intersections. In many cases, empirical approaches intersections can quickly become very complicated and often the
have been used. For instance the relationships for AWSC (All subject of simulation programs. The comments in Chapter 10
Way Stop Controlled) intersections are empirical. The time are particularly relevant here.
between successive departures of vehicles on the subject
roadway are related to the traffic conditions on the other roadway
elements.
8.2 Gap Acceptance Theory

8.2.1 Usefulness of Gaps the remaining time is considered 'useable.' This 'useable' time
divided by the saturation flow gives an estimate of the absorption
The gap acceptance theory commonly used in the analysis of capacity of the minor stream. As shown below, the effect of this
unsignalized intersections is based on the concept of defining the different concept is negligible.
extent drivers will be able to utilize a gap of particular size or
duration. For instance, will drivers be able to leave the stop line In the theory used in most guides for unsignalized intersections
at a minor road if the time between successive vehicles from the around the world, it is assumed that drivers are both consistent
left is 10 seconds; and, perhaps how many drivers will be able and homogeneous. A consistent driver is expected to behave the
to depart in this 10 second interval ? same way every time at all similar situations. He or she is not
expected to reject a gap and then subsequently accept a smaller
The minimum gap that all drivers in the minor stream are gap. For a homogeneous population, all drivers are expected to
assumed to accept at all similar locations is the critical gap. behave in exactly the same way. It is, of course, unreasonable to
According to the driver behavior model usually assumed, no expect drivers to be consistent and homogeneous.
driver will enter the intersection unless the gap between vehicles
in a higher priority stream (with a lower rank number) is at least The assumptions of drivers being both consistent and
equal to the critical gap, tc. For example, if the critical gap was homogeneous for either approach are clearly not realistic.
4 seconds, a driver would require a 4 second gap between Rank Catchpole and Plank (1986), Plank and Catchpole (1984),
1 stream vehicles before departing. He or she will require the Troutbeck (1988), and Wegmann (1991) have indicated that if
same 4 seconds at all other times he or she approaches the same drivers were heterogeneous, then the entry capacity would be
intersection and so will all other drivers at that intersection. decreased. However, if drivers are inconsistent then the capacity
would be increased. If drivers are assumed to be both consistent
Within gap acceptance theory, it is further assumed that a and homogeneous, rather than more realistically inconsistent and
number of drivers will be able to enter the intersection from a heterogeneous, then the difference in the predictions is only a
minor road in very long gaps. Usually, the minor stream few percent. That is, the overall effect of assuming that drivers
vehicles (those yielding right of way) enter in the long gaps at are consistent and homogeneous is minimal and, for simplicity,
headways often referred to as the "follow-up time", tf . consistent and homogeneous driver behavior is assumed.
Note that other researchers have used a different concept for the It has been found that the gap acceptance parameters tc and tf
critical gap and the follow-up time. McDonald and Armitage may be affected by the speed of the major stream traffic (Harders
(1978) and Siegloch (1973) independently described a concept 1976 and Troutbeck 1988). It also expected that drivers are
where a lost time is subtracted from each major stream gap and influenced by the difficulty of the maneuver. The more difficult
8-2
a maneuver is, the longer are the critical gap and follow-up time distribution of follow-up times and the critical gap distribution
parameters. There has also been a suggestion that drivers independently. Each group is discussed below.
require a different critical gap when crossing different streams
within the one maneuver. For instance a turn movement across Regression techniques.
a number of different streams may require a driver having a If there is a continuous queue on the minor street, then the
different critical gap or time period between vehicles in each technique proposed by Siegloch (1973) produces acceptable
stream (Fisk 1989). This is seen as a unnecessary complication results because the output matches the assumptions used in a
given the other variables to be considered. critical gap analysis. For this technique, the queue must have at
least one vehicle in it over the observation period. The process
is then:
8.2.2 Estimation of the Critical
Gap Parameters Record the size of each gap, t, and the number of
vehicles, n, that enter during this gap;
The two critical gap parameters that need to be estimated are the
critical gap tc and the follow-up time tf . The techniques used to
For each of the gaps that were accepted by only n
estimate these parameters fit into essentially two different drivers, calculate the average gap size, E(t) (See Figure
groups. The first group of techniques are based on a regression 8.1);
analysis of the number of drivers that accept a gap against
the gap size. The other group of techniques estimates the Use linear regression on the average gap size values
(as the dependent variable) against the number of
vehicles that enter during this average gap size, n; and
Figure 8.1
Data Used to Evaluate Critical Gaps and Move-Up Times
(Brilon and Grossmann 1991).
8-3
Given the slope is tf and the intercept of the gap size these minor street vehicles depart at 3.99, 6.22, 8.29, 11.13,
axis is to, then the critical gap tc is given by 13.14, and so on, then the headways between the minor street
vehicles are 6.22-3.99, 8.29-6.22, 11.13-8.29 and so on. The
average headway between this group of minor stream
tc
to tf /2 (8.1) vehicles is 2.33 sec. This process is repeated for a number of
larger major stream gaps and an overall average headway
between the queued minor stream vehicles is estimated. This
The regression line is very similar to the stepped line as shown average headway is the follow-up time, tf . If a minor stream
in Figure 8.2. The stepped line reflects the assumptions made by vehicle was not in a queue then the preceding headway would
Tanner (1962), Harders (1976), Troutbeck (1986), and others. not be included. This quantity is similar to the saturation
The sloped line reflects the assumptions made by Siegloch headway at signalized intersections.
(1973), and McDonald and Armitage (1978).
The estimation of the critical gap is more difficult. There have
Independent assessment of the critical gap and follow-up time been numerous techniques proposed (Miller 1972; Ramsey and
If the minor stream does not continuously queue, then the Routledge 1973; Troutbeck 1975; Hewitt 1983; Hewitt 1985).
regression approach cannot be used. A probabilistic approach The difficulty with the estimation of the critical gap is that it
must be used instead. cannot be directly measured. All that is known is that a driver's
individual critical gap is greater than the largest gap rejected and
The follow-up time is the mean headway between queued shorter than the accepted gap for that driver. If the accepted gap
vehicles which move through the intersection during the longer was shorter than the largest rejected gap then the driver is
gaps in the major stream. Consider the example of two major considered to be inattentive. This data is changed to a value just
stream vehicles passing by an unsignalized intersection at times below the accepted gap. Miller (1972) gives an alternative
2.0 and 42.0 seconds. If there is a queue of say 20 vehicles method of handling this inconsistent data which uses the data as
wishing to make a right turn from the side street, and if 17 of recorded. The difference in outcomes is generally marginal.
Figure 8.2
Regression Line Types.
8-4
Miller (1972), and later Troutbeck (1975) in a more limited μ and

study, used a simulation technique to evaluate a total of ten 2 are the mean and variance of the logarithm of the
different methods to estimate the critical gap distribution of individual drivers critical gaps (assuming a log-
drivers. In this study the critical gaps for 100 drivers were normal distribution), and
defined from a known distribution. The arrival times of priority f( ) and
traffic were simulated and the appropriate actions of the F( ) are the probability density function and the
"simulated" drivers were noted. This process was repeated for cumulative distribution function respectively for
100 different sets of priority road headways, but with the same the normal distribution.
set of 100 drivers. The information recorded included the size
of any rejected gaps and the size of the accepted gap and would The probability that an individual driver's critical gap will be
be similar to the information able to be collected by an engineer between ri and ai is F(ai) – F(ri). Summing over all drivers, the
at the road side. The gap information was then analyzed using likelihood of a sample of n drivers having accepted and largest
each of the ten different methods to give an estimate of the rejected gaps of (ai, ri) is
average of the mean of the drivers' critical gaps, the variance of
N [F(a ) F(r )]
n
the mean of the drivers' critical gaps, mean of the standard (8.2)
i i
deviation of the drivers' critical gaps and the variance of the i
1
standard deviation of the drivers' critical gaps. These statistics
enabled the possible bias in predicting the mean and standard
deviation of the critical gaps to be estimated. Techniques which The logarithm, L, of this likelihood is then
gave large variances of the estimates of the mean and the
M ln[F(a ) F(r )]
n
standard deviation of the critical gaps were considered to be less L
i i (8.3)
reliable and these techniques were identified. This procedure i
1
found that one of the better methods is the Maximum Likelihood
Method and the simple Ashworth (1968) correction to the
prohibit analysis being a strong alternative. Both methods are The maximum likelihood estimators, μ and 2, that maximize L,
documented here. The Probit or Logit techniques are also are given by the solution to the following equations.
acceptable, particularly for estimating the probability that a gap
L
will be accepted (Abou-Henaidy et al. 1994), but more care
0 (8.4)
needs to be taken to properly account for flows. Kyte et al μ
(1996) has extended the analysis and has found that the
Maximum Likelihood Method and the Hewitt (1983) models
gave the best performance for a wide range of minor stream and and
major stream flows.
L

0 (8.5)
The maximum likelihood method of estimating the critical gap 2
requires that the user assumes a probabilistic distribution of the
critical gap values for the population of drivers. A log-normal is
a convenient distribution. It is skewed to the right and does not Using a little algebra,
have non-negative values. Using the notation:
F(x)

f(x) (8.6)
ai = the logarithm of the gap accepted by the ith driver, μ
ai = if no gap was accepted,
ri = the logarithm of the largest gap rejected by the ith
driver,
ri = 0 if no gap was rejected,
F(x)

x μ2 f(x) (8.7)
2
2
8-5
This then leads to the following two equations which must be first gap offered without rejecting any gaps, then Equations 8.8
solved iteratively. It is recommended that the equation and 8.9 give trivial results. The user should then look at
alternative methods or preferably collect more data.
f(r ) f(a )
M F(a
n
i

0 i
(8.8)
i
1 ) F(r )
i i Another very useful technique for estimating the critical gap is
Ashworth’s (1968) procedure. This requires that the user
identify the characteristics of the probability distribution that
should be used to estimate μ given a value of 2. An initial value relates the proportion of gaps of a particular size that were
of 2 is the variance of all the ai and ri values. Using this accepted to the gap size. This is usually done using a Probit
estimate of μ from Equation 8.8, a better estimate of 2 can be analysis applied to the recorded proportions of accepted gaps.
obtained from the equation, A plot of the proportions against the gap size on probability
paper would also be acceptable. Again a log normal distribution
M (r μ̂ F(af(r)) (aF(r μ̂)

n ) ) f(ai) may be used and this would require the proportions to be plotted
i i i

0 (8.9)
i
1 i i
against the natural logarithm of the gap size. If the mean and
variance of this distribution are E(ta) and Var(t a ), then
Ashworth’s technique gives the critical gap as
where μ̂ is an estimate of μ.
E(tc)
E(ta) qp Var(ta) (8.12)
A better estimate of the μ can then be obtained from the
Equation 8.8 and the process continued until successive
estimates of μ and 2 do not change appreciably.
where qp is the major stream flow in units of veh/sec. If the log
The mean, E(tc ), and the variance, Var(tc ), of the critical gap normal function is used, then E(ta) and Var(ta) are values given
distribution is a function of the log normal distribution by the generic Equations 8.10 and 8.11. This is a very practical
parameters, viz: solution and one which can be used to give acceptable results in
the office or the field.
E(tc)
e μ0.5
2
(8.10)
8.2.3 Distribution of Gap Sizes

and
The distribution of gaps between the vehicles in the different
Var(tc )
E(tc )2 (e 1)
2
(8.11) streams has a major effect on the performance of the
unsignalized intersection. However, it is important only to look
at the distribution of the larger gaps; those that are likely to be
The critical gap used in the gap acceptance calculations is then accepted. As the shorter gaps are expected to be rejected, there
equal to E(tc ). The value should be less than the mean of the is little point in modeling these gaps in great detail.
accepted gaps.
A common model uses a random vehicle arrival pattern, that is,
This technique is a complicated one, but it does produce the inter-arrival times follow an exponential distribution. This
acceptable results. It uses the maximum amount of information, distribution will predict a large number of headways less than 1
without biasing the result, by including the effects of a large sec. This is known to be unrealistic, but it is used because these
number of rejected gaps. It also accounts for the effects due to small gaps will all be rejected.
the major stream headway distribution. If traffic flows were
light, then many drivers would accept longer gaps without This exponential distribution is known to be deficient at high
rejecting gaps. On the other hand, if the flow were heavy, all flows and a displaced exponential distribution is often
minor stream drivers would accept shorter gaps. The recommended. This model assumes that vehicle headways are
distribution of accepted gaps is then dependent on the major at least tm sec.
stream flow. The maximum likelihood technique can account for
these different conditions. Unfortunately, if all drivers accept the
8-6
Better models use a dichotomized distribution. These models In this chapter, the word "queues" is used to refer to a line of
assume that there is a proportion of vehicles that are free of stopped vehicles. On the other hand, a platoon is a group of
interactions and travel at headways greater than tm. These traveling vehicles which are separated by a short headway of tm.
vehicles are termed "free" and the proportion of free vehicles is When describing the length of a platoon, it is usual to include a
. There is a probability function for the headways of free platoon leader which will have a longer headway in front of him
vehicles. The remaining vehicles travel in platoons and again or her. A platoon of length one is a single vehicle travelling
there is a headway distribution for these bunched vehicles. One without any vehicles close-by. It is often useful to distinguish
such dichotomized headway model is Cowan's (1975) M3 model between free vehicles (or platoon leaders) and those vehicles in
which assumes that a proportion, , of all vehicles are free and the platoon but behind the leader. This latter group are called
have an displaced exponential headway distribution and the 1- bunched vehicles. The benefits of a number of different headway
bunched vehicles have the same headway of only tm. models will be discussed later.
8.3 Headway Distributions Used in Gap Acceptance Calculations

8.3.1 Exponential Headways
d[P(ht)]
f(t)

q e qt (8.16)
The most common distribution is the negative exponential dt
distribution which is sometimes referred to as simply the
"exponential distribution". This distribution is based on the
assumption that vehicles arrive at random without any This is the equation for the negative exponential distribution.
dependence on the time the previous vehicle arrived. The The parameter q can be estimated from the flow or the reciprocal
distribution can be derived from assuming that the probability of of the average headway. As an example, if there were 228
a vehicle arriving in a small time interval (t, t+ t) is a constant. headways observed in half an hour, then the flow is 228/1800 i.e.
It can also be derived from the Poisson distribution which gives q = 0.127 veh/sec. The proportion of headways expected to be
the probability of n vehicles arriving in time t, that is: greater than 5 seconds is then
P(h>5) = e –q t
e qt
P(n)
(qt) n
(8.13) = e – 5*0.127
n! = 0.531
The expected number of headways greater than 5 seconds

where q is the flow in veh/sec. For n = 0 this equation gives the observed in half an hour is then 0.531 228 or 116.
probability that no vehicle arrives in time t. The headway, h,
must be then greater than t and the probability, from Equation If the flow was 1440 veh/h or 0.4 veh/sec then the number of
8.13 is headways less than 0.1 seconds is then q [P(h>0.1)] 3600 or
56 per hour. This over-estimation of the number of very short
P(h>t)
e qt (8.14) headways is considered to be unrealistic and the displaced
exponential distribution is often used instead of the negative
exponential distribution.
The cumulative probability function of headways is then
8.3.2 Displaced Exponential Distribution

P(ht)
1 e qt (8.15)
The shifted or displaced exponential distribution assumes that
there is a minimum headway between vehicles, tm. This time can
The probability distribution function is then be considered to be the space around a vehicle that no other
8-7
vehicle can intrude divided by the traffic speed. If the flow is q there are (1–) bunched vehicles;
veh/h then in one hour q vehicles will pass and there are tm q
seconds lost while these vehicles pass. The remaining time must h¯f is the average headway for free vehicles;
then be distributed randomly after each vehicle and the average
random component is (1-tm q)/q seconds. The cumulative is the average headway for bunched or constrained
probability distribution of headways is then: h̄b vehicles;
(h t m)
F(h)
1 e (8.17) tm is the shift in the curve.
Other composite headway models have been proposed by

where, Buckley (1962; 1968). However, a better headway model for
gap acceptance is the M3 model proposed by Cowan (1975).
q

(8.18)
This model does not attempt to model the headways between the
1 tmq bunched vehicles as these are usually not accepted but rather
models the larger gaps. This headway model has a cumulative
probability distribution:
There, the terms, and tm need to be evaluated. These can be
estimated from the mean and the variance of the distribution. (t t m)
p(h t)
1 e for t > tm (8.21)
The mean headway, E(h), is given by:
E(h)
1/q and

tm 1 (8.19) p(h t) = 0 otherwise.

Where is a decay constant given by the equation
The variance of headways is 1/2. These two relationships can q

then be used to estimate and tm.
(8.22)
(1 tmq)
This distribution is conceptually better than the negative
exponential distribution but it does not account for the Cowan's headway model is rather general. To obtain the
platooning that can occur in a stream with higher flows. A displaced exponential distribution set to 1.0. For the negative
dichotomized headway distribution provides a better fit. exponential distribution, set to 1.0 and tm to 0. Cowan's model
can also give the headway distribution used by Tanner (1962) by
setting to 1–tmq, however the distribution of the number of
8.3.3 Dichotomized Headway Distributions vehicles in platoons is not the same. This is documented below.
In most traffic streams there are two types of vehicles, the first Brilon (1988) indicated that the proportion of free vehicles could
are bunched vehicles; these are closely following preceding be estimated using the equation,
vehicles. The second group are free vehicles that are travelling

e Aqp (8.23)
without interacting with the vehicles ahead. There have been a
number of dichotomized headway distributions developed over
time. For instance, Schuhl (1955) proposed a distribution where A values ranged from 6 to 9. Sullivan and Troutbeck
(1993) found that this equation gave a good fit to data, from
t/h̄f
p(ht)
1 e (1 )e (t tm)/(h̄b tm) (8.20) more than 600 of hours of data giving in excess of 400,000
vehicle headways, on arterial roads in Australia. They also
found that the A values were different for different lanes and for
where there are vehicles that are free (not in platoons); different lane widths. These values are listed in Table 8.1.
8-8
Table 8.1 theory when predicting capacity or delays. The hyper-Erlang

“A” Values for Equation 8.23 from Sullivan and distribution given by Dawson (1969) is:
Troutbeck (1993).
(t tmf / h̄f tmf )
p(ht)
1 e
t tmb x

Median All other k 1 k (8.24)
(t tmb/h̄b t mb) h¯b tmb
Lane lanes (1 )e
x
0 x!
Lane width < 3.0 meters 7.5 6.5
3.0 Lane width 3.5 meters 7.5 5.25
Lane width> 3.5 meters 7.5 3.7 8.3.4 Fitting the Different Headway
Models to Data
Typical values of the proportion of free vehicles are given in
If the mean headway is 21.5 seconds and standard deviation is
Figure 8.3.
19.55 seconds, then the flow is 1/21.5 or 0.0465 veh/seconds
(167 veh/hour). A negative exponential curve that would fit this
The hyper-Erlang distribution is also a dichotomized headway
data is then,
distribution that provides an excellent fit to headway data. It is
useful in simulation programs but has not been used in traffic p(h t)
1 e 0.0465t
Figure 8.3
Typical Values for the Proportion of Free Vehicles.
8-9
To estimate the parameters for the displaced exponential

distribution, the difference between the mean and the standard P(n)
(1 )n 1 (8.25)
deviation is the displacement, that is tm is equal to 21.49 – 19.55
or 1.94 seconds. The constant used in Equation 8.21 is the
reciprocal of the standard deviation. In this case, is equal to Under these conditions the mean platoon size is
1/19.55 or 0.0512 veh/sec. The appropriate equation is then:
1
n̄
(8.26)
p(ht)
1 e 0.0512(t 1.94)
The data and these equations are shown in Figure 8.4 which and the variance by
indicates the form of these distributions. The reader should not
1
make any conclusions about the suitability of a distribution from Var(n)
(8.27)
this figure but should rather test the appropriateness of the model 2
to the data collected.
In many cases there are a substantial number of very short Another distribution of platoons used in the analysis of
headways and a dichotomized headway distribution performs unsignalized intersections is the Borel-Tanner distribution. This
better. As only the larger gaps are likely to be accepted by platooning distribution comes from Tanner's (1962) assumptions
drivers, there is no point in modeling the shorter gaps in great where the major stream gaps are the outcome of a queuing
detail. An example of Cowan’s M3 model and headway data process with random arrivals and a minimum inter-departure
from an arterial road is shown in Figure 8.5. Figure 8.6 gives time of tm. Although the distribution of these 'revised' major
the same data and the hyper-Erlang distribution. stream gaps is given by Equation 8.21 with equal to 1–tm q,
Figure 8.4
Exponential and Displaced Exponential Curves
(Low flows example).
8 - 10
Figure 8.5
Arterial Road Data and a Cowan (1975) Dichotomized Headway Distribution
(Higher flows example).
Figure 8.6
Arterial Road Data and a Hyper-Erlang Dichotomized Headway Distribution
(Higher Flow Example).
8 - 11
the distribution of the platoon length is Borel-Tanner (Borel Haight and Breuer (1960) found the mean platoon size to be
1942; Tanner 1953; 1961; and Haight and Breuer 1960). Again, 1 / (1 - tm q ) or 1/ and the variance to be tm q / (1 – tmq )3 or
q is the flow in veh/sec. The Borel-Tanner distribution of (1- ) /3 . For the same mean platoon size, the Borel-Tanner
platoons gives the probability of a platoon of size n as distribution has a larger variance and predicts a greater number
nt mq
of longer platoons than does the geometric distribution.
e (ntmq)n 1 Differences in the platoon size distribution does not affect an
P(n)
(8.28)
n! estimate of capacity but it does affect the average delay per
vehicle as shown in Figure 8.13.
where n is an integer.
8.4 Interaction of Two Streams

For an easy understanding of traffic operations at an unsignalized Therefore, the amount of capacity which is provided by t-gaps
intersection it is useful to concentrate on the simplest case first per hour is 3600 qp f(t) g(t).
(Figure 8.7).
To get the total capacity, expressed in veh/second, we have to
All methods of traffic analysis for unsignalized intersections are integrate over the whole range of major stream gaps:
derived from a simple queuing model in which the crossing of
two traffic streams is considered. A priority traffic stream qm
qp , f(t) # g(t)dt (8.29)
(major stream) of the volume qp (veh/h) and a non-priority traffic 0
stream (minor stream) of the volume qn (veh/h) are involved in
this queuing model. Vehicles from the major stream can cross
the conflict area without any delay. Vehicles from the minor where,
stream are only allowed to enter the conflict area, if the next qm = maximum traffic volume departing from the
vehicle from the major stream is still tc seconds away (tc is the stop line in the minor stream in veh/sec,
critical gap), otherwise they have to wait. Moreover, vehicles qp = major stream volume in veh/sec,
from the minor stream can only enter the intersection tf seconds f(t) = density function for the distribution of gaps in
after the departure of the previous vehicle (tf is the follow-up the major stream, and
time). g(t) = number of minor stream vehicles which can
enter into a major stream gap of size, t .
8.4.1 Capacity Based on the gap acceptance model, the capacity of the simple
2-stream situation (Figure 8.7) can be evaluated by elementary
The mathematical derivation of the capacity qm for the minor probability theory methods if we assume:
stream is as follows. Let g(t) be the number of minor stream
vehicles which can enter into a major stream gap of duration t. (a) constant tc and tf values,
The expected number of these t-gaps per hour is 3600qp f(t)
where, (b) exponential distribution for priority stream headways
f(t) = statistical density function of the gaps in the (cf. Equation 8.15), and
major stream and
qp = volume of the major stream. (c) constant traffic volumes for each traffic stream.
8 - 12
Figure 8.7
Illustration of the Basic Queuing System.
Within assumption (a), we have to distinguish between two

1 for tc(n 1)tft<tcntf
different formulations for the term g(t). These are the reason for pn(t)
two different families of capacity equations. The first family 0 elsewhere

assumes a stepwise constant function for g(t) (Figure 8.2):

g(t)
M n p (t)
n
0

n (8.30) The second family of capacity equations assumes a continuous
linear function for g(t) . This is an approach which has first been
used by Siegloch (1973) and later also by McDonald and
Armitage (1978).
where,
pn(t)= probability that n minor stream 0 for t < t0
vehicles enter a gap in the major stream g(t)
t t0 (8.31)
for t t0
of duration t, tf
8 - 13
where, Both approaches for g(t) produce useful capacity formulae where
the resulting differences are rather small and can normally be
tf
t0
tc ignored for practical applications (cf. Figure 8.8).
2
If we combine Equations 8.29 and 8.30, we get the capacity
equation used by Drew (1968), Major and Buckley (1962), and
Once again it has to be emphasized that both in Equations 8.30 by Harders (1968), which these authors however, derived in a
and 8.31, tc and tf are assumed to be constant values for all different manner:
drivers.
Figure 8.8
Comparison Relation Between Capacity (q-m) and Priority Street Volume (q-p) .
8 - 14
qpt c
e
qm
qp (8.32) If the constant tc and tf values are replaced by realistic
qpt f
1 e distributions (cf. Grossmann 1988) we get a decrease
in capacity.
If we combine Equations 8.29 and 8.31 we get Siegloch's (1973) Drivers may be inconsistent; i.e. one driver can have
formula, different critical gaps at different times; A driver might
reject a gap that he may otherwise find acceptable. This
1 qp tc
qm

e (8.33) effect results in an increase of capacity.
tf
If the exponential distribution of major stream gaps is
replaced by more realistic headway distributions, we
These formulae result in a relation of capacity versus conflicting get an increase in capacity of about the same order of
flow illustrated by the curves shown in Figure 8.8. magnitude as the effect of using a distribution for tc and
tf values (Grossmann 1991 and Troutbeck 1986).
The idealized assumptions, mentioned above as (a), (b), (c),
however, are not realistic. Therefore, different attempts to drop Many unsignalized intersections have complicated
one or the other assumption have been made. Siegloch (1973) driver behavior patterns, and there is often little to be
studied different types of gap distributions for the priority stream gained from using a distribution for the variables tc and
(cf. Figure 8.9) based on analytical methods. Similar studies tf or complicated headway distributions. Moreover,
have also been performed by Catchpole and Plank (1986) and Grossmann could show by simulation techniques that
Troutbeck (1986). Grossmann (1991) investigated these effects these effects compensate each other so that the simple
by simulations. These studies showed capacity equations, 8.32 and 8.33, also give quite
realistic results in practice.
Note: Comparison of capacities for different types of headway distributions in the main street traffic flow for tc = 6 seconds and
tf = 3 seconds. For this example, tm has been set to 2 seconds.
Figure 8.9
Comparison of Capacities for Different Types of
Headway Distributions in the Main Street Traffic Flow.
8 - 15
More general solutions have been obtained by replacing the equation:

exponential headway distribution used in assumption (b) with a
more realistic one e. g. a dichotomized distribution (cf. Section
qp(t c t m)
8.3.3). This more general equation is: qpe
qm
(1 qptm) (8.36)
qpt f
1 e
qpe (tc tm)
qm
(8.34)
t f
1 e
If the linear relationship for g(t) according to Equation 8.37 is
used, then the associated capacity equation is
where qpe (t0 tm)

qm
(8.37)
qf tf

(8.35)
(1 tmqf )
or
(t0 t m)
This equation is illustrated in Figure 8.10. This is also similar (1 qptm)e
qm
(8.38)
to equations reported by Tanner (1967), Gipps (1982), tf
Troutbeck (1986), Cowan (1987), and others. If is set to 1
and tm to 0, then Harders' equation is obtained. If is set to
l– qp tm , then this equation reduces to Tanner's (1962) This was proposed by Jacobs (1979) .
Figure 8.10
The Effect of Changing in Equation 8.31 and Tanner's Equation 8.36.
8 - 16
1E z(G tc)/-

Tanner (1962) analyzed the capacity and delay at an intersection qm
(8.41)
where both the major and minor stream vehicles arrived at E(C)E(1/-)
random; that is, their headways had a negative exponential
distribution. He then assumed that the major stream vehicles
were restrained such that they passed through the intersection at
intervals not less than tm sec after the preceding major stream
vehicle. This allowed vehicles to have a finite length into which where,
other vehicles could not intrude. Tanner did not apply the same E(C) = mean length of a "major road cycle" C,
constraint to headways in the minor stream. He assumed the C = G + B,
same gap acceptability assumptions that are outlined above. G = gap,
Tanner considered the major stream as imposing 'blocks' and B = block,
'anti-blocks' on the minor stream. A block contains one or more - = probability (G > tc), and
consecutive gaps less than tc sec; the block starts at the first z(t) = expected number of departures within
vehicle with a gap of more than tc sec in-front of it and ends tc the time interval of duration t.
sec after the last consecutive gap less than tc sec. Tanner's
equation for the entry capacity is a particular case of a more Since these types of solutions are complicated many researchers
general equation. have tried to find realistic capacity estimations by simulation
studies. This applies especially for the German method (FGSV
An analytical solution for a realistic replacement of assumptions 1991) and the Polish method.
(a) and (b) within the same set of formulae is given by Plank and
Catchpole (1984):
qpt c 8.4.2 Quality of Traffic Operations
qpe
q m
(8.39)
q pt f
1 e In general, the performance of traffic operations at an
intersection can be represented by these variables (measures of
effectiveness, MOE):
where (a) average delay,
Var(tf ) (b) average queue lengths,
1

1 qp2 Var(tc ) cf (8.40) (c) distribution of delays,
2 (e
q pt f
1) (d) distribution of queue lengths (i.e number of vehicles
queuing on the minor road),
(e) number of stopped vehicles and number of
Var(tc ) = variance of critical gaps, accelerations from stop to normal velocity, and
Var(tf ) = variance of follow-up-times, (f) probability of the empty system (po ).
c = increment, which tends to 0, when Var(tc
) approaches 0, and Distributions can be represented by:
f = increment, which tends to 0, when Var(tf
) approaches 0. standard deviations,
percentiles, and
the whole distribution.
Wegmann (1991) developed a universal capacity formula which
could be used for each type of distribution for the critical gap, for To evaluate these measures, two tools can be used to solve the
the follow-up time and for each type of the major stream problems of gap acceptance:
headway distribution.
queuing theory and
simulation.
8 - 17
q (t t ) q t (2t q 1)
2
Each of these MOEs are a function of qp and qn; the proportion
Dmin
e p c m
tc 1 p m m p 2 (8.45)
of "free" vehicles and the distribution of platoon size length in (1 tmqp)qp qp 2(1 tmqp)
both the minor and major streams. Solutions from queuing
theory in the first step concentrate on average delays.
A general form of the equation for the average delay per vehicle
is Another solution for average delay has been given by Harders
(1968). It is not based on a completely sophisticated queuing
D
Dmin 1
Jx theory. However, as a first approximation, the following
(8.42)
1 x equation for the average delay to non-priority vehicles is quite
useful.
(q t q t )
and J are constants 1 e p c n f
where D
t (8.46)
x is the degree of saturation = qn/qm qm/3600 qn f
and Dmin has been termed Adams' delay after Adams (1936).
Adams' delay is the average delay to minor stream vehicles when with qm calculated using Equation. 8.34 or similar.
minor stream flow is very low. It is also the minimum average
delay experienced by minor stream vehicles. M/G/1 Queuing System - A more sophisticated queuing theory
model can be developed by the assumption that the simple two-
Troutbeck (1990) gives equations for , J and Dmin based on the streams system (Figure 8.7) can be represented by a M/G/1
formulations by Cowan (1987). If stream 2 vehicles are queue. The service counter is the first queuing position on the
assumed to arrive at random, then is equal to 0. On the other minor street. The input into the system is formed by the vehicles
hand, if there is platooning in the minor stream, then is greater approaching from the minor street which are assumed to arrive
than 0. at random, i.e. exponentially distributed arrival headways (i.e.
"M"). The time spent in the first position of the queue is the
For random stream 2 arrivals, J is given by service time. This service time is controlled by the priority
stream, with an unknown service time distribution. The "G" is
e
q pt f
qptf 1qp(e qpt f 1)Dmin for a general service time. Finally, the "1" in M/G/1 stands for
J
(8.43) one service channel, i.e. one lane in the minor street.
qp(e
q pt f
1)Dmin
For the M/G/1 queuing system, in general, the Pollaczek-
Khintchine formula is valid for the average delay of customers in
Note that J is approximately equal to 1.0. Dmin depends on the the queue
platooning characteristics in stream 1. If the platoon size
xW(1Cw)
2
distribution is geometric, then Dq
(8.47)
2(1 x)
(t c t m) 2tm2tm 2
tm
Dmin
e
tc 1 (8.44)
qp 2(tm)
where
W = average service time. It is the average
(Troutbeck 1986). time a minor street vehicle spends in the
first position of the queue near the
Tanner's (1962) model has a different equation for Adams' delay, intersection
because the platoon size distribution in stream 1 has a Borel- Cw = coefficient of variation of service times
Tanner distribution. This equation is
8 - 18
Var(W) D
Dmin (1) 1
J x
Cw
(8.49)
W 1 1 x
Var (W) = variance of service times

where and are documented in Troutbeck (1990).
The total average delay of minor street vehicles is then
This is similar to the Pollaczek-Khintchine formula (Equation
D = Dq + W. 8.48). The randomness constant C is given by (+)/(l+) and
the term 1/Dmin*(l+) can be considered to be an equivalent
In general, the average service time for a single-channel queuing 'capacity' or 'service rate.' Both terms are a function of the
system is: l/capacity. If we derive capacity from Equations 8.32 critical gap parameters tc and tf and the headway distributions.
and following and if we include the service time W in the total However, C, , and values are not available for all conditions.
delay, we get
For the M/G/1 system as a general property, the probability po of
1 x
D
1 C (8.48) the empty queue is given by
qm 1 x
po = 1 - x (8.50)
where This formula is of sufficient reality for practical use at

unsignalized intersections.
1Cw
2
C
2 M/G2/1 queuing system - Different authors found that the

service time distribution in the queuing system is better
described by two types of service times, each of which has a
Up to this point, the derivations are of general validity. The real specific distribution:
problem now is to evaluate C. Only the extremes can be defined
which are: W1 = service time for vehicles entering the empty system, i.e
no vehicle is queuing on the vehicle's arrival
Regular service: Each vehicle spends the same time in the
first position. This gives Var(W) = 0, Cw2 = 0, and C = W2 = service time for vehicles joining the queue when other
0.5 vehicles are already queuing.
This is the solution for the M/D/l queue. Again, in both cases, the service time is the time the vehicle
spends waiting in the first position near the stop line. The first
Random service: The times vehicles spend in the first ideas for this solution have been introduced by Kremser (1962;
position are exponentially distributed. This gives 1964) and in a comparable way by Tanner (1962), as well as by
Var(W) = E(W), Cw2 = 1, and C = 1.0 Yeo and Weesakul (1964).
This gives the solution for the M/M/1 queue. The average time which a customer spends in the queue of such
a system is given by Yeo's (1962) formula:
Unfortunately, neither of these simple solutions applies exactly
to the unsignalized intersection problem. However, as an
E(W1 ) E(W2 ) E(W2 )
2 2 2
approximation, some authors recommend the application of qn
Dq
(8.51)
Equation 8.48 with C = 1. 2 v y
Equation 8.42 can be further transformed to

where,
8 - 19
Dq = average delay of vehicles in the queue at

yE(W1 )zE(W2 )
higher positions than the first, 2 2 2
E(W1 ) qn
E(W1) = expectation of W1, D
(8.53)
v 2 vy
E(W12) = expectation of (W1*W1)
E(W2) = expectation of W2,
E(W22) = expectation of (W2* W2),
v = y + z, (Brilon 1988):Formulae for the expectations of W1 and W2
y = 1 –qn E(W2), and respectively have been developed by Kremser (1962):
z = qn E(Wl).
1 qpt
E(W1)
(e (8.54)
The probability po of the empty queue is qp
qt
po= y/v (8.52) e p c (1 qptf)
E(W2)
e
qp
The application of this formula shows that the differences against q pt c
2 qpt c
E(W1 )
(e 1 qptc)( e tf tc)tf tc
2 2 2
Equation 8.50 are quite small ( < 0.03). Refer to Figure 8.11.
qp qp
If we also include the service time ( = time of minor street q pt c
vehicles spent in the first position) in the total delay, we get E(W2 )
2 2e
2
(e qptc)(1 e qptf) qptfe
q pt c
qp
Figure 8.11
Probability of an Empty Queue: Comparison of Equations 8.50 and 8.52.
8 - 20
Kremser (1964), however, showed that the validity of these regarded as approximations and only apply for
equations is restricted to the special case of tc = tf , which is undersaturated conditions and steady state conditions.
rather unrealistic for two-way-stop-control unsignalized
intersections. Daganzo (1977) gave an improved solution for Figure 8.12 gives a graphical comparison for some of the delay
E(W2) and E(W22) which again was extended by Poeschl (1983). formulae mentioned.
These new formulae were able to overcome Kremer's (1964)
restrictions. It can, however be shown that Kremer's first Differences in the platoon size distribution affects the average
approach (Equation 8.56) also gives quite reliable approximate delay per vehicle as shown in Figure 8.13. Here, the critical gap
results for tc and tf values which apply to realistic unsignalized was 4 seconds, the follow-up time was 2 seconds, and the
intersections. The following comments can also be made about priority stream flow was 1000 veh/h. To emphasize the point,
the newer equations. the average delay for a displaced exponential priority stream is
4120 seconds, when the minor stream flow was 400 veh/h. This
The formulae are so complicated that they are far from is much greater than the values for the Tanner and exponential
being suitable for practice. The only imaginable application headway examples which were around 11.5 seconds for the same
is the use in computer programs. major stream flow. The average delay is also dependent on the
average platoon size as shown in Figure 8.14. The differences
Moreover, these formulae are only valid under assumptions in delays are dramatically different when the platoon size is
(a), (b), and (c) in Section 8.4.1 of the paper. That means changed.
that for practical purposes, the equations can only be
Note: For this example; qp = 600 veh/h, tc = 6 sec , and tf = 3 sec.
Figure 8.12
Comparison of Some Delay Formulae.
8 - 21
Figure 8.13
Average Steady State Delay per Vehicle
Calculated Using Different Headway Distributions.
Figure 8.14
Average Steady State Delay per Vehicle by Geometric
Platoon Size Distribution and Different Mean Platoon Sizes.
8 - 22
8.4.3 Queue Length calculated from these equations directly. Therefore, Wu (1994)
developed another set of formulae which approximate the above
In each of these queuing theory approaches, the average queue mentioned exact equations very closely:
length (L) can be calculated by Little's rule (Little 1961): p(0)
1 x a
(8.58)
p(n)
p(0)x a(b(n 1)1)
L = qn D (8.55)
Given that the proportion of time that a queue exists is equal to

the degree of saturation, the average queue length when there is p(n) = probability that n vehicles are queuing on the
a queue is: minor street
Lq = qn D/x = qm D (8.56)
where,
qn
The distribution of queue length then is often assumed to be x
geometric. qm
However, a more reliable derivation of the queue length x = degree of saturation (qm according to Equation
distribution was given by Heidemann (1991). The following 8.33).
version contains a correction of the printing mistakes in the
1
original paper (there: Equations 8.30 and 8.31). a
tc tf
p(0)
h1h3(qpqn) (8.57) 1 0.45 qp
tf
p(1)
p(0)h3qn e (tc tf)h2 qnh1h3
q nt f
1.51
b
tc
p(n)
p(n 1)h3qn e (tc tf)h2 1 0.68
q nt f
qp
tf

n 2
(tc tf qn)n m ( qntf )n m e
q nt f
h3 p(m) h2
m
0 (n m)! tf (n m 1)!
For the rather realistic approximation tc 2 tf , we get :
1 1.51
p(n) = probability that n vehicles are queuing on the a
b
1 0.45 qp 11.36qp
minor street
From Equation 8.58 we get the cumulative distribution function

qpt c qpt f qn
h1
e (e 1)
qp F(n)
p(Ln)
1 x a(b n1)
(8.59)
qpt c qn(t c t f)
h2
qpe
1

h2qne qpt f For a given percentile, S, (e.g. S = F(n) = 0.95) this equation
h3 can be solved for n to calculate the queue length which is only
exceeded during (1-S)*100 percent of the time (Figure 8.15).
For practical purposes, queue length can be calculated with
These expressions are based on assumptions (a), (b), and (c) in sufficient precision using the approximation of the M/M/1
Section 8.4.1. This solution is too complicated for practical use. queuing system and, hence, Wu’s equation. The 95-percentile-
Moreover, specific percentiles of the queue length is the desired queue length based on Equation 8.59 is given in Figure 8.15.
output rather than probabilities. This however, can not be
8 - 23
The parameter of the curves (indicated on the right side) is the degree of saturation ( x ).
Figure 8.15
95-Percentile Queue Length Based on Equation 8.59 (Wu 1994).
8.4.4 Stop Rate follow-up time tf , increases from some minimum value, P(0,t),
to 1 as the degree of saturation increases from 0 to 1.
The proportion of drivers that are stopped at an unsignalized
intersection with two streams was established by Troutbeck The proportion of drivers stopped for more than a short period
(1993). The minor stream vehicles were assumed to arrive at t, P(x,t), is given by the empirical equation:
random whereas the major stream headways were assumed to
have a Cowan (1975) M3 distribution. Changes of speed are P(x,t)
P(0,t)A1 P(0,t)x(1 A)1 P(0,t)x 2
(8.61)
assumed to be instantaneous and the predicted number of (1 A)(1 B)(1 x)x
stopped vehicles will include those drivers who could have
adjusted their speed and avoided stopping for very short periods.
The proportion stopped, P(x,0), is dependent upon the degree of where

saturation, x, the headways between the bunched major stream t (t t )
vehicles, tm, the critical gap, tc . and the major stream flow, qp . B
1 (1 )(1 tmqp)e a m
tf
The appropriate equation is: (t a t m)
A
1 a0e
(t c t m)
P(x,0)
1 (1 x)(1 tmqp)e (8.60)
and
(t a t m)
P(0,t)
P(0,0) qpte (8.62)
where is given by qp/(1-tmqp). The proportion of drivers
stopped for more than a short period of t, where t is less than the
8 - 24
or approximations if T is considerably greater than the expression

(t a t m) on the right side of the following equation.
P(0,t)
1 (1 tmqpqpt)e
1
T>
(8.63)
qm qn
2
If the major stream is random then a0 is equal to 1.25 and for

bunched major stream traffic, it is 1.15. The vehicles that are
stopped for a short period may be able to adjust their speed and with T = time of observation over which the average delay
these vehicles have been considered to have a “partial stop." should be estimated in seconds,
Troutbeck (1993) also developed estimates of the number of
times vehicles need to accelerate and move up within the queue. after Morse (1962).
This inequality can only be applied if qm and qn are nearly

8.4.5 Time Dependent Solution constant during time interval T. The threshold given by
Equation 8.63 is illustrated by Figure 8.16. The curves are given
Each of the solutions given by the conventional queuing theory for time intervals T of 5, 10, 15, 30, and 60 minutes. Steady
above is a steady state solution. These are the solutions that can state conditions can be assumed if qn is below the curve for the
be expected for non-time-dependent traffic volumes after an corresponding T-value. If this condition (Equation 8.63) is not
infinitely long time, and they are only applicable when the degree fulfilled, time-dependent solutions should be used.
of saturation x is less than 1. In practical terms, this means, Mathematical solutions for the time dependent problem have
the results of steady state queuing theory are only useful been developed by Newell (1982) and now need to be made
Note: The curves are given for time intervals T of 5, 10, 15, 30, and 60 minutes. Steady state cond itions can be assumed if q n is
below the curve for the corresponding T-value.
Figure 8.16
Approximate Threshold of the Length of Time Intervals For the Distinction
Between Steady-State Conditions and Time Dependent Situations.
8 - 25
more accessible to practicing engineers. There is, however, a This delay formula has proven to be quite useful to estimate
heuristic approximate solution for the case of the peak hour delays and it has a quite reliable background particularly for
effect given by Kimber and Hollis (1979) which are based on the temporarily oversaturated conditions.
ideas of Whiting, who never published his work.
A simpler equation can be obtained by using the same co-
During the peak period itself, traffic volumes are greater than ordinate transfer method. This is a more approximate method.
those before and after that period. They may even exceed The steady state solution is fine for sites with a low degree of
capacity. For this situation, the average delay during the peak saturation and the deterministic solution is satisfactory for sites
period can be estimated as: with a very high degree of saturation say, greater than three or
four. The co-ordinate transfer method is one technique to
provide estimates between these two extremes. the reader
1 should also refer to Section 9.4.
D
D1E
qm
The steady state solution for the average delay to the entering
1
D1
F 2G F vehicle is given by Equation 8.42. The deterministic equation
2 for delay, Dd, on the other hand is
1 T (q h)
F
m q n C(y
)y E
qmo qno 2 qm
2L0(xd 1)qmT
2Ty q Dd
Dmin x>1 (8.65)
G
C n (qm qn)E (8.64) 2qm
qmo qno qm
Cqno
E
qmo(qmo qno) and Dd = 0

h
qm qmoqno otherwise,
h
y
1
qn where L0 is the initial queue,
T is time the system is operating in seconds, and
qm is the entry capacity.
qm = capacity of the intersection entry during the These equations are diagrammatically illustrated in Figure 8.17.
peak period of duration T, For a given average delay the co-ordinate transformation method
qmo = capacity of the intersection entry before and gives a new degree of saturation, xt , which is related to the
after the peak period, steady state degree of saturation, xs , and the deterministic degree
qn = minor street volume during the peak period of of saturation, xd , such that
duration T, and
qno = minor street volume before and after the peak xd – xt = 1 – xs = a (8.66)
period
Rearranging Equations 8.42 and 8.65 gives two equations for xs
(each of these terms in veh/sec; delay in sec). and xd as a function of the delays Dd and Ds . These two
equations are:
C is again similar to the factor C mentioned for the M/G/1
system, where
Ds Dmin Dmin
x s
(8.67)
C = 1 for unsignalized intersections and Ds DminJDmin
C = 0.5 for signalized intersections (Kimber and Hollis
1979).
8 - 26
Figure 8.17
The Co-ordinate Transform Technique.
and
T(1 x) L
2(Dd Dmin) 2L0/qm A
0 Dmin(2 J) (8.71)
xd
1 (8.68) 2 qm
T
Using Equation 8.66, xt is given by: and

T(1 x)(1) Tx(J)
B
4Dmin
2(Dd Dmin) 2L0/qm Ds Dmin Dmin 2 2
xt
(8.69) (8.72)
T Ds DminJDmin L0
(1 J) Dmin
qm
Rearranging Equation 8.69 and setting D = Ds = Dd , x = xJ gives:

Equation 8.66 ensures that the transformed equation will
asymptote to the deterministic equation and gives a family of
1
Dt
A 2B A (8.70) relationships for different degrees of saturation and period of
2 operation from this technique (Figure 8.18).
A simpler equation was developed by Akçelik in Akcelik and

where Troutbeck (1991). The approach here is to rearrange Equation
8.42 to give:
8 - 27
Figure 8.18
A Family of Curves Produced from the Co-Ordinate Transform Technique.
Dmin(Jxs)
a
1 xs
D
1 T
(x 1) (x 1)2 8x
Ds Dmin qm 4 qmT
(8.75)
and this is approximately equal to: The average delay predicted by Equation 8.74 is dependent on
the initial queue length, the time of operation, the degree of
Dmin(Jxt)
a (8.73) saturation, and the steady state equation coefficients. This
Ds Dmin equation can then be used to estimate the average delay under
oversaturated conditions and for different initial queues. The use
of these and other equations are discussed below.
If this is used in Equation 8.66 and then rearranged then the
resulting equation of the non-steady state delay is:
8.4.6 Reserve Capacity
1 L0 (x 1)T
D Dmin
Independent of the model used to estimate average delays, the
2 qm 4 reserve capacity (R) plays an important role
(8.74)
TDmin(Jx)
(x 1)T
L0 2

2qm 4 2
R
qemax qn (8.76)
A similar equation for M/M/1 queuing system can be obtained

if J is set to 1, is set to zero, and Dmin is set to 1/qm; the result
is:
8 - 28
In the 1985 edition of the HCM but not the 1994 HCM, it is used
1 L0 RfT Rf L0 1
as the measure of effectiveness. This is based on the fact that b
1 (8.79)
average delays are closely linked to reserve capacity. This close q m Rf 2 R0 qm Rf
relationship is shown in Figure 8.19. In Figure 8.19, the average
100 # 3600
delay, D, is shown in relation to reserve capacity, R. The delay Rf
(8.80)
calculations are based on Equation 8.64 with a peak hour T
interval of duration T= 1 hour. The parameters (100, 500, and
1000 veh/hour) indicate the traffic volume, qp, on the major qn0 qm0 R0
L0

(8.81)
street. Based on this relationship, a good approximation of the R0 R0
average delay can also be expressed by reserve capacities. What
we also see is that - as a practical guide - a reserve capacity
R > 100 pcu/h generally ensures an average delay T = duration of the peak period
below 35 seconds. qm = capacity during the peak period
qn = minor street flow during the peak period
R = reserve capacity during the peak period
Brilon (1995) has used a coordinate transform technique for the
= qemax – qn
"Reserve Capacity" formulation for average delay with
L0 = average queue length in the period before and
oversaturated conditions. His set of equations can be given by
after the peak period
D
B B 2 b (8.77) qn0 = minor street flow in the period before and after
the peak period
qm0 = capacity in the period before and after the peak
where period
R0 = reserve capacity in the period before and after the
L peak period
1
B
bR 0 (8.78)
2 qm
Figure 8.19
Average Delay, D, in Relation to Reserve Capacity R.
8 - 29
All variables in these equations should be used in units of depart according to the gap acceptance mechanism. The
seconds (sec), number of vehicles(veh), and veh/sec. Any effect of limited acceleration and deceleration can, of
capacity formula to estimate qm and qm0 from Section 8.4.1 can course, be taken into account using average vehicle
be used. performance values (Grossmann 1988). The advantage of
this type of simulation model is the rather shorter computer
The numerical results of these equations as well as their degree time needed to run the model for realistic applications.
of sophistication are comparable with those of Equation 8.75. One such model is KNOSIMO (Grossmann 1988, 1991).
It is capable of being operated by the traffic engineer on
his personal computer during the process of intersection
8.4.7 Stochastic Simulation design. A recent study (Kyte et al , 1996) pointed out that
KNOSIMO provided the most realistic representation of
As mentioned in the previous chapters, analytical methods are traffic flow at unsignalized intersections among a group of
not capable of providing a practical solution, given the other models.
complexity and the assumptions required to be made to analyze
unsignalized intersections in a completely realistic manner. The KNOSIMO in its present concept is much related to
modern tool of stochastic simulation, however, is able to German conditions. One of the specialities is the
overcome all the problems very easily. The degree of reality of restriction to single-lane traffic flow for each direction of
the model can be increased to any desired level. It is only the main street. Chan and Teply (1991) found some easy
restricted by the efforts one is willing to undertake and by the modifications to adjust the model to Canadian conditions
available (and tolerable) computer time. Therefore, stochastic as well. Moreover, the source code of the model could
simulation models for unsignalized intersections were developed easily be adjusted to traffic conditions and driver behavior
very early (Steierwald 1961a and b; Boehm, 1968). More recent in other countries.
solutions were developed in the U. K. (Salter 1982), Germany
(Zhang 1988; Grossmann 1988; Grossmann 1991), Canada 2) Car Tracing Models - These models give a detailed
(Chan and Teply 1991) and Poland (Tracz 1991). account of the space which cars occupy on a road together
with the car-following process but are time consuming to
Speaking about stochastic simulation, we have to distinguish two run. An example of this type of model is described by
levels of complexity: Zhang (1988).
1) Point Process Models - Here cars are treated like points, Both types of models are useful for research purposes. The
i.e. the length is neglected. As well, there is only limited models can be used to develop relationships which can then be
use of deceleration and acceleration. Cars are regarded as represented by regression lines or other empirical evaluation
if they were "stored" at the stop line. From here they techniques.
8.5 Interaction of Two or More Streams in the Priority Road

The models discussed above have involved only two streams; a single lane with the opposing flow being equal to the sum of
one being the priority stream and the second being a minor the lane flows. This results in the following equation for
stream. The minor stream is at a lower rank than the priority capacity in veh/h:
stream. In some cases there may be a number of lanes that must
be given way to by a minor stream driver. The capacity and the qt a
3600qe
delay incurred at these intersections have been looked at by a qemax
(8.82)
qt f
number of researchers. A brief summary is given here. 1 e
If the headways in the major streams have a negative exponential

distribution then the capacity is calculated from the equation for where q is the total opposing flow.
8 - 30
Tanner (1967) developed an equation for the capacity of an 8.5.1 The Benefit of Using a
intersection where there were n major streams. The traffic in
Multi-Lane Stream Model
each lane has a dichotomized headway distribution in which
there is a proportion of vehicles in bunches and the remaining
Troutbeck (1986) calculated the capacity of a minor stream to
vehicles free of interaction. All bunched vehicles are assumed
cross two major streams which both have a Cowan (1975)
to have a headway of tm and the free vehicles have a headway
dichotomized headway distribution. The distribution of
equal to the tm plus a negative exponentially distributed (or
opposing headways is:
random) time. This is the same as Cowan's (1975) M3 model.
Using the assumption that headways in each lane are 2q1q2t
F(t)
for t < tm (8.87)
independent, Tanner reviewed the distribution of the random (q1q2)
time periods and estimated the entry capacity in veh/h as:
(t a t m) and
3600 (1 tm1q1)(1 tm2q2)(1 tm1q1)e
qemax
t f
(8.83) (t t m)
1 e F(t)
1 e for t > tm (8.88)
where = 1 + 2+ . . . . . + n (8.84)

where
i = iqi / (1-tm qi) (8.85) 1q1(1 q2tm) 2q2(1 q1tm)

(8.88a)
(q1q2)
qi is the flow in the major stream i in veh/sec.
i is the proportion of free vehicles in the major stream i.
or after a little algebra,
This equation by Tanner is more complicated than an earlier
N (1 q t )
n
equation (Tanner 1962) based on an implied assumption that the
proportion of free vehicles, i, is a function of the lane flow. q
i m (8.88b)
i
1
That is
i = (1-tm qi) and

' = 1 + 2 (8.89)
and then i reduces to qi. Fisk (1989) extended this earlier work
of Tanner (1962) by assuming that drivers had a different critical
gap when crossing different streams. While this would seem to As an example, if there were two identical streams then the
be an added complication it could be necessary if drivers are distribution of headways between vehicles in the two streams is
crossing some major streams from the left before merging with given by Equations 8.87 and 8.88. This is also shown in figures
another stream from the right when making a left turn. from Troutbeck (1991) and reported here as Figure 8.20.
Gap acceptance procedures only require that the longer

Her equation for capacity is: headways or gaps be accurately represented. The shorter gaps
need only be noted.
N (1 t
n
qit ai qt m
3600q mi q i )e e
i
1 (8.86) Consequently the headway distribution from two lanes can be
qemax
qt f
1 e represented by a single Cowan M3 model with the following
properties:
(t t m)
F(t)
1 e t > tm (8.90)
where q = q 1 + q 2+ . . . . . + q n
8 - 31
Figure 8.20
Modified 'Single Lane' Distribution of Headways (Troutbeck 1991).
and otherwise F(t) is zero. This modified distribution is also

illustrated in Figure 8.20. Values of * and tm* must be chosen (t m tm,i

1 (1 tmq1)(1 tmq2)e
)

to ensure the correct proportions and the correct mean headway tm,i1
(8.93)
q1q2
are obtained. This will ensure that the number of headways
greater than t, 1–F(t), is identical from either the one lane or the
two lane equations when t is greater than tm*.
Troutbeck (1991) gives the following equations for calculating

* and tm* which will allow the capacity to be calculated using * is then found from Equation 8.93.
a modified single lane model which are identical to the estimate
from a multi-lane model. Troutbeck (1991) also indicates that the error in calculating
Adams' delay when using the modified single lane model instead
The equations of the two lane model is small. Adams' delay is the delay to the
minor stream vehicles when the minor stream flow is close to
zero. This is shown in Figure 8.21. Since the modified
t m

(1 tmq1)(1 tmq2)e tm (8.91)

(1 tm q1 tmq2)e distribution gives satisfactory estimates of Adams' delay, it will

also give satisfactory estimates of delay.
and In summary, there is no practical reason to increase the

complexity of the calculations by using multi-lane models and a
e tm
e tm

(8.92) single lane dichotomized headway model can be used to
represent the distribution of headways in either one or two lanes.
are best solved iteratively for tm with tm,i being the ith estimate.
The appropriate equation is
8 - 32
Figure 8.21
Percentage Error in Estimating Adams' Delay Against the
Major Stream Flow for a Modified Single Lane Model (Troutbeck 1991).
8.6 Interaction of More than Two Streams of Different Ranking

8.6.1 Hierarchy of Traffic Streams at a This is illustrated in Figure 8.22 produced for traffic on the right
side. The figure illustrates that the left turners on the major road
Two Way Stop Controlled Intersection
have to yield to the through traffic on the major road. The left
turning traffic from the minor road has to yield to all other
At all unsignalized intersections except roundabouts, there is a streams but is also affected by the queuing traffic in the Rank 2
hierarchy of streams. Some streams have absolute priority stream.
(Rank 1), while others have to yield to higher order streams. In
some cases, streams have to yield to some streams which in turn
have to yield to others. It is useful to consider the streams as
having different levels of priority or ranking. These different
8.6.2 Capacity for Streams of
levels of priority are established by traffic rules. For instance, Rank 3 and Rank 4
Rank 1 stream has absolute priority and does not need to No rigorous analytical solution is known for the derivation of the
yield right of way to another stream, capacity of Rank-3-movements like the left-turner from the
minor street at a T-junction (movement 7 in Figure 8.22, right
Rank 2 stream has to yield to a Rank 1 stream, side). Here, the gap acceptance theory uses the impedance
factors p0 as an approximation. p0 for each movement is the
Rank 3 stream has to yield to a Rank 2 stream and in turn to probability that no vehicle is queuing at the entry. This is given
a Rank 1 stream, and with sufficient accuracy by Equation 8.50 or better with the two
service time Equation 8.52. Only during the part p0,rank-2 of the
Rank 4 stream has to yield to a Rank 3 stream and in turn to total time, vehicles of Rank 3 can enter the intersection due to
Rank 2 and Rank 1 streams (left turners from highway code regulations.
the minor street at a cross-intersection).
8 - 33
Note: The numbers beside the arrows indicate the enumeration of streams given by the Highway Capacity Manual (1994,
Chapter 10).
Figure 8.22
Traffic Streams And Their Level Of Ranking.
Therefore, for Rank-3-movements, the basic value qm for the For Rank-4-movements (left turners at a cross-intersection), the
potential capacity must be reduced to p0 qm to get the real dependency between the p0 values in Rank-2 and Rank-3-
potential capacity qe: movements must be empirical and can not be calculated from
analytical relations. They have been evaluated by numerous
qe,rank-3 = p0,rank-2 . qm,rank-3 (8.94) simulations by Grossmann (1991; cf. Brilon and Grossmann
1991). Figure 8.23 shows the statistical dependence between
For a T-junction, this means queues in streams of Ranks 2 and 3.
qe,7 = p0,4 . qm,7 In order to calculate the maximum capacity for the Rank-4-
movements (numbers 7 and 10), the auxiliary factors, pz,8 and
For a cross-junction, this means pz,11, should be calculated first:
py,i
qe,8 = px . qm,8 (8.95) pz,i
0.65py,i 0.6 py,i (8.97)
py,i3
qe,11 = px . qm,11 (8.96)
with
diminished to calculate the actual capacities, qe. Brilon (1988,
px = p0,1 . p0,4 cf. Figures 8.7 and 8.8) has discussed arguments which support
this double introduction.
Here the index numbers refer to the index of the movements
according to Figure 8.22. Now the values of p0,8 and p0,11 can be The reasons for this are as follows:
calculated according to Equation 8.50.
8 - 34
Figure 8.23
Reduction Factor to Account for the Statistical Dependence
Between Streams of Ranks 2 and 3.
During times of queuing in Rank-2 streams (e.g. left Even if no Rank-2 vehicle is queuing, these vehicles
turners from the major street), the Rank-3 vehicles (e.g. influence Rank-3 operations, since a Rank-2 vehicle
left turners from the minor street at a T-junction) cannot approaching the intersection within a time of less than tc
enter the intersection due to traffic regulations and the prevents a Rank-3 vehicle from entering the intersection.
highway code. Since the portion of time provided for
Rank-3 vehicles is p0, the basic capacity calculated from Grossmann (1991) has proven that among the possibilities
Section 8.4.1 for Rank-3 streams has to be diminished by considered, the described approach is the easiest and quite
the factor p0 for the corresponding Rank-2 streams realistic in the range of traffic volumes which occur in practical
(Equations 8.95 to 8.99). applications.
8.7 Shared Lane Formula
8.7.1 Shared Lanes on the Minor Street

M
m bi
1

(8.100)
qs i
1 qm,i
If more than one minor street movement is operating on the same
lane, the so-called "shared lane equation" can be applied. It
calculates the total capacity qs of the shared lane, if the capacities
of the corresponding movements are known. (Derivation in qs = capacity of the shared lane in veh/h,
Harders, 1968 for example.) qm,i = capacity of movement i, if it operates on a
separate lane in veh/h,
8 - 35
bi = proportion of volume of movement i of the total in Figure 8.21 may also be obstructed by queuing vehicles in
volume on the shared lane, those streams. The factors p0,1* and p 0,4* indicate the probability
m = number of movements on the shared lane. that there will be no queue in the respective shared lane. They
might serve for a rough estimate of the disturbance that can be
The formula is also used by the HCM (1994, Equation 10-9). expected and can be approximated as follows (Harders 1968):
1 p0,i
This equation is of general validity regardless of the formula for po,i
1 (8.101)
the estimation of qm and regardless of the rank of priority of the 1 qjtBj qktBk
three traffic movements. The formula can also be used if the
overall capacity of one traffic stream should be calculated, if this
stream is formed by several partial streams with different where: i = 1, j = 2 and k = 3 (cf. Figure 8.22)
capacities, e.g. by passenger cars and trucks with different
critical gaps. Kyte at al (1996) found that this procedure for or
accounting for a hierarchy of streams, provided most realistic i = 4, j = 5 and k = 6 (cf. Figure 8.22)
results.
qj = volume of stream j in veh/sec,
qk = volume of stream k in veh/sec, and
8.7.2 Shared Lanes on the Major Street tBj and tBk = follow-up time required by a vehicle in stream j
or k (s).
In the case of a single lane on the major street shared by right- (1.7 sec < tB < 2.5 sec, e.g. tB = 2 sec)
turning and through movements (movements no. 2 and 3 or 5
and 6 in Figure 8.22), one can refer to Table 8.2. In order to account for the influence of the queues in the major
street approach lanes on the minor street streams no. 7, 8, 10,
If left turns from the major street (movements no. 1 and 4 in and 11, the values p0,1 and p0,4 , according to Equation 8.47 have
Figure 8.22) have no separate turning lanes, vehicles in the to be replaced by the values p0,1* and p0,4* according to Equation
priority l movements no. 2 and 3, and no. 5 and 6 respectively 8.101. This replacement is defined in Equations 8.95 to 8.97.
8.8 Two-Stage Gap Acceptance and Priority

At many unsignalized intersections there is a space in the center the basis of an adjustment factor . The resulting set of
of the major street available where several minor street vehicles equations for the capacity of a two-stage priority situation are:
can be stored between the traffic flows of the two directions of
the major street, especially in the case of multi-lane major traffic cT
y(y k 1) [c(q5) q1] (y 1) c(q1 q2 q5)

(Figure 8.24). This storage space within the intersection enables y k1
1
the minor street driver to cross the major streams from each
direction at different behavior times. This behavior can
contribute to an increased capacity. This situation is called two- for y C 1
stage priority. The additional capacity being provided by these
wider intersections can not be evaluated by conventional cT(y
1)
k[c(q5) q1] (c(q1 q2 q5) (8.104)

capacity calculation models. k1
Brilon et al. (1996) have developed an analytical theory for the

estimation of capacities under two-stage priority conditions. It is for y = 1
based on an earlier approach by Harders (1968). In addition to cT = total capacity of the intersection for minor through traffic
the analytical theory, simulations have been performed and were (movement 8)
8 - 36
Table 8.2
Evaluation of Conflicting Traffic Volume qp
Note: The indices refer to the traffic streams denoted in Figure 8.22.
Subject Movement No. Conflicting Traffic Volume qp
Left Turn from Major Road 1 q5 + q63)
7 q2 + q33)
Right Turn from Minor 6 q22) + 0.5 q31)

Road
12 q52) + 0.5 q61)
Through Movement from 5 q2 + 0.5 q31) + q5 + q63) + q1 + q4

Minor Road
11 q2 + q33) + q5 + 0.5 q61) + q1 + q4
Left Turn from Minor Road 4 q2 + 0.5 q31) + q5 + q1 + q4 + q124)5)6) + q115)
10 q5 + 0.5 q61) + q2 + q1 + q4 + q64)5)6) + q85)
Notes
1) If there is a right-turn lane, q3 or q6 should not be considered.
2) If there is more than one lane on the major road, q2 and q5 are considered as traffic volumes on the right
lane.
3) If right-turning traffic from the major road is separated by a triangular island and has to comply with a
YIELD- or STOP-Sign, q6 and q3 need not be considered.
4) If right-turning traffic from the minor road is separated by a triangular island and has to comply with a
YIELD- or STOP-sign, q9 and q12 need not be considered.
5) If movements 11 and 12 are controlled by a STOP-sign, q11 and q12 should be halved in this equation.
Similarly, if movements 8 and 9 are controlled by a STOP-sign, q8 and q9 should be halved.
6) It can also be justified to omit q9 and q12 or to halve their values if the minor approach area is wide.
where = 1 for k=0

c(q1 q2) c (q1 q2 q5)
y
a
1 –0.32exp ( 1.3 k) for k > 0 (8.105)
c(q5) q1 c (q1 q2 q5)
Of course, here the volumes of all priority movements at part II

q1 = volume of priority street left turning traffic at part I have to be included. These are: major right (6, except if this
q2 = volume of major street through traffic coming from the movement is guided along a triangular island separated from the
left at part I through traffic) , major through (5), major left (4); numbers of
q5 = volume of the sum of all major street flows coming movements according to Figure 8.22.
from the right at part II.
8 - 37
Note: The theory is independent of the number of lanes in the major street.
Figure 8.24
Minor Street Through Traffic (Movement 8) Crossing the Major Street in Two Phases.
c(q1 + q2) = capacity at part I The same set of formulas applies in analogy for movement 7. If
c(q5) = capacity at part II both movements 7 and 8 are operated on one lane then the total
c(q1+q2+q5) = capacity at a cross intersection for capacity of this lane has to be evaluated from cT7 and cT8 using
minor through traffic with a major the shared lane formula (Equation 8.95). Brilon et al. (1996)
street traffic volume of q1+q2+q5 provide also a set of graphs for an easier application of this
theory.
(All of these capacity terms are to be calculated by any useful
capacity formula, e.g. the Siegloch-formula, Equation 8.33)
8.9 All-Way Stop Controlled Intersections
8.9.1 Richardson’s Model a service time equal to the follow-up headway for vehicles in this
approach if there are no conflicting vehicles on the cross roads
Richardson (1987) developed a model for all-way stop (to the left and right). The average service time is the time
controlled intersections (AWSC) based on M/G/1 queuing between successive approach stream vehicles being able to
theory. He assumed that a driver approaching will either have depart. If there were conflicting vehicles then the conflicting
8 - 38
vehicles at the head of their queues will depart before the The probability of no conflicting vehicles being 1–ns given by
approach stream being analysed. Consequently, Richardson
assumed that if there were conflicting vehicles then the average 1–ns = (1–s) (1–n) (8.113)
service time is the sum of the clearance time, Tc, for conflicting
vehicles and for the approach stream. hence,
For simplicity, Richardson considered two streams; northbound ns = 1 – (1–qn sn) (1–qs ss) (8.114)
and westbound. Looking at the northbound drivers, the
probability that there will be a conflicting vehicle on the cross and
road is given by queuing theory as w. The average service time ew = 1 – (1–qe se) (1–qw sw) (8.115)
for northbound drivers is then
Given the flows, qn, qs , qe , and qw and using an estimate of
sn = tm (1–w) + Tc w (8.106) service times, pns and pew can be estimated using Equations
8.114 and 8.115. The iterative process is continued with
A similar equation for the average service time for westbound Equations 8.109 to 8.112 providing a better estimate of the
drivers is service times, sn, ss, se, and sw.
sw = tm (1–e) + Tc e (8.107) Richardson used Herbert’s (1963) results in which tm was found
where, to be 4 sec and Tc was a function of the number of cross flow
i is the utilization ratio and is qi si, lanes to be crossed. The equation was
qi is the flow from approach i, tc
3.6 0.1 number of lanes
si is the service time for approach i
tm is the minimum headway, and
Tc is the total clearance time.
and Tc is the sum of thec t values for the conflicting and the
These equations can be manipulated to give a solution for sn as approach streams.
The steady-state average delay was calculated using the

qwtmTctm qwtm
2
Pollaczek-Khintchine formula with Little’s equation as:
sn
(8.108)
1 qwqn(Tc 2tmTctm)
2 2
2 2q 2Var(s)
Ws
(8.116)
2(1 )q
If there are four approaches then very similar equations are

obtained for the average service time involving the probability or
there are no cars on either conflicting stream. For instance, 2
1 q Var(s)
1 2
sn = tm (1–ew) + Tc ew (8.109) Ws
q 2(1 )
ss = tm (1–ew) + Tc ew (8.110)
This equation requires an estimate of the variance of the service

se = tm (1–ns) + Tc ns (8.111)
times. Here Richardson has assumed that drivers either had a
service time of hm or Tc. For the northbound traffic, there were
sw = tm (1–ns) + Tc ns (8.112) (1– ew) proportion of drivers with a service time of exactlymt
and ew drivers with a service time of exactly Tc . The variance
is then
8 - 39
a four way stop with single lane approaches is given in Figure

8.25. Here the southbound traffic has been set to 300 veh/h.
Var(s)n
tm(1 ew)Tc sn
2 2 2
(8.117) The east-west traffic varies but with equal flows in both
directions. In accordance with the comments above, tm was 4 sec
and Tc was 2*tc or 7.6 sec.
and
Richardson's approach is satisfactory for heavy flows where most
drivers have to queue before departing. His approach has been
sn tm extended by Horowitz (1993), who extended the number of

(8.118) maneuver types and then consequently the number of service
Tc tm time values. Horowitz has also related his model to Kyte’s
(1989) approach and found that his modified Richardson model
compared well with Kyte’s empirical data.
This then gives Figure 8.25 from Richardson's research, gives the performance
Tc sn sn tm as the traffic in one set approaches (north-south or east-
Var(s)n
tm Tc2 sn2
2
(8.119) west)increases. Typically, as traffic flow in one direction
Tc tm Tc tm increases so does the traffic in the other directions. This will
usually result in the level of delays increasing at a more rapid
rate than the depicted in this figure.
for the northbound traffic. Similar equations can be obtained for
the other approaches. An example of this technique applied to
Figure 8.25
Average Delay For Vehicles on the Northbound Approach.
8 - 40
8.10 Empirical Methods
Empirical models often use regression techniques to quantify an or even other characteristic values of the intersection layout by
element of the performance of the intersection. These models, by another set of linear regression analysis (see e.g. Kimber and
their very nature, will provide good predictions. However, at Coombe 1980).
times they are not able to provide a cause and effect The advantages of the empirical regression technique compared
relationships. to gap acceptance theory are:
Kimber and Coombe (1980), of the United Kingdom, have there is no need to establish a theoretical model.
evaluated the capacity of the simple 2-stream problem using reported empirical capacities are used.
empirical methods. The fundamental idea of this solution is as influence of geometrical design can be taken into account.
follows: Again, we look at the simple intersection (Figure 8.7) effects of priority reversal and forced priority are taken into
with one priority traffic stream and one non-priority traffic account automatically.
stream during times of a steady queue (i.e. at least one vehicle is there is no need to describe driver behavior in detail.
queuing on the minor street). During these times, the volume of
traffic departing from the stop line is the capacity. This capacity The disadvantages are:
should depend on the priority traffic volume qp during the same
time period. To derive this relationship, observations of traffic transferability to other countries or other times (driver
operations of the intersection have to be made during periods of behavior might change over time) is quite limited: For
oversaturation of the intersection. The total time of observation application under different conditions, a very big sample
then is divided into periods of constant duration, e.g. 1 minute. size must always be evaluated.
During these 1-minute intervals, the number of vehicles in both no real understanding of traffic operations at the intersection
the priority flow and the entering minor street traffic are counted. is achieved by the user.
Normally, these data points are scattered over a wide range and the equations for four-legged intersections with 12
are represented by a linear regression line. On average, half of movements are too complicated.
the variation of data points results from the use of one-minute the derivations are based on driver behavior under
counting intervals. In practice, evaluation intervals of more than oversaturated conditions.
1-minute (e.g. 5-minutes) cannot be used, since this normally each situation to be described with the capacity formulae
leads to only few observations. must be observed in reality. On one hand, this requires a
large effort for data collection. On the other hand, many of
As a result, the method would produce linear relations for qm: the desired situations are found infrequently, since
congested intersections have been often already signalized.
qm = b - c . qp (8.120)
Instead of a linear function, also other types of regression could 8.10.1 Kyte's Method
be used as well, e.g.
Kyte (1989) and Kyte et al. (1991) proposed another method for
qm = A . e-Bx . (8.121) the direct estimation of unsignalized intersection capacity for
both AWSC and TWSC intersections. The idea is based on the
Here, the regression parameters A and B could be evaluated out fact that the capacity of a single-channel queuing system is the
of the data points by adequate regression techniques. This type inverse of the average service time. The service time, tW, at the
of equation is of the same form as Siegloch's capacity formula unsignalized intersection is the time which a vehicle spends in
(Equation 8.33). This analogy shows that A=3600/tf . the first position of the queue. Therefore, only the average of
these times (tW) has to be evaluated by observations to get the
In addition to the influence of priority stream traffic volumes on capacity.
the minor street capacity, the influence of geometric layout of the
intersection can be investigated. To do this, the constant values Under oversaturated conditions with a steady queue on the minor
b and c or A and B can be related to road widths or visibility street approach, each individual value of this time in the first
8 - 41
position can easily be observed as the time between two PLT and PRT are the proportion of left and right turners;
consecutive vehicles crossing the stop line. In this case, hHV-adj is the adjustment factor for heavy vehicles; and
however, the observations and analyses are equivalent to the PHV is the proportion of heavy vehicles.
empirical regression technique .
The average departure headway, d , is first assumed to be four
Assuming undersaturated conditions, however, the time each of seconds and the degree of saturation, x , is the product of the
the minor street vehicles spends in the first position could be flow rate, V and d . A second iterative value of d is given by
measured as well. Again, the inverse of the average of these the equation:
times is the capacity. Examples of measured results are given by
M P(C )h
5
Kyte et al. (1991). d
i i
i
1
From a theoretical point of view, this method is correct. The
problems relate to the measurement techniques (e.g. by video
taping). Here it is quite difficult to define the beginning and the where P(Ci) is the probability that conflict Ci occurs. These
end of the time spent in the first position in a consistent way. If values also depend on estimates of d and the hi values. The
this problem is solved, this method provides an easy procedure service time is given by the departure headway minus the move-
for estimating the capacity for a movement from the minor street up time.
even if this traffic stream is not operating at capacity.
Kyte et al. (1996) recognizes that capacity can be evaluated from
Following a study of AWSC intersections, Kyte et al. (1996) two points of view. First, the capacity can be estimated
developed empirical equations for the departure headways from assuming all other flows remain the same. This is the approach
an approach for different levels of conflict. that is typically used in Section 8.4.1. Alternatively capacity can
be estimated assuming the ratio of flow rates for different
hi = hb-i + hLT-adj PLT + hRT-adj PRT + hHV-adj PHV (8.122) movements for all approaches remain constant. All flows are
where: incrementally increased until one approach has a degree of
saturation equal to one.
hi is the adjusted saturation headway for the
degree of conflict case i; The further evaluation of these measurement results corresponds
hb-i is the base saturation headway for case i; to the methods of the empirical regression techniques. Again,
hLT-adj regression techniques can be employed to relate the capacity
and hRT-adj are the headway adjustment factors for left estimates to the traffic volumes in those movements with a
and right turners respectively; higher rank of priority.
8.11 Conclusions
This chapter describes the theory of unsignalized intersections has been extended to predict delays in the simpler
which probably have the most complicated intersection control conditions.
mechanism. The approaches used to evaluate unsignalized
intersections fall into three classes. (b) Queuing theory in which the service time attributes are
described. This is a more abstract method of describing
(a) Gap acceptance theory which assumes a mechanism for driver departure patterns. The advantages of using
drivers departure. This is generally achieved with the queuing theory is that measures of delay (and queue
notion of a critical gap and a follow on time. This lengths) can be more easily defined for some of the more
attributes of the conflicting stream and the non priority complicated cases.
stream are also required. This approach has been
successfully used to predict capacity (Kyte et al. 1996) and
8 - 42
(c) Simulation programs. These are now becoming more Research in these three approaches will undoubtably continue.
popular. However, as a word of caution, the output from New theoretical work highlights parameters or issues that should
these models is only as good as the algorithms used in be considered further. At times, there will be a number of
the model, and some simpler models can give excellent counter balancing effects which will only be identified through
results. Other times, there is a temptation to look at the theory.
output from models without relating the results to the
existing theory. This chapter describes most of the The issues that are likely to be debated in the future include the
theories for unsignalised intersections and should assist extent that one stream affects another as discussed in Section
simulation modelers to indicate useful extension to 8.6; the similarities between signalized and unsignalized
theory. intersections; performance of oversaturated intersection and
variance associated with the performance measures.
References
Abou-Henaidy, M., S. Teply, and J. D. Hunt (1994). Gap Brilon, W. (1995). Delays at Oversaturated Unsignalized
Acceptance Investigations in Canada. Proceedings of the Intersections based on Reserve Capacities. Paper presented
Second International Symposium on Highway Capacity. at the TRB Annual Meeting, Washington D.C.
Australian Road Research Board- Transportation Research Brilon, W. and M. Grossmann (1991). The New German
Board, Vol 1. pp. 1-20. Guideline for Capacity of Unsignalized Intersections. In:
Adams (1936). Road Traffic Considered as a Random Series. Intersections without Traffic Signals II (Ed.: W. Brilon).
Journal of the Institute of Civil Engineers, London. Springer Publications, Berlin.
Vol. 4, pp. 121-130. Brilon, W, N. Wu, and K. Lemke (1996). Capacity at
Akcelik, R. and R. J. Troutbeck (1991). Implementation of the Unsignalized Two-Stage Priority Intersections. Paper
Australian Roundabout Analysis Method in SIDRA. In: 961280. Presented at the 75th TRB Annual Meeting,
U. Brannolte (ed.) Highway Capacity and Level of Service, Washington D.C.
Proceedings of the International Symposium on Highway Borel, G. (1942). Sur L'emploi du Théorème De Bernoulli
Capacity, Karlsruhe, A. A. Balkema, Rotterdam, pp.17-34. Pour Faciliter Le Calcul D'un Infinité De Co-Efficients.
Ashworth, R. (1968). A Note on the Selection of Gap Application Au Problème De L'attente À Un Guichet.
Acceptance Criteria for Traffic Simulation Studies. C. R. Acad. Sci. Paris Vol. 214, pp. 452-456.
Transportation Research, 2, pp. 171- 175, 1968. Buckley, D. J. (1962). Road Traffic Headway Distribution.
Boehm, H. (1968). Die Anwendung der Monte-Carlo- Proceedings, 1st ARRB Conf., Vol. 1(1), pp. 153-186.
Methode in der Strassenverkehrstechnik. Teil I: Buckley, D. J. (1968). A Semi-Poisson Model of Traffic Flow.
Untersuchungen an ungesteuerten Knoten (Application of Transportation Science, Vol. 2(2), pp. 107-132.
the Monte- Carlo-Method in Traffic Engineering. Part I: Catchpole, E. A. and A. W. Plank (1986). The Capacity of a
Investigations at Unsignalized Intersections). Priority Intersection. Transportation Research Board, 20B
Schriftenreihe Strassenbau und Strassenverkehrstechnik, (6), pp. 441-456
Vol. 73. Bonn 1968. Chan, P. L. and S. Teply, S. (1991). Simulation of Multilane
Brilon, W. (1988). Recent Developments in Calculation Stop-Controlled T-Intersections by Knosimo in Canada. In:
Methods for Unsignalized Intersections in West Germany. Intersections without Traffic Signals II (Ed.: W. Brilon).
In: Intersections Without Traffic Signals (Ed.: W. Brilon). Springer Publications, Berlin.
Springer publications, Berlin 1988. Cowan, R. J. (1975). Useful Headway Models.
Brilon, W. (Ed.) (1991). Intersections Without Traffic Signals Transportation Research, 9(6), pp. 371-375.
II. Springer Publications, Berlin. Cowan, R. J. (1987). An Extension of Tanner's Results on
Uncontrolled Intersections. Queuing Systems, Vol 1., pp.
249-263.
8 - 43
Daganzo, C. F. (1977). Traffic Delay at Unsignalized Hewitt, R. H. (1985). A Comparison Between Some Methods
Intersections: Clarification of Some Issues. Transportation of Measuring Critical Gap. Traffic Engineering & Control,
Science, Vol. 11. 26(1), pp. 13-21.
Dawson, R. F. (1969). The Hyperlang Probability Heidemann, D. (1991). Queue Lengths and Waiting-Time
Distribution - A Generalized Traffic Headway Model. Distributions at Priority Intersections. Transportation
Proceedings of the FourthISTTT in Karsruhe, Strassenbau Research B, Vol 25B, (4) pp. 163-174.
und Strassenverkehehrstechnik, No 89 1969, pp 30-36. Horowitz, A. J. (1993). Revised Queueing Model of Delay at
Drew, D. R. (1968). Traffic Flow Theory and Control. All-Way Stop-Controlled Intersections. Transportation
McGraw-Hill Book Company, New York. Research Record, 1398, pp. 49-53.
FGSV (1991). Merkblatt zur Berechnung der Jacobs, F. (1979). Leistungsfähigkeitsberechnung von
Leistungsfähigkeit von Knotenpunkten ohne Knotenpunkten ohne Lichtsignalanlagen (Capacity
Lichtsignalanlagen (Guideline for the Estimation of Calculations for Unsignalised Intersections).
Capacity at Unsignalized Intersections). German Vorlesungsmanuskript, Stuttgart.
Association for Roads and Traffic (FGSV), Cologne. Kimber, R. M. and E. M. Hollis (1979). Traffic Queues and
Fisk, C. S. (1989). Priority Intersection Capacity - A Delays at Road Junctions. TRRL Laboratory Report,
Generalization of Tanner's Formula. Transportation LR909.
Research, Vol. 23 B, pp. 281-286. Kimber, R. M. and R. D. Coombe (1980). The Traffic
Gipps, P. G. (1982). Some Modified Gap Acceptance Capacity of Major/Minor Priority Junctions. TRRL Report
Formulae. Proceedings of the 11th ARRB Conference, Vol. SR 582.
11, part 4, p.71. Kremser, H. (1962). Ein Zusammengesetztes
Grossmann, M. (1988). Knosimo - a Practicable Simulation Wartezeitproblem Bei Poissonschen Verkehrsströmen (A
Model for Unsignalized Intersections. In: Intersections Complex Problem of Delay with Poisson-Distributed Traffic
without Traffic Signals (Ed.: W. Brilon). Springer Flows). Osterreichisches Ingenieur-Archiv, 16.
Publications, Berlin. Kremser, H. (1964). Wartezeiten Und Warteschlangen Bei
Grossmann, M. (1991). Methoden zur Berechnung und Einfadelung Eines Poissonprozesses in Einen Anderen
Beurteilung von Leistungsfähigkeit und Verkehrsqualitat an Solchen Prozess (Delays and Queues with One Poisson
Knotenpunkten ohne Lichtsignalanlagen (Methods for Process Merging into Another One). Österreichisches
Calculation and Judgement of Capacity and Traffic Quality Ingenieur-Archiv, 18.
at Intersections Without Traffic Signals). Ruhr-University Kyte, M. (1989). Estimating Capacity and Delay at an All-
Bochum, Germany, Chair of Traffic Engineering, Vol. 9,. Way Stop-Controlled Intersection. Research Report,
Haight, F. A. and M. A. Breuer, (1960). The Borel-Tanner TRANSNOW, Moscow/Idaho.
Distribution. Biometrika, Vol. 47 (1 and 2), pp. 143-150. Kyte, M., J. Zegeer, and K. Lall (1991). Empirical Models for
Harders, J. (1968). Die Leistungsfähigkeit Nicht Estimating Capacity and Delay at Stop-Controlled
Signalgeregelter Städtischer Verkehrsknoten (The Capacity Intersections in the United States. In: Intersections Without
of Unsignalized Urban Intersections). Schriftenreihe Traffic Signals II (Ed.: W. Brilon), Springer Publications,
Strassenbau und Strassenverkehrstechnik, Vol. 76. Berlin.
Harders, J. (1976). Grenz- und Folgezeitlücken als Grundlage Kyte et al. (1996). Capacity and Level of Service at
für die Leistungsfähigkeit von Landstrassen (Critical Gaps Unsignalized Intersections. Final Report for National
and Move-Up Times as the Basis of Capacity Calculations Cooperative Highway Research Program Project 3-46.
for Rural Roads). Schriftenreihe Strassenbau und Little, J. (1961). A Proof of the Queueing Formula L = W.
Strassenverkehrstechnik, Vol. 216. Operations Research 9, pp. 383-387.
HCM (1985). Highway Capacity Manual. Transportation McDonald, M. and D. J. Armitage (1978). The Capacity of
Research Board, Special Report 209, Washington, DC. Roundabouts. Traffic Engineering & Control. Vol. 19(10),
Herbert, J. (1963). A Study of Four Way Stop Intersection pp. 447-450.
Capacities. Highway Research Record, 27, Highway
Research Board, Washington, DC.
Hewitt, R. H. (1983). Measuring Critical Gap.
Transportation Science, 17(1), pp. 87-109.
8 - 44
Miller, A. J. (1972). Nine Estimators of Gap Acceptance Tanner, J. C. (1962). A Theoretical Analysis of Delays At An
Parameters. In: Traffic Flow and Transportation (Ed. Uncontrolled Intersection. Biometrica 49(1 and 2),
Newell). Proceedings International Symposium on the pp. 163-70.
Theory of Traffic Flow and Transportation, American Tanner, J. C. (1967). The Capacity of an Uncontrolled
Elsevier Publishing Co. Intersection. Biometrica, 54(3 and 4), pp. 657-658.
Morse, P. M. (1962). Queues, Inventories and Maintenance. Tracz, M. (1991). Polish Guidelines for Capacity Analysis of
John Wiley. Priority Intersections. In: Intersections Without Traffic
Newell, G. F. (1971). Applications of Queueing Theory. Signals II (Ed.: W. Brilon), Springer Publications, Berlin.
Chapman and Hall Ltd., London. Troutbeck, R. J. (1975). A Review of the Ramsey-Routledge
Newell, G. F. (1982). Applications of Queueing Theory. 2nd Method for Gap Acceptance Times. Traffic Engineering &
Ed. Chapman and Hall Ltd., London. Control, 16(9), pp. 373-375.
Plank, A. W. and E. A. Catchpole (1984). A General Capacity Troutbeck. R. J. (1986). A verage Delay at an Unsignalized
Formula for an Uncontrolled Intersection. Traffic Intersection with Two Major Streams Each Having a
Engineering Control. 25(6), pp. 327-329. Dichotomized Headway Distribution. Transportation
Poeschl, F. J. (1983). Die Nicht Signalgesteuerte Science, 20(4), pp. 272-286.
Nebenstrassenzufahrt Als Verallgemeinertes M/G/ 1- Troutbeck, R. J. (1988). Current and Future Australian
Warteschlangensystem (The Unsignalized Minor Street Practices for the Design of Unsignalized Intersections. In:
Entry as a Generalized M/G/1 Queueing System). Zeitschrift Intersections without Traffic Signals (Ed.: W. Brilon),
für Operations Research, Vol. 27 B. Springer Publications, Berlin.
Ramsey, J. B. H. and I. W. Routledge (1973). A New Troutbeck, R. J. (1990). Roundabout Capacity and
Approach to the Analysis of Gap Acceptance Times. Traffic Associated Delay. In: Transportation and Traffic Theory.
Engineering Control, 15(7), pp. 353-357. Proceedings of the Eleventh International Symposium on
Richardson, A. J. (1987). A Delay Model for Multiway Stop- Transportation and Traffic Theory, in Yokohama, Japan (Ed:
Sign Intersections. Transportation Research Record, 1112, M Koshi), Elsevier.
pp. 107-112. Troutbeck, R. J. (1991). Unsignalized Intersection and
Salter, R. J. (1982). Simulation of Delays and Queue Lengths Roundabouts in Australia: Recent Developments. In:
at Over-Saturated Priority Highway Junctions. Australian Intersections without Traffic Signals II (Ed.: W. Brilon),
Road Research, Vol. 12. No. 4. Springer Publications, Berlin.
Schuhl, A. (1955). The Probability Theory Applied to the Sullivan, D. and R. J. Troutbeck (1993). Relationship
Distribution of Vehicles on Two-Lane Highways. Poisson Between the Proportion of Free Vehicles and Flow Rate on
and Traffic. The Eno Foundation for Highway Traffic Arterial Roads. Physical Infrastructure Centre Report, 92-
Control. Sangatuck, CT. 21, Queensland University of Technology, Brisbane.
Siegloch, W. (1973). Die Leistungsermittlung an Wegmann, H. (1991). A General Capacity Formula for
Knotenpunkten Ohne Lichtsignalsteuerung (Capacity Unsignalized Intersections. In: Intersections without Traffic
Calculations for Unsignalized Intersections). Schriftenreihe Signals II (Ed.: W. Brilon), Springer Publications, Berlin
Strassenbau und Strassenverkehrstechnik, Vol. 154. 1991.
Steierwald, G. (1961). Die Anwendung Der Monte-Carlo- Wu, N. (1994). An Approximation for the Distribution of
Methode in Der Strassenverkehrsforschung (The Queue Lengths at Signalized Intersections. Proceedings of
Application of the Monte-Carlo Method in Traffic the Second International Symposium on Highway Capacity.
Engineering Research). Habilitation thesis, RWTH Aachen. Australian Road Research Board - Transportation Research
Steierwald, G. (1961). Die Leistunsfähigkeit von Board, Vol 2., pp. 717-736.
Knotenpurlkten des Strassenverkehrs. Schriftenreihe Yeo, G. F. (1962). Single-Server Queues with Modified
Strassenbau und Strassenverkehrstechnik, Vol. 11, Bonn. Service Mechanisms. Journal Australia Mathematics
Tanner, J. C. (1953). A Problem of Interference Between Two Society, Vol 2, pp. 499-502.
Queues. Biometrika, 40, pp. 58-69. Yeo, G. F. and B. Weesakul (1964). Delays to Road Traffic
Tanner, J. C. (1961) A Derivation of the Borel Distribution. at an Intersection. Journal of Applied Probability,
Biometrica, Vol. 48, p. 222. pp. 297-310.
8 - 45
Zhang, X. (1988). The Influence of Partial Constraint on Chodur, J. and S. Gaca (1988). Simulation Studies of the
Delay at Priority Junctions. In: Intersections without Effects of Some Geometrical and Traffic Factors on the
Traffic Signals (Ed.: W. Brilon), Springer Publications, Capacity of Priority Intersections. In: Intersections Without
Berlin. Traffic Signals (Ed.: W. Brilon). Springer Publications,
Berlin.
Cowan, R. J. (1984). Adams' Formula Revised. Traffic
Additional References Engineering & Control, Vol. 25(5), pp 272- 274.
Fisk C. S. and H. H. Tan (1989). Delay Analysis for Priority
(Not cited in the Chapter 8) Intersections. Transportation Research, Vol. 23 B,
pp 452-469.
Anveden, J. (1988). Swedish Research on Unsignalized Fricker, J. D., M. Gutierrez, and D. Moffett (1991). Gap
Intersections. In: Intersections Without Traffic Signals (Ed.: Acceptance and Wait Time at Unsignalized Intersections.
W. Brilon). Springer Publications, Berlin. In: Intersections without Traffic Signals II (Ed.: W. Brilon).
Ashworth, R. (1969). The Capacity of Priority-Type Springer Publications, Berlin.
Intersections with a Non-Uniform Distribution of Critical Hansson, A. (1978). Swedish Capacity Manual. Statens
Acceptance Gaps. Transportation Research, Vol. 3. Vägverk (National Road Administration) Intern Rapport NR
Ashworth, R. (1970). The Analysis and Interpretation of Gap 24.
Acceptance Data. Transportation Research. No. 4, Hansson, A. (1987). Berakningsmetoder För Olika Effektmått
pp. 270-280. I Korningar, Del Iii: Korsningar Utan Trafiksignaler.
Baas, K. G. (1987). The Potential Capacity of Unsignalized (Procedures for Estimating Performance Measures of
Intersections. ITE-Journal, pp. 43-56. Intersections. Part III: Intersections Without Signal
Bang, K. L., A. Hansson, and B. Peterson (1978). Swedish Control.) Statens Vägverk (National Road Administration).
Capacity Manual. Transportation Research Record 667, pp Hansson, A. and T. Bergh (1988). A New Swedish Capacity
1-28. Manual/CAPCAL 2. Proceedings 14th Australian Road
Brilon, W. (1981). Anmerkungen Zur Leistungsermittlung von Research Board Conference, Vol. 14(2), pp. 38-47.
Knotenpunkten Ohne Lichtsignalanlagen. (Some Notes on Hawkes, A. G. (1965). Queueing for Gaps in Traffic.
Capacity Calculations for Unsignalized Intersections). Biometricia 52 (1 and 2). pp. 79-85.
Strassenverkehrstechnik, Heft 1, pp. 20-25. Hondermarcq, H. (1968). Essais De Priorité Á Gancle. 9th
Brilon, W. (Ed.) (1988). Intersections Without Traffic Signals. Intentional Study Week on Road Traffic Flow and Safety.
Springer Publications, Berlin. Munich.
Brilon, W. and M. Grossmann (1989). Entwicklung Eines Japan Society of Traffic Engineers (1988). The Planning and
Simulationsmodells Für Knotenpunkte Ohne Design of At-Grade Intersections.
Lichtsignalanlagen (Development of a Simulation Model Jessen, G. D. (1968). Ein Richtlinienvorschlag Für Die
for Intersections Without Traffic Signals). Schriftenreihe Behandlung Der Leistungsfähigkeit von Knotenpunkten
Strassenbau und Strassenverkehrstechnik, Vol. 554, Bonn. Ohne Signalregelung (A Guideline Suggested for Capacity
Brilon, W. and M. Grossmann (1991). Aktualisiertes Calculations for Unsignalized Intersections).
Berechnungsverfahren Für Knotenpunkte Ohne Strassenverkehrstechnik, No. 7/8.
Lichtsignalanlagen (Actualized Calculation Procedure for Jirava, P. and P. Karlicky (1988). Research on Unsignalized
Intersections Without Traffic Signals). Schriftenreihe Intersections with Impact on the Czechoslovak Design
Strassenbau und Strassenverkehrstechnik, Vol. 596, Bonn. Standard. In: Intersections without Traffic Signals (Ed.: W.
Brilon, W., M. Grossmann, and B. Stuwe (1991). Towards a Brilon). Springer Publications, Berlin.
New German Guideline for Capacity of Unsignalized Kimber, R. M., M. Marlow, and E. W. Hollis (1977).
Intersections. Transportation Research Record, No. 1320, Flow/Delay Relationships for Major/Minor Priority
Washington, DC. Junctions. Traffic Engineering & Control, 18(11),
Catling, I. (1977). A Time Dependent Approach to Junction pp. 516-519.
Delays. Traffic Engineering & Control, Vol. 18(11).
Nov. 77, pp. 520-523, 536.
8 - 46
Kimber, R. M., I. Summersgill, and I. J. Burrow (1986). Tracz, M., J. Chodur, and St. Gondek (1990). The Use of
Delay Processes at Unsignalized Junctions: The Reserve Capacity and Delay as the Complementary
Interrelation Between Queueing and Geometric and Measures of Junction Effectiveness. Proceedings of the
Queueing Delay. Transportation Research Board, 20B(6), International Symposium 'Transportation and Traffic
pp. 457-476. Theory", Yokohama.
Lassarre, S., P. Lejeune, and J. C. Decret (1991). Gap Tracz, M. and J. Chodur (1991). Comparative Analysis of
Acceptance and Risk Analysis at Unsignalized Intersections. Major/Minor Priority Intersection Capacity Methods. In:
In: Intersections Without Traffic Signals II, Springer- Highway Capacity and Level of Service, A. A. Balkema,
Verlag, Berlin. Rotterdam.
Middelham, F. (1986). Manual for the Use of FLEXSYT-I TRB (1991). Interim Material on Unsignalized Intersection
With FLEXCOL-76. Ministerie van Verkeer en Waterstaat, Capacity. Transportation Research Circular No. 373.
The Hague. Troutbeck, R. J. (1989). Evaluating the Performance of a
Semmens, M. C. (1985). PICADY 2: an Enhanced Program Roundabout. Australian Road Research Board Special
to Model Capacities, Queues and Delays at Major/Minor Report, 45.
Priority Junctions. TRRL Report RR36. Troutbeck, R. J. (1990). Traffic Interactions at Roundabouts.
Siegloch, W. (1974). Ein Richtlinienvorschlag Zur Proceedings of the 15th Australian Road Research Board
Leistungsermittlung an Knotenpunkten Ohne Conference, 15(5), pp. 17-42.
Lichtsignalsteuerung (Capacity Calculations for Troutbeck, R. J. (1992). Estimating the Critical Acceptance
Unsignalized Intersections). Strassenverkehrstechnik, Gap from Traffic Movements. Research Report, 92-5.
Vol. 1. Turner, D. J., R. Singh, and Y. H. Cheong (1984). The
Tracz, M. (1988). Research of Traffic Performance of Development of Empirical Equations for Capacity Analysis
Major/Minor Priority Intersections. In: Intersections at Priority Junctions in Singapore. Singapore Transport,
Without Traffic Signals (Ed.: W. Brilon), Springer Vol. 3, No. 4, pp. 17-21.
Publications, Berlin. Vasarhely, B. (1976). Stochastic Simulation of the Traffic of
an Uncontrolled Road Intersection. Transportation
Research, Vol. 10.
8 - 47

TRAFFIC FLOW AT
SIGNALIZED INTERSECTIONS
BY NAGUI ROUPHAIL15
ANDRZEJ TARKO16
JING LI17
15
Professor, Civil Engineering Department, North Carolina State University, Box 7908, Raleigh, NC
276-7908
16
Assistant Professor, Purdue University, West LaFayette, IN 47907
17
Principal, TransSmart Technologies, Inc., Madison, WI 53705
variance of the number of arrivals per cycle

I
mean number of arrivals per cycle
Ii = cumulative lost time for phase i (sec)

L = total lost time in cycle (sec)
q = A(t) = cumulative number of arrivals from beginning of cycle starts until t,
B = index of dispersion for the departure process,
variance of number of departures during cycle
B
mean number of departures during cycle
c = cycle length (sec)

C = capacity rate (veh/sec, or veh/cycle, or veh/h)
d = average delay (sec)
d1 = average uniform delay (sec)
d2 = average overflow delay (sec)
D(t) = number of departures after the cycle starts until time t (veh)
eg = green extension time beyond the time to clear a queue (sec)
g = effective green time (sec)
G = displayed green time (sec)
h = time headway (sec)
i = index of dispersion for the arrival process
q = arrival flow rate (veh/sec)
Q0 = expected overflow queue length (veh)
Q(t) = queue length at time t (veh)
r = effective red time (sec)
R = displayed red time (sec)
S = departure (saturation) flow rate from queue during effective green (veh/sec)
t = time
T = duration of analysis period in time dependent delay models
U = actuated controller unit extension time (sec)
Var(.) = variance of (.)
Wi = total waiting time of all vehicles during some period of time i
x = degree of saturation, x = (q/S) / (g/c), or x = q/C
y = flow ratio, y = q/S
Y = yellow (or clearance) time (sec)
= minimum headway
9.
TRAFFIC FLOW AT SIGNALIZED INTERSECTIONS
9.1 Introduction
The theory of traffic signals focuses on the estimation of delays have received greater attention since the pioneering work by
and queue lengths that result from the adoption of a signal Webster (1958) and have been incorporated in many
control strategy at individual intersections, as well as on a intersection control and analysis tools throughout the world.
sequence of intersections. Traffic delays and queues are
principal performance measures that enter into the This chapter traces the evolution of delay and queue length
determination of intersection level of service (LOS), in the models for traffic signals. Chronologically speaking, early
evaluation of the adequacy of lane lengths, and in the estimation modeling efforts in this area focused on the adaptation of steady-
of fuel consumption and emissions. The following material state queuing theory to estimate the random component of delays
emphasizes the theory of descriptive models of traffic flow, as and queues at intersections. This approach was valid so long as
opposed to prescriptive (i.e. signal timing) models. The the average flow rate did not exceed the average capacity rate.
rationale for concentrating on descriptive models is that a better In this case, stochastic equilibrium is achieved and expectations
understanding of the interaction between demand (i.e. arrival of queues and delays are finite and therefore can be estimated by
pattern) and supply (i.e. signal indications and types) at traffic the theory. Depending on the assumptions regarding the
signals is a prerequisite to the formulation of optimal signal distribution of traffic arrivals and departures, a plethora of
control strategies. Performance estimation is based on steady-state queuing models were developed in the literature.
assumptions regarding the characterization of the traffic arrival These are described in Section 9.3 of this chapter.
and service processes. In general, currently used delay models
at intersections are described in terms of a deterministic and As traffic flow rate approaches or exceeds the capacity rate, at
stochastic component to reflect both the fluid and random least for a finite period of time, the steady-state models
properties of traffic flow. assumptions are violated since a state of stochastic equilibrium
cannot be achieved. In response to the need for improved
The deterministic component of traffic is founded on the fluid estimation of traffic performance in both under and oversaturated
theory of traffic in which demand and service are treated as conditions, and the lack of a theoretically rigorous approach to
continuous variables described by flow rates which vary over the the problem, other methods were pursued. A prime example is
time and space domain. A complete treatment of the fluid the time-dependent approach originally conceived by Whiting
theory application to traffic signals has been presented in (unpublished) and further developed by Kimber and Hollis
Chapter 5 of the monograph. (1979). The time-dependent approach has been adopted in many
capacity guides in the U.S., Europe and Australia. Because it is
The stochastic component of delays is founded on steady-state currently in wide use, it is discussed in some detail in Section 9.4
queuing theory which defines the traffic arrival and service time of this chapter.
distributions. Appropriate queuing models are then used to
express the resulting distribution of the performance measures. Another limitation of the steady-state queuing approach is the
The theory of unsignalized intersections, discussed in Chapter 8 assumption of certain types of arrival processes (e.g Binomial,
of this monograph, is representative of a purely stochastic Poisson, Compound Poisson) at the signal. While valid in the
approach to determining traffic performance. case of an isolated signal, this assumption does not reflect the
impact of adjacent signals and control which may alter the
Models which incorporate both deterministic (often called pattern and number of arrivals at a downstream signal. Therefore
uniform) and stochastic (random or overflow) components of performance in a system of signals will differ considerably from
traffic performance are very appealing in the area of traffic that at an isolated signal. For example, signal coordination will
signals since they can be applied to a wide range of traffic tend to reduce delays and stops since the arrival process will be
intensities, as well as to various types of signal control. They are different in the red and green portions of the phase. The benefits
approximations of the more theoretically rigorous models, in of coordination are somewhat subdued due to the dispersion of
which delay terms that are numerically inconsequential to the platoons between signals. Further, critical signals in a system
final result have been dropped. Because of their simplicity, they could have a metering effect on traffic which proceeds
9-1
downstream. This metering reflects the finite capacity of the without reference to their impact on signal performance. The
critical intersection which tends to truncate the arrival manner in which these controls affect performance is quite
distribution at the next signal. Obviously, this phenomenon has diverse and therefore difficult to model in a generalized
profound implications on signal performance as well, fashion. In this chapter, basic methodological approaches and
particularly if the critical signal is oversaturated. The impact of concepts are introduced and discussed in Section 9.6. A
upstream signals is treated in Section 9.5 of this chapter. complete survey of adaptive signal theory is beyond the scope of
this document.
With the proliferation of traffic-responsive signal control
technology, a treatise on signal theory would not be complete
9.2 Basic Concepts of Delay Models at Isolated Signals
As stated earlier, delay models contain both deterministic and time is that portion of green where flows are sustained at the
stochastic components of traffic performance. The deterministic saturation flow rate level. It is typically calculated at the
component is estimated according to the following assumptions: displayed green time minus an initial start-up lost time (2-3
a) a zero initial queue at the start of the green phase, b) a seconds) plus an end gain during the clearance interval (2-4
uniform arrival pattern at the arrival flow rate (q) throughout the seconds depending on the length of the clearance phase).
cycle c) a uniform departure pattern at the saturation flow rate
(S) while a queue is present, and at the arrival rate when the A simple diagram describing the delay process in shown in
queue vanishes, and d) arrivals do not exceed the signal capacity, Figure 9.1. The queue profile resulting from this application is
defined as the product of the approach saturation flow rate (S) shown in Figure 9.2. The area under the queue profile
and its effective green to cycle ratio (g/c). The effective green diagram represents the total (deterministic) cyclic delay. Several
Figure 9.1
Deterministic Component of Delay Models.
9-2
Figure 9.2
Queuing Process During One Signal Cycle
(Adapted from McNeil 1968).
performance measures can be derive including the average delay Interestingly, at extremely congested conditions, the stochastic
per vehicle (total delay divided by total cyclic arrivals) the queuing effect are minimal in comparison with the size of
number of vehicle stopped (Qs ), the maximum number of oversaturation queues. Therefore, a fluid theory approach may
vehicles in the queue (Qmax) , and the average queue length be appropriate to use for highly oversaturated intersections.
(Qavg). Performance models of this type are applicable to low This leaves a gap in delay models that are applicable to the
flow to capacity ratios (up to about 0.50), since the assumption range of traffic flows that are numerically close to the signal
of zero initial and end queues is not violated in most cases. capacity. Considering that most real-world signals are timed to
operate within that domain, the value of time-dependent models
As traffic intensity increases, however, there is a increased are of particular relevance for this range of conditions.
likelihood of “cycle failures”. That is, some cycles will begin to
experience an overflow queue of vehicles that could not In the case of vehicle actuated control, neither the cycle length
discharge from a previous cycle. This phenomenon occurs at nor green times are known in advance. Rather, the length of the
random, depending on which cycle happens to experience green is determined partly by controller-coded parameters such
higher-than-capacity flow rates. The presence of an initial queue as minimum and maximum green times, and partly by the pattern
(Qo) causes an additional delay which must be considered in the of traffic arrivals. In the simplest case of a basic actuated
estimation of traffic performance. Delay models based on queue controller, the green time is extended beyond its minimum so
theory (e.g. M/D/n/FIFO) have been applied to account for this long as a) the time headway between vehicle arrivals does not
effect. exceed the controller s unit extension (U), and b) the maximum
green has not been reached. Actuated control models are
discussed further in Section 9.6.
9-3
9.3 Steady-State Delay Models

9.3.1 Exact Expressions The departure process is described by a flexible service mecha-
nism and may include the effect of an opposing stream by defin-
This category of models attempts to characterize traffic delays ing an additional queue length distribution caused by this factor.
based on statistical distributions of the arrival and departure Although this approach leads to expressions for the expected
processes. Because of the purely theoretical foundation of the queue length and expected delay, the resulting models are
models, they require very strong assumptions to be considered complex and they include elements requiring further modeling
valid. The following section describes how delays are estimated such as the overflow queue or the additional queue component
for this class of models, including the necessary data mentioned earlier. From this perspective, the formula is not of
requirements. practical importance. McNeil (1968) derived a formula for the
expected signal delay with the assumption of a general arrival
The expected delay at fixed-time signals was first derived by process, and constant departure time. Following his work, we
Beckman (1956) with the assumption of the binomial arrival express the total vehicle delay during one signal cycle as a sum
process and deterministic service: of two components
c g Q c g1
W
W1 W2,
d
[ o ] (9.1) (9.3)
c(1 q/S) q 2
where, where
W1 = total delay experienced in the red phase and
c = signal cycle, W2 = total delay experienced in the green phase.
g = effective green signal time,
q = traffic arrival flow rate, (c g)
W1
[Q(0) A(t)] dt
S = departure flow rate from queue during green, 20 (9.4)
Qo = expected overflow queue from previous cycles.
The expected overflow queue used in the formula and the and
restrictive assumption of the binomial arrival process reduce the
practical usefulness of Equation 9.1. Little (1961) analyzed the W2
c
Q(t)dt
expected delay at or near traffic signals to a turning vehicle 2(c g) (9.5)
crossing a Poisson traffic stream. The analysis, however, did not
include the effect of turners on delay to other vehicles. Darroch
(1964a) studied a single stream of vehicles arriving at a where,
fixed-time signal. The arrival process is the generalized Poisson Q(t) = vehicle queue at time t,
process with the Index of Dispersion: A(t) = cumulative arrivals at t,
var(A)
I
(9.10) Taking expectations in Equation 9.4 it is found that:
qh
1
E(W1)
(c g) Qo q (c g)2. (9.6)
2
where,
var(.)= variance of ( . )
q = arrival flow rate,
h = interval length, Let us define a random variable Z2 as the total vehicle delay
A = number of arrivals during interval h = qh. experienced during green when the signal cycle is infinite. The
9-4
variable Z2 is considered as the total waiting time in a busy Equations 9.9, 9.11, and 9.12 yield:
period for a queuing process Q(t) with compound Poisson
1
arrivals of intensity q, constant service time 1/S and an initial E(W2)
[(1I q/S q/S)g (c g)
system state Q(t=t0). McNeil showed that provided q/S<1: 2S (1 q/S)2
(1 q/S)(2 q(c g) Qo (9.13)
q 2 (c g)2 q (c g)I)]
(1 I q/S q/S) E[Q(t0)] E[Q 2(t0)]
E(Z2)
. (9.7)
2S (1 q/S) 2 2S (1 q/S)
and using Equations 9.3, 9.4 and 9.13, the following is obtained:
(c g) Q0
Now W2 can be expressed using the variable Z2: E(W)
c g qc 1 1 I
(9.14)
c(1 q/S) q 2 S 1 q/S
E(W2)
E[Z2 Q(t
c g)] E[Z2 Q(t
c)] (9.8)
The average vehicle delay d is obtained by dividing E(W) by the
and average number of vehicles in the cycle (qc):
(1 I g/Sq/S) E[Q(c g) Q(c)] c g 2 1 I
E[W2]
d
[(c g) Qo (1 )] (9.15)
2S (1 q/S)2 2 c(1 q/S) q S 1 q/S
(9.9)
E[Q 2(c g)] E[Q 2(c)]
.
2S (1 q/S)
which is in essence the formula obtained by Darroch when the
departure process is deterministic. For a binomial arrival
The queue is in statistical equilibrium, only if the degree of process I=1-q/S, and Equation 9.15 becomes identical to that
saturation x is below 1, obtained by Beckmann (1956) for binomial arrivals. McNeil
and Weiss (in Gazis 1974) considered the case of the compound
q/S
x
< 1. (9.10) Poisson arrival process and general departure process obtaining
g/c the following model:
(c g) 2 (1 q/S)(1 B 2)
d
(c g) 1 Qo
For the above condition, the average number of arrivals per cycle 2c(1 q/S) q 2S
can discharge in a single green period. In this case E[ Q(0) (9.16)
1 IB 2q/S
] = E [ Q(c) ] and E [ Q 2(0) ] = E [Q 2(c) ]. Also Q (c-g) = (1 )
Q(0) + A(c), so that: S 1 q/S
E[Q(c g) Q(c)]
E[A(c g)]
q(c g) (9.11) An examination of the above equation indicates that in the case
of no overflow (Qo= 0), and no randomness in the traffic process
(I=0), the resultant delay becomes the uniform delay component.
and This component can be derived from a simple input-output
model of uniform arrivals throughout the cycle and departures as
described in Section 9.2. The more general case in Equation
E[Q 2(c g) Q 2(c)]
2 E[A(c g)] E[Q(0)] 9.16 requires knowledge of the size of the average overflow
E[A 2(c g)] queue (or queue at the beginning of green), a major limitation on
(9.12)

2 q(c g) Qo q 2 (c g)2 the practical usefulness of the derived formulae, since these are
q (c g)I usually unknown.
9-5
A substantial research effort followed to obtain a closed-form signal performance, since vehicles are served only during the
analytical estimate of the overflow queue. For example, Haight effective green, obviously at a higher rate than the capacity rate.
(1959) specified the conditional probability of the overflow The third term, calibrated based on simulation experiments, is a
queue at the end of the cycle when the queue at the beginning of corrective term to the estimate, typically in the range of 10
the cycle is known, assuming a homogeneous Poisson arrival percent of the first two terms in Equation 9.17.
process at fixed traffic signals. The obtained results were then
modified to the case of semi-actuated signals. Shortly thereafter, Delays were also estimated indirectly, through the estimation of
Newell (1960) utilized a bulk service queuing model with an Qo, the average overflow queue. Miller (1963) for example ob-
underlying binomial arrival process and constant departure time, tained a approximate formulae for Qo that are applicable to any
using generating function technique. Explicit expressions for arrival and departure distributions. He started with the general
overflow queues were given for special cases of the signal split. equality true for any general arrival and departure processes:
Other related work can be found in Darroch (1964a) who used Q(c)
Q(0) A C C (9.18)
a more general arrival distribution but did not produce a closed
form expression of queue length, and Kleinecke (1964), whose
work included a set of exact but complicated series expansion
for Qo, for the case of constant service time and Poisson arrival where,
process. Q(c) = vehicle queue at the end of cycle,
Q(0) = vehicle queue at the beginning of cycle,
A = number of arrivals during cycle,
9.3.2 Approximate Expressions C = maximum possible number of departures
during green,
The difficulty in obtaining exact expressions for delay which are C = reserve capacity in cycle equal to
reasonably simple and can cover a variety of real world condi- (C-Q(0)-A) if Q(0)+A < C , zero otherwise.
tions, gave impetus to a broad effort for signal delay estimation
using approximate models and bounds. The first, widely used Taking expectation of both sides of Equation 9.18, Miller
approximate delay formula was developed by Webster (1961, obtained:
reprint of 1958 work with minor amendments) from a E(C)
E(C A), (9.19)
combination of theoretical and numerical simulation approaches:
1
c(1 g/c)2 2
d
x 0.65( c2 ) 3 x 25(g/c) (9.17)
2[1 (g/c)x] 2q(1 x) q
since in equilibrium Q(0) = Q(c).
where, Now Equation 9.18 can be rewritten as:

d = average delay per vehicle (sec),
c = cycle length (sec),
g = effective green time (sec), Q(c) [C E(C)]
Q(0) [C A E(C A)] (9.20)
x = degree of saturation (flow to capacity ratio),
q = arrival rate (veh/sec).
Squaring both sides, taking expectations, the following is
The first term in Equation 9.17 represents delay when traffic can obtained:
be considered arriving at a uniform rate, while the second term
E[Q(c)]2 2E[Q(c)] E(C] Var(C)
makes some allowance for the random nature of the arrivals.

This is known as the "random delay", assuming a Poisson arrival (9.21)
E[Q(0)]2 Var(C A)
process and departures at constant rate which corresponds to the
signal capacity. The latter assumption does not reflect actual
9-6
For equilibrium conditions, Equation 9.21 can be rearranged as which can now be substituted in Equation 9.15. Further
follows: approximations of Equation 9.15 were aimed at simplifying it for
practical purposes by neglecting the third and fourth terms which
Var(C A) Var(C) are typically of much lower order of magnitude than the first two
Qo
(9.22)
2E(C A) terms. This approach is exemplified by Miller (1968a) who
proposed the approximate formula:
(1 g/c) 2Q0
d
c(1 g/c) (9.27)
where, 2(1 q/s) q
C = maximum possible number of departures in

one cycle, which can be obtained by deleting the second and third terms in
A = number of arrivals in one cycle, McNeil's formula 9.15. Miller also gave an expression for the
C = reserve capacity in one cycle. overflow queue formula under Poisson arrivals and fixed service
time during the green:
The component Var(C) is positive and approaches 0 when
E(C) approaches E(A). Thus an upper bound on the expected
exp 1.33 Sg(1 x)/x
overflow queue is obtained by deleting that term. Thus: Qo
. (9.28)
2(1 x)
Var(C A)
Qo (9.23)
2E(C A)
Equations 9.15, 9.16, 9.17, 9.27, and 9.28 are limited to specific
arrival and departure processes. Newell (1965) aimed at devel-
For example, using Darroch's arrival process (i.e. E(A)=qc, oping delay formulae for general arrival and departure distribu-
Var(A)=Iqc) and constant departure time during green tions. First, he concluded from a heuristic graphical argument
(E(C)=Sg, Var(C)=0) the upper bound is shown to be: that for most reasonable arrival and departure processes, the
total delay per cycle differs from that calculated with the
Ix
Qo (9.24) assumption of uniform arrivals and fixed service times (Clayton,
2(1 x) 1941), by a negligible amount if the traffic intensity is sufficient-
ly small. Then, by assuming a queue discipline LIFO (Last In
First Out) which does not effect the average delay estimate, he
where x=(qc)/(Sg). concluded that the expected delay when the traffic is sufficiently
heavy can be approximated:
Miller also considered an approximation of the excluded term
c(1 g/c)2 Q
Var(C). He postulates that: d
o. (9.29)
2(1 q/S) q
Var(C)
I (9.25)
E(C A)
This formula gives identical results to formula (Equation 9.15)
if one neglects components of 1/S order in (Equation 9.15) and
and thus, an approximation of the overflow queue is when 1-q/S=1-g/c. The last condition, however, is never met if
equilibrium conditions apply. To estimate the overflow queue,
(2x 1)I
Qo , x 0.50 (9.26) Newell (1965) defines FQ as the cumulative distribution of the
2(1 x) overflow queue length, FA-D as the cumulative distribution of the
overflow in the cycle, where the indices A and D represent
cumulative arrivals and departures, respectively. He showed that
under equilibrium conditions:
9-7
where,
FQ(x)
FQ(z)dFA D(x z)
20 (9.30)
Sg qc
μ
. (9.33)
The integral in Equation 9.30 can be solved only under the (ISg)1/2
restrictive assumption that the overflow in a cycle is normally
distributed. The resultant Newell formula is as follows:
The function H(μ) has been provided in a graphical form.
Moreover, Newell compared the results given by expressions

tan2
Qo
qc(1 x) /2
d. (Equation 9.29) and (Equation 9.31) with Webster's formula and
20 1exp[Sg(1 x)2/(2cos2)]
(9.31)
added additional correction terms to improve the results for
medium traffic intensity conditions. Newell's final formula is:
A more convenient expression has been proposed by Newell in

c(1 g/c)2
o (1 g/c)I 2 .
Q
the form: d
(9.34)
2(1 q/S) q 2S(1 q/S)
IH(μ)x
Qo
. (9.32)
2(1 x)
Table 9.1
Maximum Relative Discrepancy between the Approximate Expressions
and Ohno's Algorithm (Ohno 1978).
Range of y = 0.0 M 0.5 Range of g/c = 0.4

1.0
Approximate Expressions s = 0.5 v/s s = 1.5 v/s s = 0.5 v/s s = 1.5 v/s
(Equation #, Q0 computed according
to Equation #)
c = 90 s c = 30 s c = 90 s c = 30 s c = 30 s c = 90 s
g = 46 s g = 16 s g = 45.33 s g = 15.33 s q = 0.2 s q = 0.6 s
Modified Miller's expression (9.15, 0.22 2.60 -0.53 0.22 2.24 0.26
9.28)
Modified Newell's expression (9.15, 0.82 2.53 0.25 0.82 2.83 0.25-
9.31)
McNeil's expression (9.15, Miller 1969) 0.49 1.79 0.12 0.49 1.51 0.08
Webster's full expression (9.17) -8.04 -21.47 3.49 -7.75 119.24 1381.10
Newell's expression (9.34, 9.31) -4.16 10.89 -1.45 -4.16 -15.37 -27.27
9-8
More recently, Cronje (1983b) proposed an analytical overflow queue calculated with the method described by
approximation of the function H(μ): Miller 1969), McNeil's formula with overflow queue according
to Miller (Equation 9.28) (modified Miller's expression),
H(μ)
exp[ μ (μ2/2)] (9.35) McNeil's formula with overflow queue according to Newell
(Equation 9.31) (modified Newell's expression), Webster
expression (Equation 9.17) and the original Newell expression
where, (Equation 9.34). Comparative results are depicted in Table 9.1
and Figures 9.3 and 9.4. Newell's expression appear to be more
μ
(1 x) (Sg)1/2. (9.36) accurate than Webster, a conclusion shared by Hutchinson
(1972) in his evaluation of three simplified models (Newell,
Miller, and Webster). Figure 9.3 represents the percentage
He also proposed that the correction (third) component in Equa- relative errors of the approximate delay models measured against
tion 9.34 could be neglected. Ohno’s algorithm (Ohno 1978) for a range of flow ratios. The
modified Miller's and Newell's expressions give almost exact
Earlier evaluations of delay models by Allsop (1972) and average delay values, but they are not superior to the original
Hutchinson (1972) were based on the Webster model form. McNeil formula. Figure 9.4 shows the same type of errors,
Later on, Ohno (1978) carried out a comparison of the existing categorized by the g/c ratio. Further efforts to improve on their
delay formulae for a Poisson arrival process and constant estimates will not give any appreciable reduction in the errors.
departure time during green. He developed a computational The modified Miller expression was recommended by Ohno
procedure to provide the basis for evaluating the selected because of its simpler form compared to McNeil's and Newell's.
models, namely McNeil's expression, Equation 9.15 (with
9.4 Time-Dependent Delay Models

The stochastic equilibrium assumed in steady-state models parabolic, or triangular functions) and calculates the correspond-
requires an infinite time period of stable traffic conditions ing delay. In May and Keller (1967) delay and queues are calcu-
(arrival, service and control processes) to be achieved. At low lated for an unsignalized bottleneck. Their work is nevertheless
flow to capacity ratios equilibrium is reached in a reasonable representative of the deterministic modeling approach and can
period of time, thus the equilibrium models are an acceptable be easily modified for signalized intersections. The general
approximation of the real-world process. When traffic flow ap- assumption in their research is that the random queue
proaches signal capacity, the time to reach statistical equilibrium fluctuations can be neglected in delay calculations. The model
usually exceeds the period over which demand is sustained. defines a cumulative number of arrivals A(t):
Further, in many cases the traffic flow exceeds capacity, a
situation where steady-state models break down. Finally, traffic
flows during the peak hours are seldom stationary, thus violating
A(t)
t
q(-)d-
an important assumption of steady-state models. There has 20 (9.37)
been many attempts at circumventing the limiting assumption
of steady-state conditions. The first and simplest way is to deal
with arrival and departure rates as a function of time in a
deterministic fashion. Another view is to model traffic at signals, and departures D(t) under continuous presence of vehicle
assuming stationary arrival and departure processes but not queue over the period [0,t]:
necessarily under stochastic equilibrium conditions, in order to
D(t)
t
S(-)d-
estimate the average delay and queues over the modeled period 20 (9.38)
of time. The latter approach approximates the time-dependent
arrival profile by some mathematical function (step-function,
9-9
Figure 9.3 Figure 9.4

Percentage Relative Errors for Approximate Delay Relative Errors for Approximate Delay Models
Models by Flow Ratios (Ohno 1978). by Green to Cycle Ratios (Ohno 1978).
9 - 10
The current number of vehicles in the system (queue) is where d1 is the delay experienced at very low traffic
intensity, (uniform delay) T = analysis period over which
flows are sustained.
Q(t)
Q(0)A(t) D(t) (9.39)
steady-state delay models are asymptotic to the y-axis (i.e
generate infinite delays) at unit traffic intensity (x=1). The
and the average delay of vehicles queuing during the time period coordinate transformation method shifts the original
[0,T] is steady-state curve to become asymptotic to the
deterministic oversaturation delay line--i.e.-- the second
1
d
T
Q(t)dt term in Equation 9.41--see Figure 9.5. The horizontal
A(T) 20
(9.40)
distance between the proposed delay curve and its
asymptote is the same as that between the steady-state
curve and the vertical line x=1.
The above models have been applied by May and Keller to a
trapezoidal-shaped arrival profile and constant departure rate. There are two restrictions regarding the application of the
One can readily apply the above models to a signal with known formula: (1) no initial queue exists at the beginning of the
signal states over the analysis period by substituting C(-) for interval [0,T], (2) traffic intensity is constant over the interval
S(-) in Equation 9.38: [0,T]. The time-dependent model behaves reasonably within the
period [0,T] as indicated from simulation experiments. Thus,
C(-) = 0 if signal is red, this technique is very useful in practice. Its principal drawback,
= S(-) if signal is green and Q(-) > 0, in addition to the above stated restrictions (1) and (2) is the lack
= q(-) if signal is green and Q(-) = 0. of a theoretical foundation. Catling overcame the latter diffi-
culties by approximating the actual traffic intensity profile with
Deterministic models of a single term like Equation 9.39 yield a step-function. Using an example of the time-dependent
acceptable accuracy only when x<<1 or x>>1. Otherwise, they version of the Pollaczek-Khintchine equation (Taha 1982), he
tend to underestimate queues and delays since the extra queues illustrated the calculation of average queue and delay for each
causes by random fluctuations in q and C are neglected. time interval starting from an initial, non-zero queue.
According to Catling (1977), the now popular coordinate Kimber and Hollis (1979) presented a computational algorithm
transformation technique was first proposed by Whiting, who did to calculate the expected queue length for a system with random
not publish it. The technique when applied to a steady-state arrivals, general service times and single channel service
curve derived from standard queuing theory, produces a time- (M/G/1). The initial queue can be defined through its
dependent formula for delays. Delay estimates from the new distribution. To speed up computation, the average initial queue
models when flow approaches capacity are far more realistic is used unless it is substantially different from the queue at
than those obtained from the steady-state model. The following equilibrium. In this case, the full computational algorithm
observations led to the development of this technique. should be applied. The non-stationary arrival process is approxi-
mated with a step-function. The total delay in a time period is
At low degree of saturation (x<<1) delay is almost equal calculated by integrating the queue size over time. The
to that occurring when the traffic intensity is uniform coordinate transformation method is described next in some
(constant over time). detail.
At high degrees of saturation (x>>1) delay can be
satisfactory described by the following deterministic model Suppose, at time T=0 there are Q(0) waiting vehicles in queue
with a reasonable degree of accuracy: and that the degree of saturation changes rapidly to x. In a deter-
ministic model the vehicle queue changes as follows:
T
d
d1 (x 1) (9.41)
2
Q(T)
Q(0) (x 1)CT. (9.42)
9 - 11
Figure 9.5
The Coordinate Transformation Method.
The steady-state expected queue length from the modified The following derivation considers the case of exponential
Pollaczek-Khintczine formula is: service times, for which 2 = μ2 , B =1. Let xd be the degree of
saturation in the deterministic model (Equation 9.42), x refers to
Bx 2 the steady-state conditions in model (Equation 9.44), while xT
Q
x (9.43)
1 x refers to the time-dependent model such that Q(x,T)=Q(xT,T).
To meet the postulate of equal distances between the curves and
the appropriate asymptotes, the following is true from Figure
where B is a constant depending on the arrival and departure 9.5:
processes and is expressed by the following equation.
1 x
xd xT (9.45)
B
0.5 1
2
(9.44)
μ2
and hence
x
xT (xd 1)
where 2 and μ are the variance and mean of the service time (9.46)
distribution, respectively.
9 - 12
and from Equation 9.42:

4 [Q(0) xCT][CT (1 B)(Q(0) x CT)]
b
. (9.54)
Q(T) Q(0) CT (1 B)
xd
1, (9.47)
CT
the transformation is equivalent to setting: The equation for the average delay for vehicles arriving during
the period of analysis is also derived starting from the average
Q(T) Q(0)
x
xT . (9.48) delay per arriving vehicle dd over the period [0,T],
CT
1
[Q(0)1] (x 1)CT
2 (9.55)
dd
C
From Figure 9.5, it is evident that the queue length at time T,
Q(T) is the same at x, xT, and xd . By substituting for Q(T) in
Equation 9.44, and rewriting Equation 9.48 gives: and the steady-state delay ds,

xT Q(T) Q(0)
1 Bx
Q(T) ds
(1 ). (9.56)
1Q(T) CT
(9.49) C 1 x
By eliminating the index T in xT and solving the second degree The transformed time dependent equation is
polynomial in Equation 9.49 for Q(T), it can be shown that:
1
d
[(a 2b)1/2 a] (9.57)
1
Q(T)
[(a 2b)1/2 a] (9.50)
2
2
with the corresponding parameters:

where
T 1
a
(1 x) [Q(0) B2] (9.58)
a
(1 x)CT1 Q(0) (9.51) 2 C
and and
b
4 [Q(0) xCT]. 4 T 1 Q(0) 1
(9.52) b
[ (1 x) xT B (1 B)]. (9.59)
C 2 2 C
If the more general steady state Equation 9.43 is used, the result
for Equation 9.51 and 9.52 is: The derivation of the coordinate transformation technique has
been presented. The steady-state formula (Equation 9.43) does
(1 x)(CT)2[1 Q(0)]CT 2(1 B)[Q(0)xCT]
a
(9.53) not appear to adequately reflect traffic signal performance, since
CT(1 B) a) the first term (queue for uniform traffic) needs further
elaboration and b) the constant B must be calibrated for cases
that do not exactly fit the assumptions of the theoretical queuing
and models.
9 - 13
Akçelik (1980) utilized the coordinate transformation technique approximation is relevant to high degrees of saturation x and its
to obtain a time-dependent formula which is intended to be more effect is negligible for most practical purposes.
applicable to signalized intersection performance than Kimber-
Hollis's. In order to facilitate the derivation of a time-dependent Following certain aspects of earlier works by Haight (1963),
function for the average overflow queue Qo, Akçelik used the Cronje (1983a), and Miller (1968a); Olszewski (1990) used
following expression for undersaturated signals as a simple non-homogeneous Markov chain techniques to calculate the
approximation to Miller's second formula for steady-state queue stochastic queue distribution using the arrival distribution P(t,A)
length (Equation 9.28): and capacity distribution P(C). Probabilities of transition from
a queue of i to j vehicles during one cycle are expressed by the
1.5(x xo) following equation:
when x> xo,
Qo
1 x (9.60)
0 otherwise
Pi,j(t)
M P (t,C)P(C)
C
0
i,j (9.64)
where and
Sg C i
xo
0.67
600
(9.61)
Pi,0(t,C)
M P(t,A
k)
k
0
when iC,
(9.65)
0 otherwise
Akçelik's time-dependent function for the average overflow

queue is and
CT 12(x xo) P(t, A

j iC) when j i C,
[(x 1) (x 1)2 ] when x>xo, Pi, j(t,C)
(9.66)
Qo
4 CT (9.62) 0 otherwise .
0 otherwise.
The probabilities of queue states transitions at time t form the

The formula for the average uniform delay during the interval transition matrix P(t). The system state at time t is defined with
[0,T] for vehicles which arrive in that interval is the overflow queue distribution in the form of a row vector PQ(t).
The initial system state variable distribution at time t =0 is
c(1 g/c)2
when x<1 Qo assumed to be known: PQ(0)=[P1(0), P2(0),...Pm(0)], where Pi(0)
d
2(1 q/S) . (9.63) is the probability of queue of length i at time zero. The vector of
C
(c g)/2 when x1 state probabilities in any cycle t can now be found by matrix
multiplication:
PQ(t)
PQ(t 1) P(t) . (9.67)
Generalizations of Equations 9.60 and 9.61 were discussed by
Akçelik (1988) and Akçelik and Rouphail (1994). It should be
noted that the average overflow queue, Q0 is an approximation
of the McNeil (Equation 9.15) and Miller (Equation 9.28) Equation 9.67, when applied sequentially, allows for the calcula-
formulae applied to the time-dependent conditions, and differs tion of queue probability evolution from any initial state.
from Newell's approximations Equation 9.29 and Equation 9.34
of the steady-state conditions. According to Akçelik (1980), this In their recent work, Brilon and Wu (1990) used a similar
computational technique to Olszewski's (1990a) in order to
9 - 14
evaluate existing time-dependent formulae by Catling (1977), which incorporate the impact of the arrival profile shape (e.g. the
Kimber-Hollis (1979), and Akçelik (1980). A comparison of peaking intensity) on delay. In this examination of delay models
the models results is given in Figures 9.6 and 9.7 for a parabolic in the time dependent mode, delay is defined according to the
arrival rate profile in the analysis period To. They found that the path trace method of measurement (Rouphail and Akçelik
Catling method gives the best approximation of the average 1992a). This method keeps track of the departure time of each
delay. The underestimation of delays observed in the Akçelik's vehicle, even if this time occurs beyond the analysis period T.
model is interpreted as a consequence of the authors' using an The path trace method will tend to generate delays that are
average arrival rate over the analyzed time period instead of the typically longer than the queue sampling method, in which
step function, as in the Catling's method. When the peak flow stopped vehicles are sampled every 15-20 seconds for the
rate derived from a step function approximation of the parabolic duration of the analysis period. In oversaturated conditions, the
profile is used in Akçelik's formula, the results were virtually measurement of delay may yield vastly different results as
indistinguishable from Brilon and Wu's (Akçelik and Rouphail vehicles may discharge 15 or 30 minutes beyond the analysis
1993). period. Thus it is important to maintain consistency between
delay measurements and estimation methods. For a detailed
Using numeric results obtained from the Markov Chains discussion of the delay measurement methods and their impact
approach, Brilon and Wu developed analytical approximate (and on oversaturation delay estimation, the reader is referred to
rather complicated) delay formulae of a form similar to Akçelik's Rouphail and Akçelik (1992a).
Figure 9.6
Figure 9.7
Comparison of Delay Models Evaluated by Brilon
Comparison of Delay Models Evaluated by Brilon
and Wu (1990) with Moderate Peaking (z=0.50).
and Wu (1990) with High Peaking (z=0.70).
9 - 15
9.5 Effect of Upstream Signals

The arrival process observed at a point located downstream of q2(t2)dt2 = total number of vehicles passing some
some traffic signal is expected to differ from that observed point downstream of the signal in the
upstream of the same signal. Two principal observations are interval (t, t+dt),
made: a) vehicles pass the signal in "bunches" that are separated q1(t1)dt1 = total number of vehicles passing the
by a time equivalent to the red signal (platooning effect), and b) signal in the interval (t, t+dt), and
the number of vehicles passing the signal during one cycle does f(t2-t1) = probability density of travel time (t2 - t1 )
not exceed some maximum value corresponding to the signal according to Equation 9.68.
throughput (filtering effect).
The discrete version of the diffusion model in Equation 9.69 is
9.5.1 Platooning Effect On Signal

q2(j)
M q (i)g(j i)
i 1 (9.70)
Performance
where i and j are discrete intervals of the arrival histograms.
The effect of vehicle bunching weakens as the platoon moves
downstream, since vehicles in it travel at various speeds, Platoon diffusion effects were observed by Hillier and Rothery
spreading over the downstream road section. This phenomenon, (1967) at several consecutive points located downstream of
known as platoon diffusion or dispersion, was modeled by Pacey signals (Figure 9.8). They analyzed vehicle delays at pretimed
(1956). He derived the travel time distribution f(-) along a road signals using the observed traffic profiles and drew the following
section assuming normally distributed speeds and unrestricted conclusions:
overtaking:
the deterministic delay (first term in approximate delay
D D
( ) 2 formulae) strongly depends on the time lag between the
D - - start of the upstream and downstream green signals
f(-)
exp (9.68)
(offset effect);
-2 ) 2 % 2 )2
the minimum delay, observed at the optimal offset,
where, increases substantially as the distance between signals
increases; and
D= distance from the signal to the point where arrivals
are observed, the signal offset does not appear to influence the
overflow delay component.
-= individual vehicle travel time along distance D,
-= mean travel time, and
The TRANSYT model (Robertson 1969) is a well-known
)= standard deviation of speed.
example of a platoon diffusion model used in the estimation of
deterministic delays in a signalized network. It incorporates the
The travel time distribution is then used to transform a traffic
Robertson's diffusion model, similar to the discrete version of the
flow profile along the road section of distance D:
Pacey's model in Equation 9.70, but derived with the assumption
q2(t2 ) dt2
q1(t1) f(t2 t1) dt1 dt2 of the binomial distribution of vehicle travel time:
2t1 (9.69)
1 1
q2(j)
q (j) (1 ) q (j 1) (9.71)
1a - 1 1a - 2
where,
where - is the average travel time and a is a parameter which
must be calibrated from field observations. The Robertson
model of dispersion gives results which are satisfactory for the
9 - 16
Figure 9.8
Observations of Platoon Diffusion
by Hillier and Rothery (1967).
purpose of signal optimization and traffic performance analysis In the TRANSYT model, a flow histogram of traffic served
in signalized networks. The main advantage of this model over (departure profile) at the stopline of the upstream signal is fir st
the former one is much lower computational demand which is a constructed, then transformed between two signals using model
critical issue in the traffic control optimization for a large size (Equation 9.71) in order to obtain the arrival patterns at the
network. stopline of the downstream signal. Deterministic delays at
the downstream signal are computed using the transformed
arrival and output histograms.
9 - 17
To incorporate the upstream signal effect on vehicle delays, the The remainder of this section briefly summarizes recent work
Highway Capacity Manual (TRB 1985) uses a progression factor pertaining to the filtering effect of upstream signals, and the
(PF) applied to the delay computed assuming an isolated signal. resultant overflow delays and queues that can be anticipated at
A PF is selected out of the several values based on a platoon downstream traffic signals.
ratio fp . The platoon ratio is estimated from field measurement
and by applying the following formula:
9.5.2 Filtering Effect on Signal Performance
PVG
fp
(9.72) The most general steady-state delay models have been derived
g/c
by Darroch (1964a), Newell (1965), and McNeil (1968) for the
binomial and compound Poisson arrival processes. Since these
where, efforts did not deal directly with upstream signals effect, the
question arises whether they are appropriate for estimating
PVG = percentage of vehicles arriving during the overflow delays in such conditions. Van As (1991) addressed
effective green, this problem using the Markov chain technique to model delays
g = effective green time, and arrivals at two closely spaced signals. He concluded that the
c = cycle length. Miller's model (Equation 9.27) improves random delay estima-
tion in comparison to the Webster model (Equation 9.17).
Courage et al. (1988) compared progression factor values Further, he developed an approximate formula to transform the
obtained from Highway Capacity Manual (HCM) with those dispersion index of arrivals, I , at some traffic signal into the
estimated based on the results given by the TRANSYT model. dispersion index of departures, B, from that signal:
They indicated general agreement between the methods,
although the HCM method is less precise (Figure 9.9). To avoid B
I exp( 1.3 F 0.627) (9.73)
field measurements for selecting a progression factor, they
suggested to compute the platoon ratio fp from the ratios of with the factor F given by
bandwidths measured in the time-space diagram. They showed
that the proposed method gives values of the progression factor Qo
F
(9.74)
comparable to the original method. Ia qc
Rouphail (1989) developed a set of analytical models for direct

estimation of the progression factor based on a time-space This model (Equation 9.73) can be used for closely spaced
diagram and traffic flow rates. His method can be considered a signals, if one assumes the same value of the ratio I along a road
simplified version of TRANSYT, where the arrival histogram section between signals.
consists of two uniform rates with in-platoon and out-of-platoon
traffic intensities. In his method, platoon dispersion is also based Tarko et al. (1993) investigated the impact of an upstream signal
on a simplified TRANSYT-like model. The model is thus on random delay using cycle-by-cycle macrosimulation. They
sensitive to both the size and flow rate of platoons. More found that in some cases the ratio I does not properly represent
recently, empirical work by Fambro et al. (1991) and theoretical the non-Poisson arrival process, generally resulting in delay
analyses by Olszewski (1990b) have independently confirmed overestimation (Figure 9.10).
the fact that signal progression does not influence overflow
queues and delays. This finding is also reflected in the most They proposed to replace the dispersion index I with an
recent update of the Signalized Intersections chapter of the adjustment factor f which is a function of the difference between
Highway Capacity Manual (1994). More recently, Akçelik the maximum possible number of arrivals mc observable during
(1995a) applied the HCM progression factor concept to queue one cycle, and signal capacity Sg:
length, queue clearance time, and proportion queued at signals.
9 - 18
Figure 9.9
HCM Progression Adjustment Factor vs Platoon Ratio
Derived from TRANSYT-7F (Courage et al. 1988).
9 - 19
Figure 9.10
Analysis of Random Delay with Respect to the Differential Capacity Factor (f)
and Var/Mean Ratio of Arrivals (I)- Steady State Queuing Conditions (Tarko et al. 1993) .
model which considers a bulk service queuing model and a set

a(mc Sg)
of arrival distribution transformations. They concluded that
f
1 e (9.75) Newell's model estimates provide a close upper bound to the
results from their model. The review of traffic delay models at
where a is a model parameter, a < 0. fixed-timed traffic signals indicate that the state of the art has
shifted over time from a purely theoretical approach grounded
A recent paper by Newell (1990) proposes an interesting in queuing theory, to heuristic models that have deterministic
hypothesis. The author questions the validity of using random and stochastic components in a time-dependent domain. This
delay expressions derived for isolated intersections at internal move was motivated by the need to incorporate additional factors
signals in an arterial system. He goes on to suggest that the sum such as non-stationarity of traffic demand, oversaturation, traffic
of random delays at all intersections in an arterial system with no platooning and filtering effect of upstream signals. It is
turning movements is equivalent to the random delay at the anticipated that further work in that direction will continue,
critical intersection, assuming that it is isolated. Tarko et al. with a view towards using the performance-based models for
(1993) tested the Newell hypothesis using a computational signal design and route planning purposes.
9.6 Theory of Actuated and Adaptive Signals

The material presented in previous sections assumed fixed time traffic-adaptive systems requires new delay formulations that are
signal control, i.e. a fixed signal capacity. The introduction of sensitive to this process. In this section, delay models for
traffic-responsive control, either in the form of actuated or actuated signal control are presented in some detail, which
9 - 20
incorporate controller settings such as minimum and maximum and maximum greens, the phase will be extended for each
greens and unit extensions. A brief discussion of the state of the arriving vehicle, as long as its headway does not exceed the
art in adaptive signal control follows, but no models are value of unit extension. An intersection with two one-way
presented. For additional details on this topic, the reader is streets was studied. It was found that, associated with each
encouraged to consult the references listed at the end of the traffic flow condition, there is an optimal vehicle interval for
chapter. which the average delay per vehicle is minimized. The value of
the optimal vehicle interval decreases and becomes more critical,
as the traffic flow increases. It was also found that by using the
9.6.1 Theoretically-Based Expressions constraints of minimum and maximum greens, the efficiency and
capacity of the signal are decreased. Darroch (1964b) also
As stated by Newell (1989), the theory on vehicle actuated investigated a method to obtain optimal estimates of the unit
signals and related work on queues with alternating priorities is extension which minimizes total vehicle delays.
very large, however, little of it has direct practical value. For
example, "exact" models of queuing theory are too idealized to The behavior of vehicle-actuated signals at the intersection of
be very realistic. In fact the issue of performance modeling of two one-way streets was investigated by Newell (1969). The
vehicle actuated signals is too complex to be described by a arrival process was assumed to be stationary with a flow rate just
comprehensive theory which is simple enough to be useful. slightly below the saturation rate, i.e. any probability
Actuated controllers are normally categorized into: fully- distributions associated with the arrival pattern are time
actuated, semi-actuated, and volume-density control. To date, invariant. It is also assumed that the system is undersaturated
the majority of the theoretical work related to vehicle actuated but that traffic flows are sufficiently heavy, so that the queue
signals is limited to fully and semi-actuated controllers, but not lengths are considerably larger than one car. No turning
to the more sophisticated volume-density controllers with movements were considered. The minimum green is
features such as variable initial and extension intervals. Two disregarded since the study focused on moderate heavy traffic
types of detectors are used in practice: passage and presence. and the maximum green is assumed to be arbitrarily large. No
Passage detectors, also called point or small-area detectors, specific arrival process is assumed, except that it is stationary.
include a small loop and detect motion or passage when a
vehicle crosses the detector zone. Presence detectors, also called Figure 9.11 shows the evolution of the queue length when the
area detectors, have a larger loop and detect presence of vehicles queues are large. Traffic arrives at a rate of q1, on one approach,
in the detector zone. This discussion focuses on traffic actuated and q 2 , on the other. r j , g j , and Yj represent the effective red,
intersection analysis with passage detectors only. green, and yellow times in cycle j. Here the signal timings are
random variables, which may vary from cycle to cycle. For any
Delays at traffic actuated control intersections largely depend on specific cycle j, the total delay of all cars Wij is the area of a
the controller setting parameters, which include the following triangular shaped curve and can be approximated by:
aspects: unit extension, minimum green, and maximum green. q1
Unit extension (also called vehicle interval, vehicle extension, or E{W1j}
(E{rj}Y)2Var(rj)
gap time) is the extension green time for each vehicle as it 2(1 q1/S1)
(9.76)
arrives at the detector. Minimum green: summation of the initial
I1(E{rj}Y) V1
interval and one unit extension. The initial interval is designed
to clear vehicles between the detector and the stop line. S1(1 q1/S1) S1q1
Maximum green: the maximum green times allowed to a specific
phase, beyond which, even if there are continuous calls for the
current phase, green will be switched to the competing approach. q2
E{W2j}
[(E{gj}Y)2
The relationship between delay and controller setting parameters 2(1 q2/S2)
(9.77)
for a simple vehicle actuated type was originally studied by
I2[E(gj)Y] V2
Morris and Pak-Poy (1967). In this type of control, minimum Var(gj)
and maximum greens are preset. Within the range of minimum S2(1 q2/S2) S2q2
9 - 21
Figure 9.11
Queue Development Over Time Under
Fully-Actuated Intersection Control (Newell 1969).
where
Yq2/S2
E{r}
(9.81)
E{W1j}, E{W2j} = the total wait of all cars during 1 q1/S1 q2/S2
cycle j for approach 1 and 2;
S1, S2 = saturation flow rate for approach 1
and 2;
Yq1/S1
E{rj}, E{gj} = expectation of the effective red E{g}
(9.82)
and green times; 1 q1/S1 q2/S2
Var(rj), Var(gj) = variance of the effective red and
green time;
I1 , I2 = variance to mean ratio of arrivals Variances of r and g were also derived, they are not listed here
for approach 1 and 2; and for the sake of brevity. Extensions to the multiple lane case
V1 , V2 = the constant part of the variance of were investigated by Newell and Osuna (1969).
departures for approach 1 and 2.
A delay model with vehicle actuated control was derived by
Since the arrival process is assumed to be stationary, Dunne (1967) by assuming that the arrival process follows a
binomial distribution. The departure rates were assumed to be
E{rj}E{r}, E{gj}E{g} (9.78) constant and the control strategy was to switch the signal when
the queue vanishes. A single intersection with two one-lane one-
Var(rj)Var(r), Var(gj)Var(g) way streets controlled by a two phase signal was considered.
(9.79)
E{Wkj}E{Wk}, k
1,2 (9.80) For each of the intervals (k-, k-+-), k=0,1,2... the probability of
one arrival in approach i = 1, 2 is denoted by qi and the
probability of no arrival by pi=1-qi. The time interval, -, is
The first moments of r and g were also derived based on the taken as the time between vehicle departures. Saturation flow
properties of the Markov process: rate was assumed to be equal for both approaches. Denote
9 - 22
W (2)r as the total delay for approach 2 for a cycle having effective sized bunches separated by inter-bunch headways. All bunched
red time of length r, then it can be shown that: vehicles are assumed to have the same headway of 1 time unit.
All inter-bunch headways follow the exponential distribution.
W (2)r1
W (2) r μ[ 1c 2] (9.83) Bunch size was assumed to have a general probability
distribution with mean, μj, and variance, 2j . The cumulative
probability distribution of a headway less than t seconds, F(t), is
where c is the cycle length, 1, and 2 are increases in delay at
F(t)
1 Qe for t
'(t )
the beginning and at the end of the cycle, respectively, when one (9.88)
vehicle arrives in the extra time unit at the beginning of the 0 for t<
phase and:
where,
μ
0 with probability p2 ,
(9.84) = minimum headway in the arrival stream, =1
1 with probability q2 . time unit;
3 = proportion of free (unbunched) vehicles; and
Equation 9.83 means that if there is no arrivals in the extra time = a delay parameter.
unit at the beginning of the phase, then W (2)r+1=W (2)r, otherwise
W (2)r+1=W (2)r + 1 + c + 2. Formulae for average signal timings (r and g) and average delays
for the cases of j = 0 and j > 0 are derived separately. j = 0
Taking the expectation of Equation 9.83 and substituting for means that the green ends as soon as the queues for the approach
E( 1), E( 2): clear while j > 0 means that after queues clear there will be a
post green time assigned to the approach. By analyzing the
E(Wr1)
E(Wr )q2(r1)/p2
(2) (2)
(9.85) property of Markov process, the following formula are derived
for the case of j = 0.
Solving the above difference equation for the initial condition W

(2) q1L
0=0 gives, E(g1)
(9.89)
1 q1 q2
(2)
E(Wr )
q2(r r)/2p2
2
(9.86)
Finally, taking the expectation of Equation 9.86 with respect to q2L

r gives E(g2)
(9.90)
1 q1 q2
E(W (2))
q2{var[r]E 2[r]E[r]}/(2p2) (9.87)
Therefore, if the mean and variance of (r) are known, delay can
be obtained from the above formula. E(W (i)) for approach 2 is q2L
obtained by interchanging the subscripts. E(r1)
l2 (9.91)
1 q1 g2
Cowan (1978) studied an intersection with two single-lane one-
way approaches controlled by a two-phase signal. The control
policy is that the green is switched to the other approach at the
earliest time, t, such that there is no departures in the interval q1L
[t- i -1, t]. In general i 0. It was assumed that departure E(r2)
l1 (9.92)
1 q1 q2
headways are 1 time unit, thus the arrival headways are at least
1 time unit. The arrival process on approach j is assumed to
follow a bunched exponential distribution. It comprises random-
9 - 23
where, The saturated portion of green period can be estimated from the
following formula:
E(g1), E(g2) = expected effective green for
f q yr
approach 1 and 2; g s
(9.97)
E(r1), E(r2) = expected effective red for approach 1 1 y
and 2;
L = lost time in cycle; where,
l1, l2 = lost time for phase 1 and 2; and
q 1, q 2 = the stationary flow rate for fq = queue length calibration factor to allow for
approach 1 and 2. variations in queue clearance time;
S = saturation flow;
The average delay for approach 1 is: r = red time; and
y = q/S, ratio of arrival to saturation flow rate.
L(1 q2) q1 (1 q2)22(2μ2)(1 q1)3(1 q2)1(1μ1)
2 2 2 2 2
(9.93)
2(1 q1 q2) 2(1 q1 q2)(1 q1 g22q1q2)
The average extension time beyond the saturated portion can be
estimated from:
Akçelik (1994, 1995b) developed an analytical method for
eg
n g h g et (9.98)
estimating average green times and cycle time at a basic vehicle
actuated controller that uses a fixed unit extension setting by
assuming that the arrival headway follows the bunched where,
exponential distribution proposed by Cowan (1978). In his ng = average number of arrivals before a gap
model, the minimum headway in the arrival stream is not change after queue clearance;
equal to one. The delay parameter, ', is taken as Qqt/, where hg = average headway of arrivals before a gap
q t is the total arrival flow rate and =1-qt . In the model, the change after queue clearance; and
free (unbunched) vehicles are defined as those with headways et = terminating time at gap change (in most case
greater than the minimum headway . Further, all bunched it is equal to the unit extension U).
vehicles are assumed to have the same headway . Akçelik
(1994) proposed two different models to estimate the proportion For the case when et = U, Equation 9.98 becomes
of free (unbunched) vehicles Q. The total time, g, allocated to a
movement can be estimated as where gmin is the minimum green eg

1
( 1 )e q(U ) (9.99)
time and g e, the green extension time. This green time, g, is q 3 q
subject to the following constraint
ggmax ge<gemax (9.94)
9.6.2 Approximate Delay Expressions
where gmax and gemax are maximum green and extension time
settings separately. If it is assumed that the unit extension is set Courage and Papapanou (1977) refined Webster's (1958) delay
so that a gap change does not occur during the saturated portion model for pretimed control to estimate delay at vehicle-actuated
of green period, the green time can be estimated by: signals. For clarity, Webster's simplified delay formula is
g
g s eg (9.95) restated below.
c(1 g/c)2 2
d
0.9(d1d2)
0.9[ x ] (9.100)
where gs is the saturated portion of the green period and eg is the 2(1 q/s) 2q(1 x)
extension time assuming that gap change occurs after the queue
clearance period. This green time is subject to the boundaries:
gminggmax (9.96) Courage and Papapanou used two control strategies: (1) the
available green time is distributed in proportion to demand on
9 - 24
the critical approaches; and (2) wasted time is minimized by

terminating each green interval as the queue has been properly mx
d2
900Tx 2[(x 1) (x 1)2 ] (9.105)
serviced. They propose the use of the cycle lengths shown in CT
Table 9.2 for delay estimation under pretimed and actuated
signal control:
where, d, d1, d2, g, and c are as defined earlier and
Table 9.2 DF = delay factor to account for signal coordination and

Cycle Length Used For Delay Estimation for Fixed- controller type;
Time and Actuated Signals Using Webster’s x = q/C, ratio of arrival flow rate to capacity;
Formula (Courage and Papapanou 1977). m = calibration parameter which depends on the arrival
pattern;
Cycle Length Cycle Length C = capacity in veh/hr; and
Type of Signal in 1st Term in 2nd Term T = flow period in hours (T=0.25 in 1994 HCM).
Pretimed Optimum Optimum The delay factor DF=0.85, reduces the queuing delay to account
for the more efficient operation with fully-actuated operation
Actuated Average Maximum when compared to isolated, pretimed control. In an upcoming
The optimal cycle length, c0, is Webster's: revision to the signalized intersection chapter in the HCM, the
delay factor will continue to be applied to the uniform delay term
1.5L5 only.
c0
1
yci (9.101)
As delay estimation requires knowledge of signal timings in the
average cycle, the HCM provides a simplified estimation
where L is total cycle lost time and yci is the volume to saturation method. The average signal cycle length is computed from:
flow ratio of critical movement i. The average cycle length, ca is Lxc
ca
xc
yci
defined as: (9.106)
1.5L
ca
1
yci (9.102)
where xc = critical q/C ratio under fully-actuated control (xc=0.95
and the maximum cycle length, cmax, is the controller maximum in HCM). For the critical lane group i, the effective green:
cycle setting. Note that the optimal cycle length under pretimed yci
control will generally be longer than that under actuated control. gi
ca (9.107)
The model was tested by simulation and satisfactory results
xc
obtained for a wide range of operations.
In the U. S. Highway Capacity Manual (1994), the average This signal timing parameter estimation method has been the
approach delay per vehicle is estimated for fully-actuated subject of criticism in the literature. Lin (1989), among others,
signalized lane groups according to the following: compared the predicted cycle length from Equation 9.106 with
field observations in New York state. In all cases, the observed
d
d1 DF d2 (9.103) cycle lengths were higher than predicted, while the observed xc
ratios were lower.
c(1 g/c)2
d1
(9.104) Lin and Mazdeysa(1983) proposed a general delay model of the
2(1 xg/c) following form consistent with Webster's approximate delay
formula:
9 - 25
where w is the average time required for each queuing vehicle to

g start moving after the green phase starts, L is the average
c(1 K1 )2
c 3600(K2x)2 vehicle length, a is the vehicle acceleration rate from a standing
d
0.9 (9.10) position., and s is the detector setback. If tn0, there is no
!
g 2q(1 K2x)
2(1 K1 K2 x) moving queue exists and thus ei=0; otherwise the green will be
c
extended by the moving queue. Let s be the rate at
which the queuing vehicle move across the detector.
where g, c, q, x are as defined earlier and K1 and K2 are two Considering that additional vehicles may join the queue during
coefficients of sensitivity which reflect different sensitivities of the time interval tn, if tn>0 and s>0, then:
traffic actuated and pretimed delay to both g/c and x ratios. In
stn
this study, K1 and K 2 are calibrated from the simulation model eni
(9.110)
for semi-actuated and fully-actuated control separately. More s q
importantly, the above delay model has to be used in conjunction
with the method for estimating effective green and cycle length. To account for the probability that no moving queues exist
In earlier work, Lin (1982a, 1982b) described a model to upstream of the detector at the end of the initial interval, the
estimate the average green duration for a two phase fully- expected value of eni, ei is expressed as:
actuated signal control. The model formulation is based on the Pj (n/Ti) eni
following assumption: (1) the detector in use is small area ei
(9.111)
passage detector; (2) right-turn-on-red is either prohibited or its
M
n
nmin 1 pj ( n< nmin)
effect can be ignored; and (3) left turns are made only from
exclusive left turn lanes. The arrival pattern for each lane was
assumed to follow a Poisson distribution. Thus, the headway where nmin is the minimum number of vehicles required to form
distribution follows a shifted negative exponential distribution. a moving queue.
Figure 9.12 shows the timing sequence for a two phase fully To estimate Ei, let us suppose that after the initialmini
G and
actuated controller. For phase i, beyond the initial green additional green eni have elapsed, there is a sequence of k
interval, gmini, green extends for Fi based on the control logic and consecutive headways that are shorter than U followed by a
the settings of the control parameters. Fi can be further divided headway longer than U. In this case the green will be extended
into two components: (1) eni — the additional green extended by k times and the resultant green extension time is kJ+U with
n vehicles that form moving queues upstream of the detectors probability [F(h U)]k F(h U), where J is the average length
after the initial interval Gmini; (2) Eni — the additional green of each extension and F(h) is the cumulative headway
extended by n vehicles with headways of no more than one unit distribution function.
extension, U, after Gmini or eni. Note that eni and Eni are random
Ui
variables that vary from cycle to cycle. Lin (1982a, 1982b) tf(t) dt

developed the procedures to estimate ei and Ei , the expected J

(9.112)
F(h<Ui)
value of eni and Eni , as follows. A moving queue upstream of a
detector may exist when Gmini is timed out in case the flow rate
of the critical lane qc is high. If there are n vehicles arriving in and therefore
the critical lane during time Ti, then the time required for the nth
vehicle to reach the detector after Gmini is timed out can be
Ei
k
0(kJU)[F(hU)]kF(h>U)
[ 1 ]e q(U )
1
estimated by the following equation: (9.113)
q q
2(nL si) (9.109)

tn
nw Gmini where is the minimum headway in the traffic stream.
a
9 - 26
Figure 9.12
Example of a Fully-Actuated Two-Phase Timing Sequence (Lin 1982a).
Referring to Figure 9.12, after the values of T1 and T2 are control. The proposed approach uses the delay format in the
obtained, Gi can be estimated as: 1994 HCM (Equations 9.104 and 9.105) with some variations,
namely a) the delay factor, DF, is taken out of the formulation
of delay model and b) the multiplier x2 is omitted from the
Gi

n
0(GminieiEi) P(n/Ti) (9.114)
formulation of the overflow delay term to ensure convergence to
the deterministic oversaturated delay model. Thus, the overflow
delay term is expressed as:
subject to
8kx
Gmini ei Ei (Gmax)i (9.115) d2
900T[(x 1) (x 1)2 ] (9.116)
CT
where P(n/Ti) is the probability of n arrivals in the critical lane

of the ith phase during time interval Ti. Since both T1 and T2 are where the parameter (k) is derived from a numerical calibration
unknown, an iterative procedure was used to determine G1 and of the steady-state for of Equation 9.105 as shown below.
G2.
kx
d 2
(9.117)
C(1 x)
Li et al. (1994) proposed an approach for estimating overflow
delays for a simple intersection with fully-actuated signal
9 - 27
This expression is based on a more general formula by Akçelik models would satisfy the requirement that both controls yield
(1988) and discussed by Akçelik and Rouphail (1994). The identical performance under very light and very heavy traffic
calibration results for the parameter k along with the overall demands. Recent work along these lines has been reported by
statistical model evaluation criteria (standard error and R2) are Akçelik and Chung (1994, 1995).
depicted in Table 9.3. The parameter k which corresponds to
pretimed control, calibrated by Tarko (1993) is also listed. It is
noted that the pretimed steady-state model was also calibrated 9.6.3 Adaptive Signal Control
using the same approach, but with fixed signal settings. The first
and most obvious observation is that the pretimed model Only a very brief discussion of the topic is presented here.
produced the highest k (delay) value compared to the actuated Adaptive signal control systems are generally considered
models. Secondly, the parameter was found to increase with the superior to actuated control because of their true demand
size of the controller's unit extension (U). responsiveness. With recent advances in microprocessor
technology, the gap-based strategies discussed in the previous
Procedures for estimating the average cycle length and green section are becoming increasingly outmoded and demonstrably
intervals for semi-actuated signal operations have been inefficient. In the past decade, control algorithms that rely on
developed by Lin (1982b, 1990) and Akçelik (1993b). Recently, explicit intersection/network delay minimization in a time-
Lin (1992) proposed a model for estimating average cycle length variant environment, have emerged and been successfully tested.
and green intervals under semi-actuated signal control operations While the algorithms have matured both in Europe and the U.S.,
with exclusive pedestrian-actuated phase. Luh (1991) studied evident by the development of the MOVA controller in the U.K.
the probability distribution of and delay estimation for semi- (Vincent et al. 1988), PRODYN in France (Henry et al. 1983),
actuated signal controllers. and OPAC in the U.S. (Gartner et al. 1982-1983), theoretical
work on traffic performance estimation under adaptive control
In summary, delay models for vehicle-actuated controllers are is somewhat limited. An example of such efforts is the work by
derived from assumptions related to the traffic arrival process, Brookes and Bell (1991), who investigated the use of Markov
and are constrained by the actuated controller parameters. The Chains and three heuristic approaches in an attempt to calculate
distribution of vehicle headways directly impact the amount of the expected delays and stops for discrete time adaptive signal
green time allocated to an actuated phase, while controller control. Delays are computed by tracing the queue evolution
parameters bound the green times within specified minimums process over time using a `rolling horizon' approach. The main
and maximums. In contrast to fixed-time models, performance problem lies in the estimation (or prediction) of the initial queue
models for actuated have the additional requirement of in the current interval. While the Markov Chain approach yields
estimating the expected signal phase lengths. Further research theoretically correct answers, it is of limited value in practice due
is needed to incorporate additional aspects of actuated operations to its extensive computational and storage requirements.
such as phase skipping, gap reduction and variable maximum Heuristics that were investigated include the use of the mean
greens. Further, there is a need to develop generalized models queue length, in the last interval as the starting queue in this
that are applicable to both fixed time and actuated control. Such interval; the `two-spike' approach, in which the queue length
Table 9.3
Calibration Results of the Steady-State Overflow Delay Parameter (k) (Li et al. 1994).
Control Pretimed U=2.5 U=3.5 U=4.0 U=5.0

k (m=8k) 0.427 0.084 0.119 0.125 0.231
s.e. NA 0.003 0.002 0.002 0.006
2
R 0.903 0.834 0.909 0.993 0.861
9 - 28
distribution has non-zero probabilities at zero and at an integer Overall, the latter method was recommended because it not only
value closest to the mean; and finally a technique that propagates produces estimates that are sufficiently close to the theoretical
the first and second moment of the queue length distribution estimate, but more importantly it is independent of the traffic
from period to period. arrival distribution.
9.7 Concluding Remarks
In this chapter, a summary and evolution of traffic theory arrival process at the intersection, and of traffic metering
pertaining to the performance of intersections controlled by which may causes a truncation in the departure distribution from
traffic signals has been presented. The focus of the discussion a highly saturated intersection. Next, an overview of delay
was on the development of stochastic delay models. models which are applicable to intersections operating under
vehicle actuated control was presented. They include stochastic
Early models focused on the performance of a single intersection models which characterize the randomness in the arrival and
experiencing random arrivals and deterministic service times departure process-- capacity itself is a random variable which
emulating fixed-time control. The thrust of these models has can vary from cycle to cycle, and fixed-time equivalent models
been to produce point estimates--i.e. expectations of-- delay and which treat actuated control as equivalent pretimed models
queue length that can be used for timing design and quality of operating at the average cycle and average splits.
service evaluation. The model form typically include a
deterministic component to account for the red-time delay and a Finally, there is a short discussion of concepts related to adaptive
stochastic component to account for queue delays. The latter signal control schemes such as the MOVA systems in the United
term is derived from a queue theory approach. Kingdom and OPAC in the U.S. Because these approaches
focus primarily on optimal signal control rather
While theoretically appealing, the steady-state queue theory than performance modeling, they are somewhat beyond the
approach breaks down at high degrees of saturation. The scope of this document.
problem lies in the steady-state assumption of sustained arrival
flows needed to reach stochastic equilibrium (i.e the probability There are many areas in traffic signal performance that deserve
of observing a queue length of size Q is time-independent) . In further attention and require additional research. To begin with,
reality, flows are seldom sustained for long periods of time and the assumption of uncorrelated arrivals found in most models is
therefore, stochastic equilibrium is not achieved in the field at not appropriate to describe platooned flow--where arrivals are
high degrees of saturation. highly correlated. Secondly, the estimation of the initial
overflow queue at a signal is an area that is not well understood
A compromise approach, using the coordinate transformation and documented. There is also a need to develop queuing/delay
method was presented which overcomes some of these models that are constrained by the physical space available for
difficulties. While not theoretically rigorous, it provides a means queuing. Michalopoulos (1988) presented such an application
for traffic performance estimation across all degrees of saturation using a continuous flow model approach. Finally, models that
which is also dependent on the time interval in which arrival describe the interaction between downstream queue lengths and
flows are sustained. upstream departures are needed. Initial efforts in this direction
have been documented by Prosser and Dunne (1994) and
Further extensions of the models were presented to take into Rouphail and Akçelik (1992b).
account the impact of platooning, which obviously alter the
9 - 29
References
Akçelik, R. (1980). Time-Dependent Expressions for Delay, Brookes, D. and M. G. Bell (1991). Expected Delays and Stop
Stop Rate and Queue Length at Traffic Signals. Australian Calculation for Discrete Adaptive Traffic Signal Control.
Road Research Board, Internal Report, AIR 367-1. Proceedings, First International Symposium on Highway
Akçelik, R. (1988). The Highway Capacity Manual Delay Capacity, Karlsruhe, Germany, U. Brannolte, Ed.
Formula for Signalized Intersections. ITE Journal 58(3), Catling, I. (1977). A Time-Dependent Approach To Junction
pp. 23-27. Delays. Traffic Engineering and Control, 18(11),
Akçelik, R. and N. Rouphail (1993). Estimation of Delays at pp. 520-523,526.
Traffic Signals for Variable Demand Conditions. Clayton, A. (1941). Road Traffic Calculations. J. Inst. Civil.
Transportation Research-B, Vol. 27B, No. 2, pp. 109-131. Engrs, 16, No.7, pp.247-284, No. 8, pp. 588-594.
Akçelik, R. (1994). Estimation of Green Times and Cycle Courage, K. G. and P. P. Papapanou (1977). Estimation of
Time for Vehicle-Actuated Signals. Transportation Research Delay at Traffic-Actuated Signals. Transportation Research
Board 1457, pp.63-72. Record, 630, pp. 17-21.
Akçelik, R. and E. Chung (1994). Traffic Performance Courage, K. G., C. E. Wallace, and R. Alqasem (1988).
Models for Unsignalized Intersections and Fixed-Time Modeling the Effect of Traffic Signal Progression on Delay.
Signals. Proceedings, Second International Symposium on Transportation Research Record, 1194, TRB, National
Highway Capacity, Sydney, Australia, ARRB, Volume I, Research Council, Washington, DC,
pp. 21-50. pp. 139-146.
Akçelik, R. and N. Rouphail (1994). Overflow Queues and Cowan, R. (1978). An Improved Model for Signalized
Delays with Random and Platoon Arrivals at Signalized Intersections with Vehicle-Actuated Control. J. Appl. Prob.
Intersections, Journal of Advanced Transportation, Volume 15, pp. 384-396.
28(3), pp. 227-251. Cronje, W. B. (1983a). Derivation of Equations for Queue
Akçelik, R. (1995a). Extension of the Highway Capacity Length, Stops, and Delays for Fixed Time Traffic Signals.
Manual Progression Factor Method for Platooned Arrivals, Transportation Research Record, 905, pp. 93-95.
Research Report ARR No. 276, ARRB Transport Research Cronje, W. B. (1983b). Analysis of Existing Formulas for
Ltd., Vermont South, Australia. Delay, Overflow, and Stops. Transportation Research
Akçelik, R. (1995b). Signal Timing Analysis for Vehicle Record, 905, pp. 89-93.
Actuated Control. Working Paper WD TE95/007, ARRB Darroch, J. N. (1964a). On the Traffic-Light Queue. Ann.
Transport Research Ltd., Vermont South, Australia. Math. Statist., 35, pp. 380-388.
Akçelik, R. and E. Chung (1995). Calibration of Performance Darroch, J. N., G. F. Newell, and R. W. J. Morris (1964b).
Models for Traditional Vehicle-Actuated and Fixed-Time Queues for a Vehicle-Actuated Traffic Light. Operational
Signals. Working Paper WD TO 95/103, ARRB Transport Research, 12, pp. 882-895.
Research Ltd., Vermont South, Australia. Dunne, M. C. (1967). Traffic Delay at a Signalized
Allsop, R. (1972). Delay at a Fixed Time Traffic Signal-I: Intersection with Binomial Arrivals. Transportation
Theoretical Analysis. Transportation Science, 6, Science, 1, pp. 24-31.
pp. 260-285. Fambro, D. B., E. C. P. Chang, and C. J. Messer (1991).
Beckmann, M. J., C. B. McGuire, and C. B. Winsten (1956). Effects of the Quality of Traffic Signal Progression on
Studies in the Economics in Transportation. New Haven, Delay. National Cooperative Highway Research Program
Yale University Press. Report 339, Transportation Research Board, National
Bell, M. G. H. and D. Brookes (1993). Discrete Time- Research Council, Washington, DC.
Adaptive Traffic Signal Control. Transportation Research, Gartner, N. (1982-1983). Demand Responsive Decentralized
Vol. 1C, No. 1, pp. 43-55. Urban Traffic Control. Part I: Single Intersection Policies;
Brilon, W. and N. Wu (1990). Delays At Fixed-time Traffic Part II: Network Extensions, Office of University Research,
Signals Under Time Dependent Traffic Conditions. Traffic U.S.D.O.T.
Engineering and Control, 31(12), pp. 623-63. Gazis, D. C. (1974). Traffic Science. A Wiley-Intersection
Publication, pp. 148-151.
9 - 30
Gerlough, D. L. and M. J. Huber (1975). Traffic Flow Theory. Lin, F. (1992). Modeling Average Cycle Lengths and Green
Transportation Research Board Special Report, 165, Intervals of Semi-actuated Signal Operations with Exclusive
National Research Council, Washington, DC. Pedestrian-actuated Phase. Transportation Research, Vol.
Haight, F. A. (1959). Overflow At A Traffic Flow. 26B, No. 3, pp. 221-240.
Biometrika. Vol. 46, Nos. 3 and 4, pp. 420-424. Lin, F. B. and F. Mazdeyasa (1983). Delay Models of Traffic
Haight, F. A. (1963). Mathematical Theories of Traffic Flow. Actuated Signal Controls. Transportation Research Record,
Academic Press, New York. 905, pp. 33-38.
Henry, J., J. Farges, and J. Tuffal (1983). The PRODYN Luh, J. Z. and Chung Yee Lee (1991). Stop Probability and
Real-Time Traffic Algorithm. 4th IFAC-IFIC--IFORS on Delay Estimations at Low Volumes for Semi-Actuated
Control in Transportation Systems, Baden-Baden. Traffic Signals. Transportation Science, 25(1), pp. 65-82.
Hillier, J. A. and R. Rothery (1967). The Synchronization of May, A. D. and H. M. Keller (1967). A Deterministic
Traffic Signals for Minimum Delays. Transportation Queuing Model. Transportation Research, 1(2),
Science, 1(2), pp. 81-94. pp. 117-128.
Hutchinson, T. P. (1972). Delay at a Fixed Time Traffic May, A. D. (1990). Traffic Flow Fundamentals. Prentice
Signal-II. Numerical Comparisons at Some Theoretical Hall, Englewood Cliffs, New Jersey, pp. 338-375.
Expressions. Transportation Science, 6(3), pp. 286-305. McNeil, D. R. (1968). A Solution to the Fixed-Cycle Traffic
Kimber, R. M. and E. M. Hollis (1978). Peak Period Traffic Light Problem for Compound Poisson Arrivals. J. Appl.
Delay at Road Junctions and Other Bottlenecks. Traffic Prob. 5, pp. 624-635.
Engineering and Control, Vol. 19, No. 10, pp. 442-446. Miller, A. J. (1963). Settings for Fixed-Cycle Traffic Signals.
Kimber, R. and E. Hollis (1979). Traffic Queues and Delays Operational Research Quarterly, Vol. 14, pp. 373-386.
at Road Junctions. TRRL Laboratory Report, 909, U.K. Miller, A. J. (1968a). Australian Road Capacity Guide -
Li, J., N. Rouphail, and R. Akçelik (1994). Overflow Delay Provisional Introduction and Signalized Intersections.
Estimation for Intersections with Fully-Actuated Signal Australian Road Research Board Bulletin No.4, (Superseded
Control. Presented at the 73rd Annual Meeting of TRB, by ARRB report ARR No. 123, 1981).
Washington, DC. Miller, A. J. (1968b). The Capacity of Signalized
Lighthill, M. H. and G. B. Whitham (1957). On Kinematic Intersections in Australia. Australian Road Research Board,
Waves: II. A Theory of Traffic Flow on Long Crowded ARRB Bulletin No.3.
Roads. Proceedings of the Royal Society, London Series Morris, R. W. T. and P. G. Pak-Boy (1967). Intersection
A229, No. 1178, pp. 317-345. Control by Vehicle Actuated Signals. Traffic Engineering
Little, J. D. C. (1961). Approximate Expected Delays for and Control, No.10, pp. 288-293.
Several Maneuvers by Driver in a Poisson Traffic. Newell, G. F. (1960). Queues for a Fixed-Cycle Traffic Light.
Operations Research, 9, pp. 39-52. The Annuals of Mathematical Statistics, Vol.31, No.3,
Lin, F. B. (1982a). Estimation of Average Phase Duration of pp. 589-597.
Full-Actuated Signals. Transportation Research Record, Newell, G. F. (1965). Approximation Methods for Queues with
881, pp. 65-72. Application to the Fixed-Cycle Traffic Light. SIAM
Lin, F. B. (1982b). Predictive Models of Traffic-Actuated Review, Vol.7.
Cycle Splits. Transportation Research - B, Vol. 16B, Newell, G. F. (1969). Properties of Vehicle Actuated Signals:
No. 5, pp. 361-372. I. One-Way Streets. Transportation Science, 3, pp. 30-52.
Lin, F. (1989). Application of 1985 Highway Capacity Newell, G. F. (1989). Theory of Highway Traffic Signals.
Manual for Estimating Delays at Signalized Intersection. UCB-ITS-CN-89-1, Institute of Transportation Studies,
Transportation Research Record, 1225, TRB, National University of California.
Research Council, Washington DC, pp. 18-23. Newell, G. F. (1990). Stochastic Delays on Signalized
Lin, F. (1990). Estimating Average Cycle Lengths and Green Arterial Highways. Transportation and Traffic Theory,
Intervals of Semiactuated Signal Operations for Level-of- Elsevier Science Publishing Co.,Inc., M. Koshi, Ed.,
Service Analysis. Transportation Research Record, 1287, pp. 589-598.
TRB, National Research Council, Washington DC,
pp. 119-128.
9 - 31
Newell, G. F. and E. E. Osuna (1969). Properties of Vehicle- Rouphail, N. and R. Akçelik (1992a). Oversaturation Delay
Actuated Signals: II. Two-Way Streets. Transportation Estimates with Consideration of Peaking. Transportation
Science, 3, pp. 99-125. Research Record, 1365, pp. 71-81.
Ohno, K. (1978). Computational Algorithm for a Fixed Cycle Rouphail, N. and R. Akçelik(1992b). A Preliminary Model of
Traffic Signal and New Approximate Expressions for Queue Interaction at Signalized Paired Intersections,
Average Delay. Transportation Science, 12(1), pp. 29-47. Proceedings, 16th ARRB Conference 16(5), Perth, Australia,
Olszewski, P. S. (1988). Efficiency of Arterial Signal pp. 325-345.
Coordination. Proceedings 14th ARRB Conference, 14(2), Stephanopoulos, G., G. Michalopoulos, and G.
pp. 249-257. Stephanopoulos (1979). Modelling and Analysis of Traffic
Olszewski, P. (1990a). Modelling of Queue Probability Queue Dynamics at Signalized Intersections.
Distribution at Traffic Signals. Transportation and Traffic Transportation Research, Vol. 13A, pp. 295-307.
Flow Theory, Elsevier Science Publishing Co., Inc., M. Tarko, A., N. Rouphail, and R. Akçelik (1993b). Overflow
Koshi, Ed., pp. 569-588. Delay at a Signalized Intersection Approach Influenced by
Olszewski, P. (1990b). Traffic Signal Delay Model for Non- an Upstream Signal: An Analytical Investigation.
Uniform Arrivals. Transportation Research Record, 1287, Transportation Research Record, No. 1398, pp. 82-89.
pp. 42-53. Transportation Research Board (1985). Highway Capacity
Pacey, G. M. (1956). The Progress of a Bunch of Vehicles Manual, Special Report 209, National Research Council,
Released from a Traffic Signal. Research Note No. Washington, DC.
Rn/2665/GMP. Road Research Laboratory, London Transportation Research Board (1993). Highway Capacity
(mimeo). Manual, Chapter 9 (Signalized Intersections), National
Potts, R. B. (1967). Traffic Delay at a Signalized Intersection Research Council, Washington, DC.
with Binomial Arrivals. pp. 126-128. Van As, S. C (1991). Overflow Delay in Signalized
Robertson, D. I. (1969). TRANSYT: A Traffic Network Study Networks. Transportation Research - A, Vol. 25A,
Tool. Road Research Laboratory Report LR 253, No. 1, pp. 1-7.
Crowthorne. Vincent, R. A. and J. R. Pierce (1988). MOVA: Traffic
Rorbech, J. (1968). Determining The Length of The Approach Responsive, Self Optimizing Signal Control for Isolated
Lanes Required at Signal-Controlled Intersections on Intersections. TRRL Lab Research Report, 170.
Through Highways. Transportation Research Record, 1225, Webster, F. V. (1958). Traffic Signal Settings. Road
TRB, National Research Council, Washington, DC, pp. Research Laboratory Technical Paper No. 39, HMSO,
18-23. London.
Rouphail, N. (1989). Progression Adjustment Factors at Wirasinghe, S. C. (1978). Determination of Traffic Delays
Signalized Intersections. Transportation Research Record, from Shock-Wave Analysis. Transportation Research, Vol.
1225, pp. 8-17. 12, pp. 343-348.
9 - 32
TRAFFIC SIMULATION
BY EDWARD LIEBERMAN18
AJAY K. RATHI
18
President, KLD Associates, Inc. 300 Broadway, Huntington Station, NY 11746
af = acceleration response of follower vehicle to some stimulus
v = instantaneous speed of lead vehicle
vf = instantaneous speed of follower vehicle
d = projected maximum deceleration rate of lead vehicle
df = projected maximum deceleration rate of follower vehicle
Rf = reaction time lag of driver in follower vehicle
Ii = ith replicate of specified seed, S0
RI = ith random number
h = headway separating vehicles (sec)
H = mean headway (sec)
hmin = minimum headway (sec)
R = random number
XI = ith observation (sample) of an MOE
μx = mean of sample
2 = variance
= estimate of variance
tn-1,1- /2 = upper 1-/2 critical point of the t distribution with n-1 degrees of freedom
xj(m) = mean of m observation of jth batch
x = grand sample mean across batches
Ni = number of replications of the ith strategy
var(X ) = variance of statistic, X
10.
TRAFFIC SIMULATION
10.1 Introduction
Simulation modeling is an increasingly popular and effective tool for this purpose as an integral element of the ATMS research
for analyzing a wide variety of dynamical problems which are and development activity.
not amenable to study by other means. These problems are
usually associated with complex processes which can not readily 2. Testing new designs
be described in analytical terms. Usually, these processes are Transportation facilities are costly investments. Simulation
characterized by the interaction of many system components or can be applied to quantify traffic performance responding to
entities. Often, the behavior of each entity and the interaction of different geometric designs before the commitment of
a limited number of entities, may be well understood and can be resources to construction.
reliably represented logically and mathematically with
acceptable confidence. However, the complex, simultaneous 3. As an element of the design process
interactions of many system components cannot, in general, be The classical iterative design paradigm of conceptual design
adequately described in mathematical or logical forms. followed by the recursive process of evaluation and design
refinement, can benefit from the use of simulation. Here, the
Simulation models are designed to "mimic" the behavior of such simulation model can be used for evaluation; the detailed
systems. Properly designed models integrate these separate statistics provided can form the basis for identifying design
entity behaviors and interactions to produce a detailed, flaws and limitations. These statistics augmented with
quantitative description of system performance. Specifically, animation displays can provide invaluable insights guiding
simulation models are mathematical/logical representations (or the engineer to improve the design and continue the process.
abstractions) of real-world systems, which take the form of
software executed on a digital computer in an experimental 4. Embed in other tools
fashion. In addition to its use as a stand-alone tool, simulation sub-
models can be integrated within software tools designed to
The user of traffic simulation software specifies a “scenario” perform other functions. Examples include: (1) the flow
(e.g., highway network configuration, traffic demand) as model model within the TRANSYT-7F signal optimization; (2) the
inputs. The simulation model results describe system DYNASMART simulation model within a dynamic traffic
operations in two formats: (1) statistical and (2) graphical. The assignment; (3) the simulation component of the
numerical results provide the analyst with detailed quantitative INTEGRATION assignment/control model; (4) the
descriptions of what is likely to happen. The graphical and CORSIM model within the Traffic Research Laboratory
animated representations of the system functions can provide (TreL) developed for FHWA; and (5) the simulation module
insights so that the trained observer can gain an understanding of the EVIPAS actuated signal optimization program.
of why the system is behaving this way. However, it is the
responsibility of the analyst to properly interpret the wealth of 5. Training personnel
information provided by the model to gain an understanding of Simulation can be used in the context of a real-time
cause-and-effect relationships. laboratory to train operators of Traffic Management Centers.
Here, the simulation model, which is integrated with a real-
Traffic simulation models can satisfy a wide range of time traffic control computer, acts as a surrogate for the real-
requirements: world surveillance, communication and traffic environments.
1. Evaluation of alternative treatments 6. Safety Analysis

With simulation, the engineer can control the experimental Simulation models to “recreate” accident scenarios have
environment and the range of conditions to be explored. proven to be indispensable tools in the search to build safer
Historically, traffic simulation models were used initially to vehicles and roadways. An example is the CRASH program
evaluate signal control strategies, and are currently applied used extensively by NHTSA.

%5$)),& $,08/$7,21
This compilation of applications indicates the variety and scope potential. Unlike the other chapters of this monograph, we will
of traffic simulation models and is by no means exhaustive. not focus exclusively on theoretical developments -- although
Simulation models can also be supportive of analytical models fundamental simulation building blocks will be discussed.
such as PASSER, and of computational procedures such as the Instead, we will describe the properties, types and classes of
HCS. While these and many other computerized tools do not traffic simulation models, their strengths and pitfalls, user
include simulation sub-models, users of these tools can enhance caveats, and model-building fundamentals. We will emphasize
their value by applying simulation to evaluate their performance. how the user can derive the greatest benefits from simulation
through proper interpretation of the results, with emphasis on the
This chapter is intended for transportation professionals, need to adequately calibrate the model and to apply rigorous
researchers, students and technical personnel who either statistical analysis of the results.
currently use simulation models or who wish to explore their
10.2 When Should the Use of Simulation Models be Considered?

Since simulation models describe a dynamical process in in order to explain why the resulting statistics were
statistical and pictorial formats., they can be used to analyze a produced.
wide range of applications wherever...
Congested conditions persist over a significant time.
Mathematical treatment of a problem is infeasible or
inadequate due to its temporal or spatial scale, and/or It must be emphasized that traffic simulation, by itself, cannot be
the complexity of the traffic flow process. used in place of optimization models, capacity estimation
procedures, demand modeling activities and design practices.
The assumptions underlying a mathematical Simulation can be used to support such undertakings, either as
formulation (e.g., a linear program) or an heuristic embedded submodels or as an auxiliary tool to evaluate and
procedure (e.g., those in the Highway Capacity extend the results provided by other procedures. Some
Manual) cast some doubt on the accuracy or representative statistics (called Measures of Effectiveness,
applicability of the results. MOE) provided by traffic simulation models are listed in Table
10.1.
The mathematical formulation represents the dynamic
traffic/control environment as a simpler quasi steady- Such statistics can be presented for each specified highway
state system. section (network link) and for each specified time period, to
yield a level of detail that is both spatially and temporally
There is a need to view vehicle animation displays to disaggregated. Aggregations of these data, by subnetwork and
gain an understanding of how the system is behaving network-wide, and over specified time periods, may also be
provided.
10.3 Examples of Traffic Simulation Applications

Given the great diversity of applications that are suitable for the 10.3.1 Evaluation of Signal Control
use of traffic simulation models, the following limited number of
Strategies
examples provides only a limited representation of past
experience.
This study (Gartner and Hou, 1992) evaluated and compared the
performance of two arterial traffic control strategies,

%5$)),& $,08/$7,21
Table 10.1
Simulation Output Statistics: Measures of Effectiveness
Measure for Each Link and for Entire Network
Travel: Vehicle-Miles Bus Travel Time
Travel Time: Vehicle-minutes Bus Moving Time
Moving Time: Vehicle-minutes Bus Delay
Delay Time: Vehicle-minutes Bus Efficiency: Moving Time

persons-minutes Total Travel Time
Efficiency: Moving Time Bus Speed

Total Travel Time
Mean Travel Time per Vehicle-Mile Bus Stops
Mean Delay per Vehicle-Mile Time bus station capacity exceeded
Mean Travel Time per vehicle Time bus station is empty
Mean Time in Queue Fuel consumed
Mean Stopped Time CO Emissions
Mean Speed HC Emissions
Vehicle Stops NOX Emissions
Link Volumes Occupancy
Mean Link Storage Area Consumed
Number of Signal Phase Failures
Average Queue Length
Maximum Queue Length
Lane Changes
Bus Trips
Bus Person Trips

%5$)),& $,08/$7,21
MULTIBAND and MAXBAND, employing the TRAF- system optimal (SO) equilibrium calculations for a specified
NETSIM simulation model. The paper describes the statistical network, over a range of traffic loading conditions from
analysis procedures, the number of simulation replications unsaturated to oversaturated. This is an example of traffic
executed and the resulting 95 percent confidence intervals, and simulation used as a component of a larger model to perform a
the results of the analysis. complex analysis of an ITS initiative.
Figure 10.1 is taken from this paper and illustrates how Figure 10.2 which is taken from the cited paper illustrates how
simulation can provide objective, accurate data sufficient to simulation can produce internally consistent results for large
distinguish between the performance of alternative analytical scale projects, of sufficient resolution to distinguish between two
models, within the framework of a controlled experiment. comparable equilibrium assignment approaches.
10.3.2 Analysis of Equilibrium Dynamic 10.3.3 Analysis of Corridor Design

Assignments (Mahmassani and Alternatives (Korve Engineers
Peeta, 1993) 1996)
This analysis employed the WATSim simulation model to

This large-scale study used the DYNASMART simulation-
evaluate alternative scenarios for increasing capacity and
assignment model to perform both user equilibrium (UE) and
improving traffic flow on a freeway connection, SR242, in
Figure 10.1
Average Delay Comparison, Canal Street,
MULTIBAND & MAXBAND (KLD-242).

%5$)),& $,08/$7,21
Figure 10.2
Comparison of average trip times (minutes) of SO & UE...(KLD-243).
California and ensuring a balanced design relative to freeway could mitigate the extent and duration of congestion within the
SR4 on the north and I680 to the south. Design alternatives area, thereby improving performance and productivity. Here, a
considered for three future periods (years 2000, 2010, 2020) new control concept was tested on a real-world test-bed: a
included geometric changes, widening, HOV lanes and ramp section of Manhattan. This example illustrates the value in
metering. Given the scale of this 20-mile corridor and the strong testing new “high risk” ideas with simulation without exposing
interactions of projected design changes for the three highways, the public to possible adverse consequences, and prior to
the use of simulation provided a statistical basis for quantifying expending resources to implement these concepts.
the operational performance of the corridor sections for each
alternative. These examples certainly do not represent the full range of traffic
simulation applications. Yet, they demonstrate the application
This example illustrates the use of simulation as an element of of traffic simulation in the areas of (1) traffic control; (2)
the design process with the capability of analyzing candidate transportation planning; (3) design; and (4) research.
designs of large-scale highway systems in a manner that lies
beyond the capabilities of a straight-forward HCM analysis.
10.4 Classification of Simulation
10.3.4 Testing New Concepts Models
The TRAF-NETSIM simulation model was used by Rathi and Almost all traffic simulation models describe dynamical systems
Lieberman (1989) to determine whether the application of -- time is always the basic independent variable. Continuous
metering control along the periphery of a congested urban area simulation models describe how the elements of a system change

%5$)),& $,08/$7,21
state continuously over time in response to continuous stimuli. processes. The duration of the lane-change maneuver can also
Discrete simulation models represent real-world systems (that be calculated.
are either continuous or discrete) by asserting that their states
change abruptly at points in time. There are generally two types A mesoscopic model generally represents most entities at a high
of discrete models: level of detail but describes their activities and interactions at a
much lower level of detail than would a microscopic model. For
Discrete time example, the lane-change maneuver could be represented for
Discrete event individual vehicles as an instantaneous event with the decision
based, say, on relative lane densities, rather than detailed vehicle
The first, segments time into a succession of known time interactions.
intervals. Within each such interval, the simulation model
computes the activities which change the states of selected A macroscopic model describes entities and their activities and
system elements. This approach is analogous to representing an interactions at a low level of detail. For example, the traffic
initial-value differential equation in the form of a finite- stream may be represented in some aggregate manner such as a
difference expression with the independent variable, t. statistical histogram or by scalar values of flow rate, density and
speed. Lane change maneuvers would probably not be
Some systems are characterized by entities that are "idle" much represented at all; the model may assert that the traffic stream is
of the time. For example, the state of a traffic signal indication properly allocated to lanes or employ an approximation to this
(say, green) remains constant for many seconds until its state end.
changes instantaneously to yellow. This abrupt change in state
is called an event. Since it is possible to accurately describe the High-fidelity microscopic models, and the resulting software, are
operation of the signal by recording its changes in state as a costly to develop, execute and to maintain, relative to the lower
succession of [known or computed] timed events, considerable fidelity models. While these detailed models possess the
savings in computer time can be realized by only executing these potential to be more accurate than their less detailed
events rather than computing the state of the signal second-by- counterparts, this potential may not always be realized due to the
second. For systems of limited size or those representing entities complexity of their logic and the larger number of parameters
whose states change infrequently, discrete event simulations are that need to be calibrated.
more appropriate than are discrete time simulation models, and
are far more economical in execution time. However, for Lower-fidelity models are easier and less costly to develop,
systems where most entities experience a continuous change in execute and to maintain. They carry a risk that their
state (e.g., a traffic environment) and where the model objectives representation of the real-world system may be less accurate,
require very detailed descriptions, the discrete time model is less valid or perhaps, inadequate. Use of lower-fidelity
likely to be the better choice. simulations is appropriate if:
Simulation models may also be classified according to the level The results are not sensitive to microscopic details.
of detail with which they represent the system to be studied: The scale of the application cannot accommodate the
higher execution time of the microscopic model.
Microscopic (high fidelity) The available model development time and resources
Mesoscopic (mixed fidelity) are limited.
Macroscopic (low fidelity)
Within each level of detail, the developer has wide latitude in
A microscopic model describes both the system entities and their designing the simulation model. The developer must identify the
interactions at a high level of detail. For example, a lane-change sensitivity of the model's performance to the underlying features
maneuver at this level could invoke the car-following law for the of the real-world process. For example, if the model is to be
subject vehicle with respect to its current leader, then with used to analyze weaving sections, then a detailed treatment of
respect to its putative leader and its putative follower in the lane-change interactions would be required, implying the need
target lane, as well as representing other detailed driver decision for a micro- or mesoscopic model. On the other hand, if the

75$)),& 6,08/$7,21
model is designed for freeways characterized by limited merging Traffic simulation models have taken many forms depending on
and no weaving, describing the lane-change interactions in great their anticipated uses. Table 10.2 lists the TRAF family of
detail is of lesser importance, and a macroscopic model may be models developed for the Federal Highway Administration
the suitable choice. (FHWA), along with other prominent models, and indicates their
respective classifications. This listing is necessarily limited.
Another classification addresses the processes represented by the Some traffic simulation models consider a single facility
model: (1) Deterministic; and (2) Stochastic. Deterministic (NETSIM, NETFLO 1 and 2: surface streets; FRESIM,
models have no random variables; all entity interactions are FREFLO: freeways; ROADSIM: two-lane rural roads;
defined by exact relationships (mathematical, statistical or (CORSIM) integrates two other simulation models, FRESIM
logical). Stochastic models have processes which include and NETSIM; INTEGRATION, DYNASMART, TRANSIMS
probability functions. For example, a car-following model can are components of larger systems which include demand models
be formulated either as a deterministic or stochastic relationship and control policies; while CARSIM is a stand-alone simulation
by defining the driver's reaction time as a constant value or as a of a car-following model. It is seen that traffic simulation models
random variable, respectively. take many forms, each of which satisfies a specific area of
application.
Table 10.2
Representative Traffic Simulation Models
Name Discrete Discrete Micro Mesoscopic Macro Deterministic Stochastic

Time Event
NETSIM X X X
NETFLO 1 X X X
NETFLO 2 X X X
FREFLO X X X
ROADSIM X X X
FRESIM X X X
CORSIM X X X
INTEGRATION X X X
DYNASMART X X X
CARSIM X X X
TRANSIMS X X X

75$)),& 6,08/$7,21
10.5 Building Traffic Simulation Models

The development of a traffic simulation model involves the 4) Calibrate the Model
following activities: - Collect/acquire data to calibrate the model.
- Introduce this data into the model.
1) Define the Problem and the Model Objectives
- State the purpose for which the model is being 5) Model Verification
developed. - Establish that the software executes in accord with
- Define the information that the model must the design specification.
provide. - Perform verification at the model component level.
2) Define the System to be Studied 6) Model Validation

- Disaggregate the system to identify its major - Collect, reduce, organize data for purposes of
components. validation.
- Define the major interactions of these - Establish that the model describes the real system
components. at an acceptable level of accuracy over its entire
- Identify the information needed as inputs. domain of operation; apply rigorous statistical
- Bound the domain of the system to be modeled. testing methods.
3) Develop the Model 7) Documentation

- Identify the level of complexity needed to satisfy the - Executive Summary
stated objectives. - Users Manual
- Classify the model and define its inputs and - Model documentation: algorithms and software
outputs.
- Define the flow of data within the model. The development of a traffic simulation model is not a “single-
- Define the functions and processes of the model pass” process. At each step in the above sequence, the analyst
components. must review the activities completed earlier to determine
- Determine the calibration requirements and form: whether a revision/extension is required before proceeding
scalars, statistical distributions, further. For example, in step 5 the analyst may verify that the
parametric dependencies. software is replicating a model component properly as designed,
- Develop abstractions (i.e., mathematical-logical- but that its performance is at a variance with theoretical
statistical algorithms) of each expectations or with empirical observations. The analyst must
major system component, their activities and then determine whether the calibration is adequate and accurate
interactions. (step 4); whether the model’s logical/mathematical design is
- Create a logical structure for integrating these correct and complete (step 3); whether all interactions with other
model components to support the flow of data model components are properly accounted for and that the
among them. specified inputs are adequate in number and accuracy (step 2).
- Select the software development paradigm, This continual feedback is essential; clearly, it would be
programming language(s), user interface, pointless to proceed with validation (step 6) if it is known that
presentation formats of model results. the verification activity is incomplete.
- Design the software: simulation, structured or
object-oriented programming language; database, Step 3 may be viewed as the most creative activity of the
relational/object oriented. development process. The simulation logic must represent all
- Document the logic and all computational relevant interactions by suitably exploring the universe of
procedures. possibilities and representing the likely outcome. These
- Develop the software code and debug. combinations of interactions are called processes which

75$)),& 6,08/$7,21
represent specified functions and utilize component models. A The most popular approach for random number generation is the
small sample of these is presented below. “linear congruential method” which employs a recursive
equation to produce a sequence of random integers S as:
10.5.1 Car-Following Si
(aSi 1 b) mod c.
One fundamental interaction present in all microscopic traffic
simulation models is that between a leader-follower pair of
vehicles traveling in the same lane. This interaction takes the where the integers chosen are defined as,
form of a stimulus-response mechanism: c is the modulus, such that c > 0,
a is the multiplier such that 0 < a < c,
b is the increment such that 0 < b < m, and
af
F (vl,vf,s,dl,df,Rf,Pi) (10.1) S o is the starting value or the Seed of the random number
generator, such that 0 < So < c.
where af , the acceleration (response) of the follower vehicle, is
The ith random number denoted by Ri is then generated as
dependent on a number of (stimulus) factors including:
vl, vf = Speeds of leader, follower vehicles, Si

respectively. Ri
.
c
s = Separation distance.
d l, d f = Projected deceleration rates of the
leader, follower vehicles, respectively.
These random number generations are typically used to generate
Rf = Reaction time of the driver in the
random numbers between 0 and 1. That is, a Uniform (0,1)
following vehicle.
random number is generated. Random variates are usually
Pi = Other parameters specific to the car-
referred to as the sample generated from a distribution other than
following model.
the Uniform (0,1). More often than not, these random variates
F() = A mathematical and logical formulation
are generated from the Uniform (0,1) random number. A
relating the response parameter to the
simulation usually needs random variates during its execution.
stimulus factors.
Based on the distribution specified, there are various analytical
methods employed by the simulation models to generate the
This behavioral model can be referenced (i.e., executed) to
random variates. The reader is referred to Law and Kelton
support other behavioral models such as lane-changing,
(1991) or Roberts (1983) for a detailed treatment on this topic.
merging, etc.
As an example, random variates in traffic simulation are used to
generate a stream of vehicles.
10.5.2 Random Number Generation

10.5.3 Vehicle Generation
All stochastic models must have the ability to generate random
numbers. Generation of random numbers has historically been
At the outset of a simulation run, the system is “empty”.
an area of interest for researchers and practitioners. Before
Vehicles are generated at origin points, usually at the periphery
computers were invented, people relied on mechanical devices
of the analysis network, according to some headway distribution
and their observations to generate random numbers. While
based on specified volumes. For example, the shifted negative
numerous methods in terms of computer programs have been
exponential distribution will yield the following expression:
devised to generate random numbers, these numbers only
“appear” to be random. This is the reason why some call them
pseudo-random numbers. h
(H hmin) [ ln(1 R)] H hmin

%5$)),& $,08/$7,21
where Driver: Aggressiveness; responsiveness to stimuli;

h = Headway (sec) separating vehicle emissions destination (route); other behavioral and decision
H = Mean headway = 3600/V, where V is the processes.
specified volume, vph
hmin = Specified minimum headway
(e.g., 1.2 sec/veh) Each attribute must be represented by the analyst, some by
R = Random number in the range (0 to 1.0), scalars (e.g., vehicle length); some by a functional relationship
obtained from a pseudo-random number (e.g., maximum vehicle acceleration as a function of its current
generator. speed); some by a probability distribution (e.g., driver gap
acceptance behavior). All must be calibrated.
Suppose the specified volume, V (vph), applies for a 15-minute
period. If the user elects to guarantee that V is explicitly
satisfied by the simulation model, it is necessary to generate N The driver-vehicle combination forms a model component, or
values of h using the above formula repeatedly, generating a new entity. This component is defined in terms of its own elemental
random number each time. Here, N = V/4 is the expected attributes and its functionality is defined in terms of the
number of vehicles to be emitted in 15 minutes. The model interactions between these elements. For example, the driver's
could then calculate the factor, K: decision to accelerate at a certain rate may be constrained by the
vehicle's operational limitations. In addition, this system
component interacts with other model entities representing the
15 x60
K
(10.2) environment under study, including:
M
N
hi
i
1
roadway geometrics
intersection configurations
The model would then multiply each of the N values of hi, by K, nearby driver-vehicle entities
so that the resulting sum of (the revised) hi will be exactly 15 control devices
minutes, ensuring that the user’s specification of demand volume lane channelization
are satisfied. However, if K C 1.0, then the resulting distribution conflicting vehicle movements
of generated vehicles is altered and one element of stochasticity
(i.e., the actual number of generated vehicles) is removed. The As an example, the driver-vehicle entity’s interaction with a
model developer must either include this treatment (i.e., control device depends on the type and current state of the device
eq.10.2), exclude it; or offer it as a user option with appropriate (e.g., a signal with a red indication), the vehicle’s speed, its
documentation. distance from stop-bar, the driver’s aggressiveness, etc. It is the
developer’s responsibility to design the model components and
their interactions in a manner that satisfies the model objectives
10.5.4 A Representative Model and is consistent with its fidelity.
Component
Consider two elements of every traffic environment: (1) a

10.5.5 Programming Considerations
vehicle and (2) its driver. Each element can be defined in terms
of its relevant attributes: Programming languages, in the context of this chapter, may be
classified as simulation and general-purpose languages.
Vehicle: Length; width; acceleration limits; deceleration Simulation languages such as SIMSCRIPT and GPSS/H greatly
limit; maximum speed; type (auto, bus, truck, ...); ease the task of developing simulation software by incorporating
maximum turn radius, etc. many features which compile statistics and perform queuing and
other functions common to discrete simulation modeling.

%5$)),& $,08/$7,21
General-purpose languages may be classified as procedural While object-oriented languages can produce more reliable
(e.g., FORTRAN, PASCAL, C, BASIC), or object-oriented software, they require a higher level of programming skill than
(e.g., SMALLTALK, C++, JAVA). Object-oriented languages do procedural languages. The developer should select a
are gaining prominence since they support the concept of language which is hardware independent, is supported by the
reusable software defining objects which communicate with one major operating systems and is expected to have a long life,
another to solve a programming task. Unlike procedural given the rapid changes in the world of software engineering.
languages, where the functions are separated from, and operate Other factors which can influence the language selection process
upon, the data base, objects encapsulate both data describing its include: (1) the expected life of the simulation model; (2) the
state, as well as operations (or “methods”) which can change its skills of the user community; (3) available budget (time and
state and interact with other objects. resources) to develop and maintain the software; and (4) a
realistic assessment of available software development skills.
10.6 An Illustration of Simulation Model Building

Given the confines of this chapter, we will illustrate the model It must be emphasized that model development is an iterative
development process by presenting the highlights of a sample process. For example, the need for the indicated input
problem, but avoiding exhaustive detail. distributions may not have been recognized during this
definition phase, but may have emerged later during the
1) Define the Problem and Model Objectives - An existing logical design. Note also that the problem is bounded -- no
microscopic stochastic simulation model of freeway traffic grades or horizontal curves are to be considered at this time.
does not consider lane-change operations. It has been See Figure 10.3 for the form of these distributions.
determined that this model’s results are unreliable as a result.
The purpose of this project is to introduce additional logic 3) Develop the Model - Since this lane-change model is to be
into the model to represent lane-changing operations. This introduced into an existing microscopic stochastic model,
addition should provide improved accuracy in estimating using a procedural language, it will be designed to utilize the
speed and delay; in addition it will compute estimates of lane existing software. The model logic moves each vehicle, each
changes by lane, by vehicle type and by direction (to the left time-step, t, starting with the farthest downstream vehicle,
and to the right). then moving the closest upstream vehicle regardless of lane
position, etc. At time, to, the vehicle states are shown in
2) Define the System - Figure 10.4(a).
a) A freeway of up to six lanes -- level tangent In developing the model, it is essential to identify the
b) Three vehicle types: passenger car; single-unit truck; independent functions that need to be performed and to
tractor-trailer truck segregate each function into a separate software module, or
c) Required inputs: traffic volume (varies with time); routine. Figure 10.5 depicts the structure -- not the flow --
distributions of free-speed, of acceptable risk (expressed of the software. This structure shows which routines are
in terms of deceleration rates if lead vehicle brakes), of logically connected, with data flowing between them. Some
motivation to change lanes, all disaggregated by vehicle routines reference others more than once, demonstrating the
type benefits of disaggregating the software into functionally
d) Drivers are randomly assigned an “aggressiveness index” independent modules.
ranging from 1 (very aggressive) to 10 (very cautious)
drawn from a uniform distribution to represent the range As indicated in Table 10.3 which presents the algorithm for
of human behavior. the Lane Change Executive Routine in both “Structured
English” or “pseudo-code” and as a flow chart, traffic

%5$)),& $,08/$7,21
Figure 10.3
Several Statistical Distributions.

%5$)),& $,08/$7,21
Figure 10.4
Vehicle Positions during Lane-Change Maneuver

%5$)),& $,08/$7,21
Figure 10.5
Structure Chart of Simulation Modules
Table 10.3
Executive Routine
For each vehicle, I:

CALL routine MOTIV to determine whether this driver is “motivated” to change lanes, now
IF so, THEN
CALL routine CANLN to identify which of neighboring lanes (if either) are
acceptable as potential target lanes
IF the lane to the right is acceptable, THEN
CALL routine CHKLC to determine whether a lane-change is feasible, now.
Set flag if so.
ENDIF
IF the lane to the left is acceptable, THEN
CALL routine CHKLC to determine whether a lane-change is feasible, now.
Set flag, if so.
ENDIF
IF both lane-change flags are set (lane-change is feasible in either direction), THEN
CALL routine SCORE to determine more favorable target lane
ELSE IF one lane-change flag is set, THEN
Identify that lane
ENDIF
IF a [favored] target lane exists, THEN
CALL routine LCHNG to execute the lane-change
Update lane-change statistics
ELSE
CALL routine CRFLW to move vehicle within this lane
Set vehicle’s process code (to indicate vehicle has been moved this time-step)
ENDIF
ELSE (no lane-change desired)
CALL routine CRFLW to move vehicle within its current lane
Set vehicle’s process code
ENDIF
continue...

%5$)),& $,08/$7,21
Figure 10.6
Flow Diagram.

%5$)),& $,08/$7,21
simulation models are primarily logical, rather than systems), are also common in models which simulate
computational in context. This property reflects the fact that human decisions.
traffic operations are largely the outcome of driver decisions
which themselves are [hopefully!] logical in context. As the LCHNG: After executing the subject vehicle’s lane-change
vehicles are processed by the model logic, they transition from activity, the logic performs some needed
one state to the next. The reader should reference Table 10.3 bookkeeping:
and Figures 10.3, 10.4, 10.5 and 10.6 to follow the discussion
given below for each routine. The “subject vehicle” is shown as At time, to + t, vehicle 19 acts as the leader for both
number 19 in Figure 10.4. vehicles 14 and 31, who must “follow” (and are
constrained by) its presence.
Executive: Controls the flow of processing, activating
(through CALLs) routines to perform the At time, to +2t, the logic asserts that the lane changer
necessary functions. Also updates lane-change (no. 19) has committed to the lane-change and no longer
statistics. influences its former follower, vehicle no. 31. Of course,
vehicle 14 now follows the lane-change vehicle, 19.
MOTIV: Determines whether a lane-change is required to
position the subject vehicle for a downstream 4) Calibrate the Model - Figures 10.3(b) and (c) are
maneuver or is desired to improve the vehicle’s distributions which represent the outcome of a calibration
operation (increase its speed). activity. The distribution of free-flow speed is site-specific
and can be quantified by direct observation (using paired
CANLN: Determines whether either or both adjoining lanes (1 loop detectors or radar) when traffic conditions are light --
and 3) are suitable for servicing the subject vehicle. LOS A.
CHKLC: Identifies vehicles 22 and 4 as the leader and The distribution of acceptable decelerations would be very
follower, respectively, in target Lane 1; and vehicles difficult to quantify by direct observation -- if not infeasible.
16 and 14, respectively, in target Lane 3. The car- Therefore, alternative approaches should be considered. For
following dynamics between the pairs of vehicles, 19 example...
and 22, then 4 and 19 are quantified to assess the
prospects for a lane-change to Lane 1. Subsequently, Gather video data (speeds, distance headway) of
the process is repeated between the pairs of vehicles, lane-change maneuvers. Then apply the car-
19 and 16; then 14 and 19; for a lane-change to Lane following model with these data to “back-out” the
3. If the gap is inadequate in Lane 1, causing an implied acceptable decelerations. From a sample
excessive, and possibly impossible deceleration by of adequate size, develop the distribution.
either the subject vehicle, 19, or the target follower,
4, to avoid a collision, then Lane 3, would be On a more macro level, gather statistics of lane-
identified as the only feasible target lane. In any case, changes for a section of highway. Execute the
CHKLC would identify either Lane 1 or Lane 3, or simulation model and adjust this distribution of
both, or neither, as acceptable target lanes at this acceptable deceleration rates until agreement is
time, depending on safety considerations. attained between the lane-changes executed by the
model, and those observed in the real world. This
SCORE: If both adjoining target lanes are acceptable, then this is tenuous since it is confounded by the other
heuristic algorithm emulates driver reasoning to model features, but may be the best viable
select the preferable target lane. It is reasonable to approach.
expect that the target lane with a higher-speed leader
and fewer vehicles -- especially trucks and buses -- It is seen that calibration -- the process of quantifying
would be more attractive. Such reasonableness model parameters using real-world data -- is often a
algorithms (which are expressed as “rules” in expert difficult and costly undertaking. Nevertheless, it is a

75$)),& 6,08/$7,21
necessary undertaking that must be pursued with Developing the experimental design of the
some creativity and tenacity. validation study, including a variety of "scenarios"
to be examined.
5) Model Verification - Following de-bugging,
verification is a structured regimen to provide Performing the validation study:
assurance that the software performs as intended - Executing the model using input data and
(Note: verification does not address the question, “Are calibration data representing the real-world
the model components and their interactions conditions.
correct?”). Since simulation models are primarily - Performing the hypothesis testing.
logical constructs, rather than computational ones, the
analyst must perform detailed logical path analyses. Identifying the causes for any failure to satisfy the
validation tests and repairing the model
Verification is performed at two levels and generally in accordingly.
the sequence given below:
Validation should be performed at the component
Each software routine (bottom-up testing) system level as well as for the model as a whole.
Integration of “trees” (top-down testing)... For example, Figure 10.7 compares the results
produced by a car-following model, with field data
When completed, the model developer should be collected with aerial photographs. Such face
convinced that the model is performing in accord with validation offers strong assurance that the model
expectations over its entire domain of application. is valid. This validation activity is iterative -- as
differences between the model results and the
6) Model Validation - Validation establishes that the real-world data emerge, the developer must
model behavior accurately and reliably represents the “repair” the model, then revalidate. Considerable
real-world system being simulated, over the range of skill (and persistence) are needed to successfully
conditions anticipated (i.e., the model's "domain"). validate a traffic simulation model.
Model validation involves the following activities:
7) Documentation - Traffic simulation models, as is the
Acquiring real-world data which, to the extent case for virtually all transportation models, are data
possible, extends over the model's domain. intensive. This implies that users must invest effort in
data acquisition and input preparation to make use of
Reducing and structuring these empirical data so these models. Consequently, it is essential that the
that they are in the same format as the data model be documented for...
generated by the model.
The end user, to provide a “friendly” interface to
Establishing validation criteria, stating the ease the burden of model application.
underlying hypotheses and selecting the statistical Software maintenance personnel.
tests to be applied. Supervisory personnel who must assess the
potential benefits of using the model.
10.7 Applying Traffic Simulation Models

Considerable skill and attention to detail must be exercised by simulation-based analysis. The following procedure is
the user in order to derive accurate and reliable results from a recommended:

75$)),& 6,08/$7,21
Figure 10.7
Comparison of Trajectories of Vehicles from Simulation Versus Field Data for Platoon 123.
Identify the Problem Domain - Pictorial: static graphics, animation.

What information is available as input and calibration data?
What highway facilities are involved? - Consider expected accuracy and reliability.
- Surface streets (grid, arterial, both), freeways, - Consider available budget for data acquisition.
rural roads, toll plaza.
What is the traffic environment? Are results needed on a relative or absolute basis?
- Autos, trucks, buses, LRT, HOV...
- Unsaturated, oversaturated conditions. What other functions and tools are involved?
What is the control environment? - Capacity analysis
- Signals (fixed time, actuated, computer- - Design
controlled), signs - Demand modeling
- Route guidance - Signal optimization
What is the size of the network and duration of the analysis Is the application real-time or off-line?
period?
Investigate Candidate Traffic Simulation Models
Define the Purpose of the Study
Identify strengths and limitations of each.
What information is needed from the simulation model? - Underlying assumptions
- Statistical: MOE sought, level of detail.

75$)),& 6,08/$7,21
- Computing requirements - Signal timing plans, actuated controller settings.

- Availability, clarity, completeness of documentation - Traffic volume and patterns; traffic composition.
- Availability, reliability, timeliness of software - Transit schedules
support - Other, as required.
Estimate extent and cost of data collection for calibration and Confirm the accuracy of these data through field observation.
input preparation.
Undertake field data collection for input and for calibration, as
Determine whether model features match problem needs. required.
Assess level of skill needed to properly apply model. Identify need for accurate operational traffic data: based on
model’s sensitivity site-specific features; accuracy requirements.
Determine compatibility with other tools/procedures needed for
the analysis. - Select representative locations to acquire these field
data.
Assess the Need to Use a Traffic Simulation Model - Collect data using video or other methods as
required: saturation flow rates at intersections; free-
Is traffic simulation necessary to perform the analysis of the flow speeds; acceptable gaps for permitted left-turns,
problem? etc.
- Are other tools adequate but less costly? - Accept model default values or other data from the
- Are your skills adequate to properly apply literature with great care if data collection is
simulation? infeasible or limited by cost considerations.
- Can the data needed by traffic simulation be
acquired? Model Calibration
Is traffic simulation highly beneficial even though not Calibration is the activity of specifying data describing traffic
necessary? operations and other features that are site-specific. These data
- Simulation results can confirm results obtained by may take the form of scalar elemets and of statistical
other tools. distributions that are referenced by the logic of stochastic
- Animation displays needed as a presentation simulation models. While traffic simulation models generally
medium. provide default value which represent average conditions for
these calibration data elements, it is the responsibility of the
If it is determined that traffic simulation is needed/advisable, analyst to quantify these data with field observations to the extent
continue. practicable rather than to accept these default values.
Select Traffic Simulation Model Model Execution
Relate relevant model attributes to problem needs. The application of a simulation model should be viewed as
performing a rigorous statistical experiment. The model must
Determine which model satisfies problem needs to the greatest first be executed to initialize its database so that the data
extent. Consider technical, and cost, time, available skills and properly represents the initial state of the traffic environment.
support, and risks factors. This requirement can reliably be realized if the environment is
initially at equilibrium.
Data Acquisition
Thus, to perform an analysis of congested conditions, the analyst
Obtain reliable records of required information. should design the experiment so that the initial state of the traffic
- Design drawings for geometrics. environment is undersaturated, and then specify the changing

%5$)),& $,08/$7,21
conditions which, over time, censors the congested state which Anomolous results (e.g., the creation and growth of
is of interest. Similarly, the final state of the traffic environment queues when conditions are believed to be
should likewise be undersaturated, if feasible. undersaturated) can be examined and traced to valid,
incongested behavior; to errant input specifications; or to
It is also essential that the analyst properly specify the dynamic model deficiencies.
(i.e., changing) input conditions which describe the traffic
environment. For example, if one-hour of traffic is to be If the selected traffic simulation lacks an animation feature or if
simulated, the analyst should always specify the variation in questions remain after viewing the animation, then the following
demand volumes -- and in other variables -- over that hour at an procedures may be applied:
appropriate level of detail rather than specifying average,
constant values of volume. Execute the model to replicate existing real-world
conditions and compare its results with observed
Finally, if animation displays are provided by the model, this behavior. This “face validation”, which is recommended
option should always be exercised, as discussed below. regardless of the model selected, can identify model or
implementation deficiencies.
Interpretation of Simulation Results
Perform “sensitivity” tests on the study network by
Quite possibly, this activity may be the most critical. It is the varying key variables and observing model responses in
analyst who must determine whether the model results constitute a carefully designed succession of model executions.
a reasonable and valid representation of the traffic environment Plot these results. A review will probably uncover the
under study, and who is responsible for any inferences drawn perceived anomalies.
from these results. Given the complex processes taking place in
the real-world traffic environment, the analyst must be alert to Table 10.1 lists representative data elements provided by traffic
the possibility that (1) the model’s features may be deficient in simulation models. Figure 10.8 shows typical graph displays
adequately representing some important process; (2) the input while Figure 10.9 displays a "snapshot" of an animation screen.
data and/or calibration specified is inaccurate or inadequate; (3)
the results provided are of insufficient detail to meet the project
objectives; (4) the statistical analysis of the results are flawed (as Note that all the graphical displays can be accessed interactively
discussed in the following section); or (5) the model has “bugs” by the user, thus affording the user an efficient means for
or some of its algorithms are incorrect. Animation displays of extracting the sought information and insights from the mass of
the traffic environment (if available) are a most powerful tool for data compiled by the simulation model.
analyzing simulation results. A careful and thorough review of
this animation can be crucial to the analyst in identifying: Proper output analysis is one of the most important aspects of
any simulation study. A variety of techniques are used,
Cause-and-effect relationships. Specifically the origins particularly for stochastic model output, to arrive at inferences
of congested conditions in the form of growing queues that are supportable by the output. A brief exposition to output
can be observed and related to the factors that caused it. analysis of simulation data is paresented next.
10.8 Statistical Analysis of Simulation Data

In most efforts on simulation studies, more often than not a large a heuristic model, which is then programmed on the computer,
amount of time and money is spent on model development and and conclude with a single run of the program to yield an
coding, but little on analyzing the output data. Simulation answer. This is a result of overlooking the fact that a simulation
practitioners therefore have more confidence in their results than is a sampling experiment implemented on a computer and
is justified. Unfortunately, many simulation studies begin with therefore needs appropriate statistical techniques for its design,

%5$)),& $,08/$7,21
Figure 10.8
Graphical Displays.

%5$)),& $,08/$7,21
Figure 10.9
Animation Snapshot.
analysis, and implementation. Also, more often than not, output Typical goals of analyzing output data from simulation
data from simulation experiments are auto correlated and experiments are to present point estimates of the measures of
nonstationary. This precludes its analysis using classical effectiveness (MOE) and form confidence intervals around these
statistical techniques which are based on independent and estimates for one particular system design, or to establish which
identically distributed (IID) observations. simulated system is the best among different alternate

%5$)),& $,08/$7,21
configurations with respect to some specified MOE. Point system using independent random number streams across
estimates and confidence intervals for the MOEs can be obtained replications.
either from one simulation run or a set of replications of the
10.9 Looking to the Future

With the traffic simulation models now mounted on high- - Combining simulation with Artificial Intelligence
performance PCS, and with new graphical user interfaces (GUI) (AI) software. The simulation can provide the
becoming available to ease the burden of input preparation, it is knowledge base in real-time or generate it in
reasonable to expect that usage of these models will continue to advance as an off-line activity.
increase significantly over the coming years. - Integrating traffic simulation models with other tools
such as: transportation demand models, signal
In addition, technology-driven advances in computers, combined optimization models, GIS, office suites, etc.
with the expanding needs of the Intelligent Transportation - Providing Internet access.
Systems (ITS) program, suggest that the new applications of
traffic simulation can contribute importantly to this program. Real-time simulators which replicate the performance of
Specifically: TMC operations. These simulators must rely upon
simulation models as “drivers” to provide the real-world
Simulation support systems of Advanced Traffic stimuli to ITS real-time software being tested. Such
Management Systems (ATMS) in the form of: simulators are invaluable for:
- Off-line emulation to test, refine, evaluate, new real- - Testing new ATMS concepts prior to deployment.
time control policies. - Testing interfaces among neighboring TMCs.
- On-line support to evaluate candidate ad-hoc - Training TMC operating personnel.
responses to unscheduled events and to advise the - Evaluating different ATMS architectures.
operators at the Traffic Management Center (TMC) - Educating practitioners and student.
as to the “best” response. -Real- - Demonstrating the benefits of ITS programs to state
time component of an advanced control/guidance and municipal officials and to the public through
strategy. That is, the simulation model would be a animated graphical displays and virtual reality.
component of the on-line strategy software.
References
Gartner, N.H. and D. L. Hou (1992). Comparative Evaluation Korve Engineers (1996). State Route 242 Widening Project -
of Alternative Traffic Control Strategies,” Transportation Operations Analysis Report to Contra Costa Transportation
Research Record 1360, Transportation Research Board. Authority.
Mahmassani, H.S. and S. Peeta (1993). Network Performance Rathi, A.K. and E.B. Lieberman (1989). Effectiveness
Under System Optimal and User Equilibrium Dynamic of Traffic Restraint for a Congested Urban Network: A
Assignments: Implications for Advanced Traveler Simulation Study. Transportation Research Record 1232.
Information Systems Transportation Research Record 1408,
Transportation Research Board.

KINETIC THEORIES
BY PAUL NELSON*
*
Professor, Department of Computer Science, Texas A&M University, College Station, TX 77843-3112

KINETIC THEORIES
Possible objectives and applications for kinetic theo-

11 Kinetic Theories ries of vehicular traffic are considered. One of these is the
traditional application to the development of continuum mod-
Criticisms and accomplishments of the Prigogine-Herman
els, with the resulting microscopically based coefficients.
kinetic theory are reviewed. Two of the latter are identified as
However, modern computing power makes it possible to con-
possible benchmarks, against which to measure proposed
sider computational solution of kinetic equations per se, and
novel kinetic theories of vehicular traffic. Various kinetic
therefore direct applications of the kinetic theory (e.g., the
theories that have been proposed in order to eliminate defi-
kinetic distribution function). It is concluded that the primary
ciencies of the Prigogine-Herman theory are assessed in this
applications are likely to be found among situations in which
light. None are found to have yet been shown to meet both of
variability between instances is an important consideration
these benchmarks.
(e.g., travel times, or driving cycles).
issues for kinetic models of vehicular traffic. The status of
11.1 Introduction various kinetic models will also be reviewed, in terms of
achieving two objectives that seem appropriate to designate as
On page 20 of their well-known monograph on the kinetic
benchmarks, primarily on the basis that the seminal kinetic
theory of vehicular traffic, Prigogine and Herman (1971)
model of Prigogine and Herman (1971) has been shown to
summarize possible alternate forms of the relaxation term in
meet those objectives.
their kinetic equation of vehicular traffic. They conclude this
It seems appropriate to view kinetic models as occu-
discussion by issuing the invitation that “the reader may, if he
pying a point on the model spectrum that is intermediate be-
is so inclined, work out the theory using other forms of the
tween continuum (e.g., hydrodynamic) models and micro-
relaxation law.” This invitation to explore alternate kinetic
scopic (e.g., car-following or cellular automata) models.1 One
models of vehicular traffic has subsequently been accepted by
of the primary applications of kinetic models is to obtain con-
a number of workers, most notably by Paveri-Fontana (1975)
tinuum models in a consistent manner from an underlying
and by Phillips (1979, 1977), and more recently by Nelson
microscopic model of driver behavior. (See Nelson (1995b)
(1995a), and by Klar and Wegener (1999). The existence of
for further thoughts on the role of kinetic models of vehicular
this variety of kinetic models of vehicular traffic raises the
traffic as a bridge from microscopic models to macroscopic
issue of how one chooses between them in any particular ap-
models.) However, computing power now has advanced to
plication; more generally there arises the issue of the types of
the point that it is practical to consider computational solution
applications for which any kinetic model has a role. In these
of kinetic equations per se. This opens the door to the realistic
lights, the primary objective of this chapter is to address ques-
possibility of applying kinetic models directly to the simula-
tions related to what might reasonably be expected from a
tion of traffic flow. This is a qualitatively different situation
good kinetic theory of vehicular traffic.
from that prevailing in the 1960’s, when kinetic models of
The approach presented here is substantially influ-
vehicular traffic were initially proposed by Prigogine, Herman
enced by the work of Nagel (1996), who gave an excellent
and co-workers. (See Prigogine and Herman, 1971, and works
review of a variety of types of models of vehicular traffic,
cited therein.)
including continuum (“hydrodynamic”), car-following and
particle hopping (cellular automata) models. In particular, he
The specific further contents of this chapter are as
has emphasized that: i) any model necessarily represents some
follows. In Section 2 below, the status of the Prigogine-
compromise in terms of its fidelity in describing the reality it is
intended to represent; ii) different types of models represent
engineering judgements as to the relative importance of reso- 1
The word mesoscopic has come into recent vogue to describe mod-
lution, fidelity and scale for the particular application at hand. els that are, in some sense, intermediate between macroscopic and
To some extent, this chapter is intended to address similar microscopic models.
11-1
KINETIC THEORIES
Herman (1971) kinetic model of vehicular traffic is reviewed. a) the zero order moment of f(x,v,t),
The intent of this review is to provide an evenhanded discus-
f
sion of both the deficiencies and signal accomplishments of
this seminal kinetic theory of vehicular traffic. Two of these c ( x, t ) ³ f ( x, v, t ) dv,
0
accomplishments are suggested as benchmarks that should
minimally be met by any proposed novel kinetic model of is vehicular density.
vehicular traffic. In Section 3 alternate kinetic models that b) the ratio of the first and zero order moments,
have been proposed in the literature are assessed against these
benchmarks, and none are found that yet have been shown to f
meet both of them. Both of these benchmarks relate to the

equilibrium solutions of the Prigogine-Herman kinetic equa-
v ( x, t ) ³ vf ( x, v, t ) dv
0
c ( x, t )
tion, and one of them relates to the recent result of Nelson and
Sopasakis (1998) to the effect that under certain circumstances is mean vehicular speed;
– particularly for sufficiently congested traffic – the Prigog- c) P is passing probability;
ine-Herman model admits a two-parameter family of equilib- d) T is the relaxation time;
rium solutions, as opposed to the one-parameter (density) fam- e) f0 is the corresponding density function for the desired
ily that would be expected classically. speed of vehicles;
In Section 4, the role of kinetic equations as a bridge f) f is the density function for the distribution of vehicles in
from microscopic to continuum models is considered. Section phase space, so that
5 is devoted to consideration of the potential applications of
v2 x2
the solution of kinetic equations per se.
³ ³ f ( x, v, t ) dv dx
11.2 Status of the Prigogine-Herman Ki- v1 x1
netic Model is the expected number of vehicles at time t that have po-
The kinetic model of Prigogine and Herman (1971) is summa- sition between x1 and x2 and speed between v1 and
rized in Subsection 2.1. A number of published criticisms of v2 ( x1 d x2 and v1 d v2 ).
this model, along with alternative models that have been sug- The second term on the left-hand side of Eq. (1), the
gested to overcome some of these criticisms, are reviewed in streaming term, represents the rate of change of the density
Subsection 2.2. In Subsection 2.3 two significant accom- function due to motion of the traffic stream, absent any
plishments of the Prigogine-Herman theory are described, and changes of velocity by vehicles. The first term on the right-
suggested as benchmarks against which novel kinetic theories hand side, which is often called the relaxation term, is the
of vehicular traffic should be measured. contribution to this rate of change that stems from changes of
vehicular speed associated with passing or other causes of
acceleration. The second term on the right-hand side, the
11.2.1 The Prigogine-Herman Model slowing-down term, stems from deceleration of vehicles that
The kinetic equation of Prigogine and Herman is overtake slower-moving vehicles. The relaxation term is phe-
nomenological in nature, in that it is based on the underlying
assumption that increases in vehicular speed cause the actual
wf wf f f0 density to “relax” toward the desired density with some char-
v c(v v)(1 P) f . (11. 1) acteristic time T. By contrast, the slowing-down term can be
wt wx T obtained from basic physical arguments, albeit with idealized
(1)
assumptions such as instantaneous deceleration, treatment of
Here the various symbols have the following meanings: vehicles as point particles (i.e., neglect of the positive length
of vehicles), and the validity of what Paveri-Fontana (1975)
terms vehicular chaos. The validity of both of these particular
forms of the rates of change due to changes of speed has been
11-2
KINETIC THEORIES
questioned, as will be briefly discussed in the following sub- variable, on the same footing as the actual speed, and he pro-
section. vides a modification of the Prigogine-Herman equation that
A kinetic equation generally is an equation that in accomplishes precisely that.
principle, subject to appropriate initial and boundary condi- Prigogine and Herman (1971, Section 3.4, and 1970)
tions, can be solved for the density function f, as defined dispute the claim of Munjal and Pahl (1969) to the effect that
above. Some kinetic equations that are alternatives to that of “the validity of the interaction term (i.e., the Prigogine-
Prigogine and Herman are discussed in Section 3 following Herman slowing-down term) is limited to traffic situations
where no vehicles are queuing” (parenthetical clarification
11.2.2 Criticisms of the Prigogine-Herman added). Current opinion seems inclined to agree with Paveri-
Model. Fontana (1975) that on balance Munjal and Pahl have the bet-
ter of this particular discussion. However, traffic on arterial
The first published serious critique of the Prigogine-Herman roads, for which signalized intersections necessarily enforce
kinetic equation seems to be the work of Munjal and Pahl the formation of platoons, is the only situation that seems thus
(1969). These workers raise a number of questions,2 but the to be definitely excluded from the domain of the Prigogine-
most fundamental of these fall into one of the following two Herman kinetic equation. In particular, it is not a priori clear
categories: that the same objection is valid for the stop-and-go traffic that
1. The validity of the slowing-down term (denoted the “in- seems to characterize congested traffic on freeways. Nelson
teraction term” by these authors) is doubtful in the pres- (1995a) has noted that a correlation model is generally needed
ence of “queues” (or “platoons”) of vehicles. This stems to obtain a kinetic equation, and vehicular chaos is simply one
from the fact that the correlation inherent in platoons in- instance of a correlation model. Other correlation models,
validates the assumption of vehicular chaos (Paveri- which would lead to a kinetic equation other than that of
Fontana, 1975), which assumption underpins the particu- Prigogine and Herman, conceivably could better treat pla-
lar form of the slowing-down term in the Prigogine- toons, at least under restricted circumstances. Approaches
Herman kinetic equation. (e.g., Prigogine and Andrews, 1960; Beylich, 1979), in which
2. The absence of a derivation of the relaxation term from multiple-vehicle density functions appear as the unknowns to
first principles raises general questions regarding its va- be determined, also offer the potential ability to treat queues
lidity. The validity of the specific expression (in terms of within the spirit of the kinetic theory.
c) used by Prigogine et al. for the relaxation time T has Nelson (1995a) introduced the concepts of a mechani-
therefore “not been proven.” Further, it is therefore also cal model and a correlation model as the fundamental ingredi-
difficult to “conceive the meaning of the relaxation time” ents of any kinetic equation. This work was motivated pre-
and therefore “define a method for its experimental de- cisely by the desire to obtain forms of the speeding-up term
termination.” that are based upon at least the same level of first principles as
the classical derivations of the Prigogine-Herman slowing-
In addition to noting the first of these concerns, Paveri- down term. Klar and Wegener (1999) used this approach to
Fontana (1975) argues forcefully that it is fundamentally in- obtain a kinetic equation for traffic flow that accounts for the
correct to treat the desired speed as a parameter, as in done in spatial extent of vehicles. The treatment of vehicles as
the Prigogine-Herman kinetic equation. Rather, he suggests “points” of zero length is an idealization underlying the
the desired speed must be taken as an additional independent Prigogine-Herman kinetic equation that seems not to have
been extensively discussed in the earlier literature on traffic
2 flow. Klar and Wegener (1999) show that including the length
Other concerns relate to: i) The necessity to include time depend-
ence in the desired speed distribution, owing to the normalization
of vehicles has a significant quantitative effect upon the value
v max
of some coefficients in associated continuum models. The
observational measurements of the relaxation time by Edie,
³f
0
0 ( x, v, t ) dv c( x, t ) ; and ii) the interpretation and func- Herman and Lam (1980) also bear mentioning.
The arguments of Paveri-Fontana (1975) that the de-
tional dependence of the passing probability, P. sired speed must appear as an independent variable in any
kinetic equation, so that the density function depends upon the
11-3
KINETIC THEORIES
desired speed, as well as position, actual speed and time, seem More recently, Nelson and Sopasakis (1998) showed
to be quite convincing. In order to avoid this complexity, that if one relaxes the assumption of Prigogine and Herman
some workers (e.g., Nelson, 1995a) choose to focus upon that there exist drivers having arbitrarily small desired speeds,
models in which all drivers have the same desired speed. then at sufficiently high densities the equilibrium solution is a
Paveri-Fontana (1975) represents his modification of the two-parameter family. This contrasts with the one-parameter
Prigogine-Herman equation, to include desired speed as an (typically taken as density) family that occurs at low densities,
independent variable, as valid only for dilute traffic. How- even at all densities under the restrictive assumption of
ever, as suggested above, it is not completely clear that this Prigogine and Herman (1971).3
restriction is required, unless the dense traffic also includes a The consequence of the equilibrium solutions of Nel-
significant fraction of the vehicles in platoons. son and Sopasakis (1998) for the attendant traffic stream
model will now be briefly described. Let
11.2.3 Accomplishments of the Prigogine- w
f 0 (v )
Herman Model F (] ; c) : ³ c(v ] ) dv,
In view of the deficiencies chronicled in the preceding subsec- w
tion, why would anyone deem the Prigogine-Herman kinetic where w- and w+ are respectively the lower and upper bound
equation to be of any interest? That question is answered in on the desired speeds. Then there exists a positive critical
this subsection, by describing two significant results that stem density, denoted ccrit, and defined as the unique root (in c) of
from the Prigogine-Herman model. the equation F(0;c) = cT(1-P), such that the dependence of
First, Prigogine and Herman (1971, Chap. 4) demon- mean speed upon density is as follows. Let ]*=]*(c) be the
strated, under the somewhat restrictive assumption that there unique root (in ] ) of F(];c) = cT(1-P). If 0d cdccrit , then
exist drivers desiring arbitrarily small speeds, that one can ]*d0, and the mean speed is given by
obtain traffic stream models (fundamental diagrams,
1
speed/density relations), say q=Q(c), from the equilibrium v v (c ) ] *.
solutions (i.e., the solutions that are independent of space and cT (1 P)
time) of their kinetic equation. (Here q is vehicular flow, and However, if c>ccrit , then ]*>0, and the mean speed is given by
c is, as above, vehicular density.) The procedure is precisely
analogous to that giving rise to the Maxwellian distribution 1
v v (c, ] ) ] ,
and the ideal gas law, when applied to the Boltzmann equation cT (1 P)
of the kinetic theory of gases. Further, at high concentrations
where now ] can take on any value such that 0 d ] d
the equilibrium solution is bimodal; that is, it displays two
(local) maxima in speed, in qualitative agreement with the min{]*,w-}. The parameter ] is the speed of the embedded
observations of Phillips (1977, 1979). (See the following sec- collective flow, and the preceding equation for the mean speed
tion for more details of these works.) One of these modes shows that the overall mean speed increases with increasing
corresponds to a modification of the distribution of desired speed of the embedded collective flow. Figure 11.1 shows a
speeds, and the other (under the assumptions of Prigogine and three-dimensional graphical representation of the resulting
Herman) to platoon flow in the rather extreme case of stopped “traffic stream model,” for a particular hypothetical desired
traffic (i.e., zero speed). This “multiphase” aspect of con-
gested traffic flow has subsequently been rediscovered by a
number of workers. Note that this approach gives rise to a
traffic stream model from an underlying microscopic model, 3
In some of their work, Prigogine and Herman (1971, Section
via the equilibrium solution of a corresponding kinetic equa- 4.4, esp. Fig. 4.8 and the related discussion) did permit posi-
tion. Such a theoretical development contrasts with statements tive lower bounds for the set of desired speeds, but for reasons
sometimes encountered to the effect that traffic stream models that seem unclear at this point their attendant analysis did not
must be based upon observational data. identify the full two-parameter range of equilibrium solutions
at higher densities.
11-4
KINETIC THEORIES
60
50
40
Mean speed (mph)
30
20
10
0
0
norm
alize 0.5
d co 1 0 −20 −40 −60 −80
ncentr 40 20
atio speed of collective flow (mph)
n
Figure 11-1 Dependence of the mean speed upon density normalized to jam density, K=c/cP, for jam density cP = 200 vpm, P=1-
K, T=WK/(1-K), with W=0.003 hours, and a uniform desired speed distribution from 40 to 80 mph.
speed distribution. See Nelson and Sopasakis (1998) for more One certainly can envision more ambitious objectives for a
details. kinetic theory of vehicular traffic than these two. Some possi-
The significance of this three-dimensional presenta- ble such objectives are discussed further in Sections 4 and 5
tion of a traffic stream model lies in the fact that it is consis- below. However, given that the seminal Prigogine-Herman
tent with the well-known tendency (e.g., Drake, Schofer and kinetic equation of vehicular traffic does accomplish at least
May, 1967) for traffic flow data to be widely scattered at high these objectives, it seems appropriate to suggest them as
densities. The effort to explain this tendency has spawned a minimal benchmarks that should be met by any alternative
number of theories (e.g., Ceder, 1976; Hall, 1987; Disbro and kinetic equations that might be proposed. In the following
Frame, 1989). The explanation in terms of an embedded col- section some of the alternative kinetic equations that have
lective flow seems possibly preferable to these, in that it de- been proposed, as described in the preceding subsection, are
rives from the kinetic theory, which is a well-known branch of assessed against these benchmarks.
traffic flow theory, as opposed to requiring some novel ad hoc
theory. 11.3 Other Kinetic Models
Thus, the Prigogine-Herman kinetic equation has
Both benchmarks suggested in the preceding subsection have
equilibrium solutions that both reproduce the observed bi-
to do with the equilibrium solutions of the kinetic equation of
modal distribution of speeds at high densities, and provide
interest. The equilibrium solutions of the Paveri-Fontana
traffic stream models that reproduce qualitatively the well-
(1975) generalization of the Prigogine-Herman kinetic equa-
known result that at sufficiently high densities mean speeds
tion, as described in Subsection 2.2, do not seem to have been
and flows do not depend exclusively upon vehicular density.
definitively ascertained. Indeed, Helbing (1996), who has
11-5
KINETIC THEORIES
extensively applied the Paveri-Fontana kinetic model in his the first of the benchmarks suggested in the preceding section.
recent works on the kinetic theory of vehicular traffic, states, However, the equilibrium solutions of the kinetic equation of
in regard to these equilibrium solutions, that “unfortunately it Nelson (1995a) are such that it clearly does not meet the sec-
seems impossible to find an analytical expression ….” He ond benchmark (i.e., does not predict scattered flow data un-
then indicates that “empirical data and microsimulations” sug- der congestion). It is possible that the underlying mechanical
gest these equilibrium solutions are “approximately a Gaus- model could be modified to attain this objective, but that has
sian.” Note that Gaussians are not bimodal. Thus, the Paveri- not been demonstrated.
Fontana model does not seem to have been shown to satisfy Klar and Wegener (1999) use numerical techniques
either of the benchmarks suggested in the preceding subsec- to obtain equilibrium solutions of their kinetic equations.
tion. They do not explicitly present corresponding traffic stream
Phillips (1977, 1979) develops yet another kinetic models. Their numerical equilibrium solutions do not display
equation that is an alternative to the original Prigogine- two modes. It might be difficult to obtain the lower mode,
Herman kinetic model. However, this development seems which typically appears as a delta function, by a strictly nu-
predicated on a form of the corresponding equilibrium solu- merical approach.
tion that ignores the considerations that led Prigogine and Table I summarizes the status of the various kinetic
Herman to the “lower mode” of their bimodal equilibrium models mentioned here, as regards their ability to meet the two
solution; cf. Eq. (4) of Phillips, 1979. Phillips compared benchmarks delineated in Subsection 2.3.
(sketchily in Phillips, 1979, but exhaustively in Phillips, 1977)
Table 11-I Status of various kinetic models with respect to
the equilibrium solution of his kinetic model against measured
the benchmarks of Subsection 11.2.3
speed distributions. With one possibly important exception,
the agreement seems reasonable. One therefore expects good Benchmark Bimodal equilib- Equilibrium with
agreement between the traffic stream model obtained theoreti- Kinetic Model rium solutions? scattered flows at
cally from the equilibrium solution and that obtained observa- high densities?
tionally, although Phillips does not explicitly effect such com- Prigogine-Herman
parisons. The exception is that a large amount of the data in- (1971) yes yes
dicates a bimodal equilibrium solution; cf. Figs. 3 and 4 of Paveri-Fontana (1975) ? ?
Phillips, 1979, and numerous figures in Phillips, 1977. Thus, Phillips (1977, 1979) no no
although the bimodal nature of an equilibrium solution is Nelson (1995a) yes no
missed by the theoretical analysis, it is supported by the asso- Klar-Wegener (1999) no? no
ciated observations. In summary, it seems likely that the ki-
netic equation of Phillips (1979, 1977) meets the first bench-
mark suggested in the preceding section, and possible that a
11.4 Continuum Models from Kinetic Equa-
mathematical reassessment of its equilibrium solutions will tions
reveal that it meets the second of these benchmarks. How- Continuum models historically have played an important role
ever, neither of these conclusions has yet been conclusively in traffic flow theory. They have been obtained either by sim-
established. ply writing them as analogs of some corresponding fluid dy-
Nelson (1995a) obtained a specific kinetic equation namical system (e.g., Kerner and Konhäuser, 1993), or by
for purposes of providing a concrete illustration of his pro- rational developments from some presumably more basic mi-
posed general methodology for obtaining speeding-up (and croscopic (e.g., car-following) model of traffic flow. In the
slowing-down) terms based on first principles (i.e., appropri- latter case the continuum equations can be developed either
ate mechanical and correlation models). In subsequent work directly from the underlying microscopic model that serves as
(Nelson, Bui and Sopasakis, 1997) it was shown that this ki- the starting point, or a kinetic model can play an intermediary
netic equation provides a theoretical traffic stream model that role between the microscopic and continuum models. For
agrees well with classical traffic stream models, except near early examples of the former approach, through the steady-
jam density. It has further been shown (Bui, Nelson and So- state solutions of car-following models, see numerous refer-
pasakis, 1996) that a simple modification of the underlying ences cited in Nelson, 1995b. Nagel (1998) presents a more
correlation model removes the incorrect behavior near jam modern approach, through appropriate formal (“fluid-
density. Thus, this kinetic equation has been shown to meet dynamical”) limits of particle-hopping models.
11-6
KINETIC THEORIES
Here the primary interest is, of course, in approaches developed the zero-order (again Lighthill-Whitham) and first-
to continuum models that use a kinetic intermediary to the order Hilbert expansions, and the second-order Chapman-
underlying microscopic model. Such approaches often (e.g., Enskog expansion, for the Prigogine-Herman kinetic model.
Helbing, 1995) follow the route of first taking the first few Interest in continuum models of traffic flow seems
(one or two) low-order polynomial moments of the kinetic likely to continue, as applications exist within the space of
equation, then achieving closure via ad hoc approximations. resolution/fidelity/scale requirements for which continuum
An alternate approach, via certain formal asymptotic expan- models are deemed most suitable. Along with this, interest in
sions (e.g., Hilbert or Chapman-Enskog expansions) is often the use of kinetic models of vehicular traffic as a basis for
used in the kinetic theory of gases (e.g., Grad, 1958). In this continuum models seems likely to continue. For example, the
approach, the number of polynomial moments of the kinetic venerable Lighthill-Whitham (1955) model is widely viewed
equation that are taken tend to be determined by the number of as the most basic continuum model of traffic flow. But a suit-
invariants that are defined by the dynamics of the microscopic able traffic stream model is an essential ingredient of the
model of the interaction between the constituent “particles” Lighthill-Whitham model. Thus, traffic stream models are an
(vehicles, for traffic flow) of the system. This approach leads important part of continuum models, as well as being of inter-
to a hierarchy of continuum models (e.g., the Euler/Navier- est in their own right. Therefore, both of the benchmarks de-
Stokes/Burnett/super-Burnett equations of fluid dynamics), as marcated in the preceding section can be viewed as related to
opposed to the single continuum equation that tends to result the issue of how well a particular kinetic model performs in
from formal limits of microscopic models. At all levels of this terms of providing a particularly low-order continuum model,
hierarchy the parameters of the resulting continuum model are specifically the Lighthill-Whitham model.
expressed in terms of those of the underlying microscopic
model. 11.5 Direct Solution of Kinetic Equations
Nelson and Sopasakis (1999) applied the Chapman-
Along with the traditional application of kinetic models of
Enskog expansion to the Prigogine-Herman (1971) kinetic
vehicular traffic to rational development of continuum models
equation. In the region below the critical density described in
from microscopic models, as described in the preceding sec-
Subsection 11.2 the lowest (zero) order expansion was found
tion, modern computers permit consideration of the utility of
to be a Lighthill-Whitham (1955) continuum model, with as-
kinetic models in their own right, rather than merely as tools
sociated traffic stream model corresponding to the one-
that can be used to construct continuum models. In this re-
parameter family of equilibrium solutions. The next highest
spect, there are two substantive issues:
(first-order) solution was found to be a diffusively corrected
i) How can one solve kinetic equations, to obtain the distri-
Lighthill-Whitham model,
bution function (f)?
wc w w ª wc º ii) Given this distribution function, what applications of it
>Q(c)@ « D (c ) » , (11.2)
can usefully be made?
wt wx wx ¬ wx ¼ As regards the first issue, Hoogendoorn and Bovy (to appear)
have employed Monte Carlo (i.e., simulation-based) tech-
where now both the flow function Q(c) and the diffusion coef-
niques for the computational solution of a kinetic equation of
ficient D(c) are known in terms of the density c and the pa-
vehicular traffic that builds upon the earlier work of Paveri-
rameters of the Prigogine-Herman kinetic model. This result
Fontana (1975). By contrast, in the kinetic theory of gases
is perhaps somewhat surprising, as one might reasonably have
there exists a significant body of knowledge (e.g., Neunzert
expected rather a continuum higher-order model of the type
and Struckmeier, 1995, and other works cited therein) relative
suggested by Payne (1971). Figure 11.2 illustrates how an
to the deterministic computational solution of kinetic equa-
initial discontinuity between an upstream higher-density re-
tions. This knowledge base undoubtedly could be invaluable
gion and a downstream lower-density region will tend to dis-
in attempting to develop similar capabilities for vehicular traf-
sipate according to the diffusively corrected Lighthill-
fic, but the equations are sufficiently different from those aris-
Whitham model, as opposed to the shock wave predicted by
ing in the kinetic theory of gases so that considerable further
ordinary Lighthill-Whitham theory, which is given by Eq.
development is likely to be necessary. This
(11.2) with D { 0. See Nelson (2000) for more details of the
example underlying this figure. Sopasakis (2000) has also
11-7
KINETIC THEORIES
250
200
density (vehicles per mile)
150
100
50 0.3
0.2
0
−2 0.1
−1.5 −1 −0.5 0 0.5 1 0
1.5 2 time (hours)
distance (miles)
Figure 11-2 Evolution of the flow, according to a diffusively corrected Lighthill-Whitham model, from initial conditions con-
sisting of 190 vpm downstream of x=0 and 30 vpm upstream. See Nelson (2000) for details of the traffic stream model (flow
function) and diffusion coefficient.
development is unlikely to occur in the absence of a relatively 1. Single-instance sampling: The sample space consists of
clear vision as to the uses that would be made of it. Therefore, all vehicles present on a specified road network at a spe-
the focus here primarily will be on the second of these issues. cific designated time.
Any consideration of applications of the distribution 2. Ensemble sampling: The sample space consists of all
function of a kinetic theory requires a consideration of its in- vehicles present, at a designated time, on one of an en-
terpretation. It is a statistical distribution function. The tradi- semble of identical road networks.
tional interpretation of such a distribution function is that it For example, the Houston freeway network at 5:00 p.m. on
describes the frequency with which certain properties occur Wednesday, July 15, 1998 would be a reasonable sample
among samples drawn from some sample space. In the kinetic space for single-instance sampling. On the other hand, the
theory of vehicular traffic, the samples are vehicles, and the Houston freeway network at 5:00 p.m. on all midweek work-
properties of interest are the positions and speeds of the vehi- days during 1998 for which dry weather conditions prevailed
cles; however, there are two fundamentally different possible would be a reasonable approximation of a sample space suit-
interpretations of the underlying sample space: able for ensemble sampling.
The difference between these two interpretations is
subtle, but it has profound consequences. Traffic theorists
11-8
KINETIC THEORIES
normally tend to think in terms that are most consistent with approach to estimation of fuel emissions (cf. Carson and Aus-
single-instance sampling. But any attempt to apply the kinetic tin, 1997).
theory within that interpretation implies the intention to pre- The crucial question underlying any potential appli-
dict, at some level of approximation, the evolution of traffic cation of kinetic models is whether a kinetic model can be
for that specific instance, given suitable initial and boundary found that has sufficient fidelity and resolution for the particu-
conditions for the distribution function. It seems somewhat lar application, and that can be solved on the necessary scale
questionable that this is attainable over any significant dura- using available computational resources. The answer to that
tion. (The ``rolling horizon’’ approach often applied to pre- clearly depends upon the specific details of the particular ap-
diction of traffic flow is a tacit admission of the significance plication, and any such proposed application of a specific ki-
of this issue.) On the other hand, the ensemble sampling in- netic model must be validated against actual observations.
terpretation implies only the intent to predict the likelihood However, data of sufficiently high quality to permit such vali-
with which various outcomes will occur. This intuitively dations are both rare and expensive to obtain. Under these
seems much more achievable (cf. p. 10 of Asimov, 1988). circumstances, it seems appropriate to use microscopic models
Thus the more subtle ensemble sampling interpretation leads (e.g., cellular automata) as a framework within which initially
to an apparently more achievable objective than does the more to vet proposed kinetic models.
obvious single-instance interpretation. For that reason, the Specifically, it seems worthwhile to employ micro-
ensemble sampling interpretation seems more likely to lead to scopic models to study the following:
potential direct applications of the kinetic theory. HYPOTHESIS: The multiparameter family of equilibrium solu-
The fundamental advantage of kinetic models over tions of the Prigogine-Herman kinetic model found by Nelson
continuum models is that the kinetic distribution function pro- and Sopasakis (1998), with its attendant traffic stream surface
vides an estimate of the variability (over various instances of (rather than the classical curve), reflects the fact that actual
the ensemble, under the ensemble-sampling interpretation) of traffic has a number of spatially homogeneous equilibrium
densities and speed at specific times and locations, whereas states (with different average speeds) corresponding to the
continuum models provide estimates of only the mean (pre- same density.
sumably over the ensemble) of these quantities. If the quantity If this hypothesis is true, then presumably different initial con-
(function of position and time) of interest in a particular appli- figurations of a traffic stream have the possibility to approach
cation is not highly variable between instances within the en- different states in the long-time limit, even though their densi-
semble, or if that variability is not of interest, then presumably ties are the same on the macroscopic scale. Results reported
one should choose a continuum model, or perhaps an even by Nagel (1996, esp. Sec. V) tend to confirm this hypothesis.
more highly aggregated model. On the other hand, if this
variability is both of significant magnitude and important to
the issue under study, then kinetic models might be a useful References
alternative to the computationally more expensive possibility
Asimov, I. (1988). Prelude to Foundation. Doubleday, New
of running a sufficiently large number of microscopic simula-
York.
tions so as to capture the nature of the variation between in-
Beylich, A. E. (1979). Elements of a Kinetic Theory of Traffic
stances.
Flow, Proceedings of the Eleventh International Sympo-
Some specific instances of quantities for which vari-
sium on Rarefied Gas Dynamics, Commissariat a
ability might be of significant interest are travel times and
L’Energie Atomique, Paris, pp. 129-138.
driving cycles. The latter require knowledge about the statis-
Bui, D. D., P. Nelson and A. Sopasakis (1996). The General-
tical distribution of accelerations, as well as velocities and
ized Bimodal Traffic Stream Model and Two-regime Flow
densities, but such acceleration information is inherent in the
Theory. Transportation and Traffic Theory, Proceedings
distribution function, along with the mechanical and correla-
of the 13th International Symposium on Transportation
tion models that underlie any kinetic model. In fact, kinetic
and Traffic Theory, Pergamon Press (Jean-Baptiste Le-
models seem to be the natural connecting link between contin-
sort, Ed.), Oxford, pp. 679-696.
uum models, which provide the “cross sectional” view of
Carson, T. R. and T. C. Austin, Development of Speed Cor-
traffic that most transportation planning is based on, and the
rection Cycles. Report prepared for the U. S. Environ-
“longitudinal” view that underlies the standard driving cycle
mental Protection Agency, Contract No. 68-C4-0056,
11-9
KINETIC THEORIES
Work Assignment No. 2-01, Sierra Research, Inc., Sac- Nelson, P. (1995a). A Kinetic Model of Vehicular Traffic and
ramento, California, April 30.. its Associated Bimodal Equilibrium Solutions. Transport
Ceder, A. (1976). A Deterministic Flow Model of the Two- Theory and Statistical Physics, Volume 24, pp. 383-409.
regime Approach. Transportation Research Record 567, Nelson, P. (1995b). On Deterministic Developments of Traffic
TRB, NRC, Washington, D.C., pp. 16-30.. Stream Models. Transportation Research B, Volume 29,
Disbro, J. E. and M. Frame (1989). Traffic Flow Theory and pp. 297-302.
Chaotic Behavior. Transportation Research Record 1225, Nelson, P. (2000). Synchronized Flow from Modified
TRB, NRC, Washington, D.C., pp. 109-115. Lighthill-Whitham Model. Physical Review E, Volume
Drake, J. S., J. L. Schofer and A. D. May, Jr. (1967). A Statis- 61, pp. R6052-R6055.
tical Analysis of Speed Density Hypothesis. Highway Re- Nelson, P., D. D. Bui and A. Sopasakis (1997). A Novel Traf-
search Record 154, TRB. NRC, Washington, D.C., pp. fic Stream Model Deriving from a Bimodal Kinetic Equi-
53-87. librium. Preprints of the IFAC/IFIP/IFORS Symposium
Edie, L. C., R. Herman and T. N. Lam (1980). Observed Mul- on Transportation Systems, Technical University of Crete
tilane Speed Distribution and the Kinetic Theory of Ve- (M. Papageorgiou and A. Pouliezos, Eds.), Volume 2, pp.
hicular Traffic. Transportation Research Volume 9, pp. 799-804.
225-235, Nelson, P. and A. Sopasakis (1998). The Prigogine-Herman
Grad, H. (1958). Principles of the Kinetic Theory of Gases. Kinetic Model Predicts Widely Scattered Traffic Flow
Handbuch der Physik, Vol. XII, Springer-Verlag, Berlin. Data at High Concentrations. Transportation Research
Hall, F. L. (1987). An Interpretation of Speed-Flow Concen- B, Volume 32, pp. 589-604.
tration Relationships Using Catastrophe Theory. Trans- Nelson, P. and A. Sopasakis (1999). The Chapman-Enskog
portation Research A, Volume 21, pp. 191-201. Expansion: A Novel Approach to Hierarchical Extension
Helbing, D. (1995). High-fidelity Macroscopic Traffic Equa- of Lighthill-Whitham Models. Transportation and Traffic
tions. Physica A, Volume 219, pp. 391-407. Theory: Proceedings of the 14th International Symposium
Helbing, D., “Gas-Kinetic Derivation of Navier-Stokes-like on Transportation and Traffic Theory, Pergamon (A.
Traffic Equations,” Physical Review E. 53, 2366-2381, Ceder, Ed.), pp. 51-80.
1996. Neunzert, H. and J. Struckmeier (1995). Particle Methods for
Hoogendoorn, S.P., and P.H.L. Bovy (to appear). Non-Local the Boltzmann Equation. Acta Numerica, Volume 4, pp.
Multiclass Gas-Kinetic Modeling of Multilane Traffic 417-457.
Flow. Networks and Spatial Theory. Paveri-Fontana, S. L. (1975). On Boltzmann-like Treatments
Kerner, B. S. and P. Konhäuser (1993). Cluster Effect in Ini- for Traffic Flow: A Critical Review of the Basic Model
tially Homogeneous Traffic Flow. Physical Review E, and an Alternative Proposal for Dilute Traffic Analysis.
Volume 48, pp. R2335-R2338. Transportation Research, Volume 9, pp. 225-235.
Klar, A. and R. Wegener (1999). A Hierarchy of Models for Payne H. J. (1971). Models of Freeway Traffic and Control.
Multilane Vehicular Traffic (Part I: Modeling and Part II: Simulation Council Procs., Simulations Council, Inc., La
Numerical and Stochastic Investigations). SIAM J. Appl. Jolla, CA, Vol. 1, pp. 51-61.
Math, Volume 59, pp. 983-1011. Phillips, W. F. (1977). Kinetic Model for Traffic Flow. Re-
Lighthill, M. J. and G. B. Whitham (1955). On Kinematic port DOT/RSPD/DPB/50-77/17, U. S. Department of
Waves II: A Theory of Traffic Flow on Long Crowded Transportation.
Roads. Proceedings of the Royal Society, Series A, Vol- Phillips, W. F. (1979). A Kinetic Model for Traffic Flow with
ume 229, pp. 317-345. Continuum Implications. Transportation Planning and
Nagel, K. (1996). Particle Hopping Models and Traffic Flow Technology, Volume 5, pp. 131-138.
Theory. Physical Review E, Volume 53, pp. 4655-4672. Prigogine I. and F. C. Andrews (1960). A Boltzmann-like Ap-
Nagel, K. (1998). From Particle Hopping Models to Traffic proach for Traffic Flow. Operations Research, Volume 8,
Flow Theory. Transportation Research Record, Volume pp. 789-797.
1644, pp. 1-9. Prigogine, I. and R. Hermann (1970). Comment on “An
Munjal, P. and J. Pahl (1969). An Analysis of the Boltzmann- Analysis of the Boltzmann-type Statistical Models for
type Statistical Models for Multi-lane Traffic Flow. Multi-lane Traffic Flow.” Transportation Research, Vol-
Transportation Research, Volume 3, pp. 151-163. ume 4, pp. 113-116.
11-10
KINETIC THEORIES
Prigogine, I. and R. Hermann (1971). Kinetic Theory of Ve-

hicular Traffic, Elsevier, New York.
Sopasakis, A. (2000). Developments in the Theory of the
Prigogine-Herman Kinetic Equation of Vehicular Traffic.
Ph.D. dissertation (mathematics), Texas A&M University,
May.
11-11
INDEX
Note: Index has not been updated to reflect revisions to Chapters 2, 5 and new 11.
A B
AASHTO Green Book, 3-21 ballistic, 3-8

acceleration of the lead car, fluctuation in the, 4-8 bifurcation behavior, 5-15
acceleration control, 3-24 block length, average, 6-20, 6-20, 6-22
acceleration noise, 7-8 blockages per hour, 6-22
actuated signals , 9-23 boundary, 5-4, 5-4, 5-10-5-11, 5-23, 5-24, 5-36
adaptive signals, 9-19 brake and carburetion systems, 7-8
adaptive signal control, 9-27 braking inputs, 3-7
aerial photography, 2-3, 6-11 braking performances, 3-20, 4-1
aerodynamic conditions, 7-9 braking performance reaction time, 3-5
aero-dynamic effects, 7-11
age, 3-16
aggregated data , 6-3 C
aggressive driving, 6-20
aging eyes, 3-16 California standards, 7-15
air pollutant levels, 7-15 CALINE-4 dispersion model, 7-15
air pollutants, 7-13, 7-14 capacity, 4-1
air pollution, 7-13 carbon monoxide, 7-13
air quality standards, 7-14, 7-15 car-following,10-2, 10-3, 10-8, 10-15
air quality models, 7-15 car following models, 4-1
air quality, 7-13, 7-15 catastrophe theory, 2-8, 2-27, 2-28
air resistance, 7-12 central city, 6-8
-relationship, 6-12 central vs. peripheral processes, 3-17
alternative fuel, 7-15 changeable message signs, 3-12
alternative fuels, 7-12 changes in cognitive performance, 3-17
altitude, 7-8, 7-8 changes in visual perception, 3-16
ambient temperature, 7-8, 7-8 chase car, 6-21, 6-22
Ambient Air Quality Standards, 7-13 Clean Air Act, 7-13, 7-13
analytical solution, 5-3, 5-3, 5-9 closed-loop braking performance, 3-21
arrival and departure patterns, 5-9, 5-9 coefficient of variation, 3-11
arterials, 5-6, 5-9 cognitive changes, 3-16
Athol, 2-2, 2-10, 2-22 collective flow regime, 6-16
auxiliary electric devices, 7-8 composite emission factors, 7-15
average block length, 6-20, 6-20, 6-22 compressibility, 5-1, 5-9
average cycle length, 6-20 compressible gases, 5-22, 5-22
average flows, 6-8 computer simulation, 6-22, 6-23
average maximum running speed, 6-17 concentration, 2-1, 2-5, 2-8, 2-20, 2-29, 4-15, 6-16,
average number of lanes per street, 6-20, 6-22 6-17, 6-20, 6-23
average road width, 6-6 concentration at maximum flow, 6-25
average signal cycle length, 6-20, 6-23 conditions, 5-4, 5-6-5-9, 5-11, 5-23, 5-27, 5-29-5-30,
average signal spacing, 6-10 5-32, 5-36, 5-38, 5-43, 5-45
average space headway, 5-6, 5-6 confidence intervals, 10-17, 10-17, 10-20, 10-21,
average speed,6-3, 6-6, 6-8, 6-10, 6-11, 6-17, 6-22, 10-26
7-9, 7-11 congested operations, 2-11, 2-22
average speed limit, 6-20 continuity equation, 5-1-5-3, 5-20, 5-22, 5-24, 5-25
average street width, 6-10, 6-11 continuous simulation models, 10-3
continuum models, 5-1-5-1, 5-3, 5-20, 5-29, 5-41
control, 3-1, 4-2
control movement time, 3-7, 3-7
control strategies, 6-22
12-1
convection motion and relaxation, 5-20 entrance or exit ramps, 5-12
convection term, 5-20, 5-22 equilibrium, 5-1-5-1, 5-3, 5-10, 5-22-5-23, 5-45
convergence, 5-11 ergodic, 6-17
coordinate transformation method., 9-11 evasive maneuvers, 3-15
correlation methods, 10-22, 10-22 expectancies, 3-7
critical gap values for unsignalized exposure time, 3-13
intersections, 3-26
cruising, 7-11
cruising speed, 7-11
F
figure/ground discrimination, 3-17, 3-17

D
filtering effect on signal performance, 9-17
first and second moments, 5-22
deceleration-acceleration cycle, 7-11, 7-11
Fitts' Law, 3-7, 3-7
decision making, 4-2
fixed-time signals , 9-23
defensive driving, 3-17-3-17
floating car procedure, 2-3
delay models at isolated signals, 9-2
floating vehicles, 6-11
delay per intersection, 6-10
flow, 1-4, 1-4, 2-1, 2-7, 2-10, 2-16, 2-18, 2-24, 2-26,
density, 1-4-1-4, 2-1- 2-3, 2-7, 2-11, 2-18, 2-21, 2-28
2-32, 2-34, 4-1
density and speed, 2-3, 2-22
flow-concentration relationship, 4-15
disabled drivers, 3-2
flow rates, 2-2, 2-4-2-5, 2-14, 2-32, 5-3, 5-6, 5-10,
discontinuity, 5-1, 5-4
5-12, 5-13, 5-19, 5-24, 5-25
discrete simulation models, 10-3
forced pacing under highway conditions, 3-17
discretization, 5-10, 5-10-5-14, 5-26, 5-30, 5-34
fraction of approaches with signal progression, 6-20
display for the driver, 3-2
fraction of curb miles with parking, 6-20, 6-23
dissipation times, 5-8
fraction of one-way streets, 6-20, 6-20, 6-22
distractors on/near roadway, 3-28
fraction of signalized approaches in progression, 6-23
disturbance, 4-15
fraction of signals actuated, 6-20
Drake et al., 2-7, 2-12, 2-20, 2-23, 2-24, 2-28, 2-36
fraction of vehicles stopped, 6-17, 6-23
driver as system manager, 3-2
free-flow speed, 6-9
driver characteristics, 7-8
fuel consumption, 6-23, 7-8, 7-9, 7-12
driver performance characteristics, 3-28
fuel consumption models, 7-8
driver response or lag to changing traffic signals, 3-9
fuel consumption rate, 7-8, 7-9, 7-11
drivers age, 3-16
fuel efficiency, 7-8, 7-9, 7-12
driving task, 3-9, 3-28
fundamental equation, 2-8, 2-10
drugs, 3-17-3-17
G
E
gap acceptance, 3-10, 3-25
Edie, 2-6, 2-18, 2-32, 2-34
gasoline type, 7-8
effective green interval, 5-6, 5-8
gasoline volatility, 7-15
effective red interval, 5-8
Gaussian diffusion equation, 7-15
electrification, 7-15
gender, 3-16, 3-16
Elemental Model, 7-9, 7-11
glare recovery, 3-17
emission control, 7-14
“good driving" rules, 4-1
emissions, 7-13
grades, 7-8
energy consumption, 7-8, 7-12
Greenberg, 2-20, 2-20, 2-21, 2-34
energy savings, 7-8
Greenshields, 2-18, 2-18, 2-34
engine size, 7-8, 7-12
guidance, 3-1
engine temperature, 7-8
12-2
H lead (Pb), 7-13
legibility , 3-9
headways, 2-2, 2-2, 2-3, 2-8 levels of service, 6-2
Hick-Hyman Law, 3-3 light losses and scattering in optic train, 3-16
high order models, 5-1-5-1, 5-15 local acceleration, 5-20, 5-26
Highway Capacity Manual, 4-1 log-normal probability density function, 3-5
highway driving, 7-11 looming, 3-13
Human Error, 3-1 loss of visual acuity, 3-16
humidity, 7-15
hysteresis phenomena, 5-15
M
I macroscopic, 6-1
macroscopic measure, 6-16
identification, 3-9, 3-15 macroscopic models, 6-6
idle flow rate, 7-12 macroscopic relations, 6-25
idling, 7-11, 7-11 macular vision, 3-17
Index of Difficulty, 3-8 maximum average speed, 6-3
individual differences in driver performance, 3-16 maximum flow, 6-11
infinitesimal disturbances, 4-15 May, 2-2-2-7, 2-9, 2-12, 2-22, 2-24, 2-33, 2-36
information filtering mechanisms, 3-17 measurements along a length of road, 2-3
information processor, 3-2 Measures of Effectiveness, 10-17, 10-17, 10-25
initial and boundary conditions, 5-5, 5-5, 5-6, 5-11 medical conditions, 3-18
inner zone, 6-10 merging, 3-25
inspection and maintenance, 7-15 meteorological data, 7-15
instantaneous speeds, 7-12 methanol, 7-15
interaction time lag, 5-12, 5-12, 5-13 microscopic, 6-22
intersection capacity, 6-11 microscopic analyses , 6-1
intersection density, 6-20 minimum fraction of vehicles stopped, 6-25
intersections per square mile, 6-20 minimum trip time per unit distance, 6-17, 6-17 mixing
intersection sight distance, 3-10, 3-27 zone, 7-16
Intelligent Transportation Systems (ITS), method of characteristics, 5-4
2-1-2-2, 2-5, 2-6, 2-8, 2-19-2-20, 2-24, model validation, 10-5
2-32-2-33, 3-1, 6-25 model verification, 10-5, 10-15
momentum equation, 5-1-5-1, 5-22, 5-26, 5-29
motion detection in peripheral vision, 3-14
J movement time, 3-7
moving observer method, 2-3, 2-3
jam concentration, 4-14 MULTSIM, 7-12
jam density, 5-3, 5-8, 5-11-5-14
N
K
navigation, 3-1
kinetic theory of traffic flow, 6-16 NETSIM , 6-22, 6-23
network capacity, 6-6
network topology, 6-1
L network concentrations, 6-22, 6-24
network features, 6-20, 6-20
network-level relationships, 6-23
lane-changing, 10-5
network-level variables, 6-25
lane miles per square mile, 6-20
12-3
network model, 6-1, 6-6 period of measurement, 7-15
network performance, 6-1 "Plain Old Driving" (POD), 3-1
network types, 6-6 platoon dynamics, 5-6
network-wide average speed, 6-8 platooning effect on signal performance, 9-15
nighttime static visual acuity, 3-11 pollutant dispersion, 7-16
nitrogen dioxide, 7-13 Positive Guidance, 3-28
non-instantaneous adaption, 5-23 positive kinetic energy, 7-11
non-linear models, 4-15 pupil, 3-16
normal or gaussian distribution, 3-5
normalized concentration, 4-15
normalized flow, 4-15
Q
number of lanes per street, 6-20
number of stops, 6-23, 7-8
quality of service, 6-20, 6-25
numerical solution, 5-9, 5-11, 5-12, 5-29, 5-31-5-33,
quality of traffic service, 6-12, 6-16
5-49
queue, 5-4, 5-7, 5-50
queue discharge flow, 2-12, 2-13, 2-15
queue length, 5-6, 5-9
O queue length stability, 5-8
object detection, 3-15

obstacle and hazard detection, 3-15
R
obstacle and hazard recognition, 3-15
obstacle and hazard identification , 3-15
radial motion, 3-13
occupancy, 1-4, 2-1, 2-9, 2-11, 2-21, 2-22,
random numbers, 10-2-10-2, 10-22, 10-26
2-25-2-26, 2-28, 2-32, 2-34, 2-36
reaction time, 3-3, 3-3, 3-4, 3-7, 3-8, 3-16, 3-17
off-peak conditions, 6-6
real-time driver information input, 3-28
Ohno's algorithm, 9-8
refueling emissions controls, 7-15
oil viscosity, 7-8
relaxation term, 5-23
oncoming collision, 3-13
resolving power, 3-11
open-loop, 3-8
response distances and times to traffic control
open-loop braking performance, 3-20
devices, 3-9
oscillatory solutions, 5-15
response time, 3-4, 3-7, 3-15, 3-16, 3-20
outer zone, 6-10
response to other vehicle dynamics, 3-13
overtaking and passing in the traffic stream, 3-24
road density, 6-15
overtaking and passing vehicles, 3-24
roadway gradient, 7-8
overtaking and passing vehicles (Opposing Traffic),
rolling friction, 7-9
3-25
rolling resistance, 7-12
oxygenated fuels/reformulated gasoline, 7-15
running (moving) time, 6-17
ozone, 7-13
running speed, 6-10, 6-10, 6-11
P S
partial differential equation, 5-4, 5-30
saturation flow, 6-10
particulate matter, 7-13
scatter in the optic train, 3-17
pavement roughness, 7-8
scattering effect of, 3-17
pavement type, 7-8
senile myosis, 3-16
peak conditions, 6-6
sensitivity coefficient , 4-15, 5-12, 5-12
perception-response time, 3-3
shock waves, 5-1, 5-1, 5-3-5-4, 5-6, 5-29, 5-30, 5-50
peripheral vs. central processes, 3-17
signalized intersection, 5-6, 5-6, 5-7
perception, 4-2
signalized links and platoon behavior, 5-9
12-4
short-term events, 6-22 stimulus-response equation, 4-3
signals, stochastic process, 10-17
actuated, 9-23 stochastic simulation, 10-5
adaptive, 9-19 stop time, 6-17, 6-17
signal control,adaptive, 9-27 stopped time, 6-10
signal densities, 6-10 stopped delay, 7-11
signal density, 6-20 stopping maneuvers, 3-15
signals per intersection, 6-22 stopping sight distance, 3-26
sign visibility and legibility, 3-11 stop-start waves, 5-15-5-15, 5-17, 5-24, 5-26, 5-36,
signage or delineation, 3-17 5-39
simulation models,building 10-5 street network, 6-20
site types, 7-15 structure chart, 10-8
smog, 7-13 substantial acceleration, 5-20, 5-20
Snellen eye chart, 3-11 sulfur dioxide, 7-13
sound velocity, 5-22 summer exodus to holiday resorts, 5-17
source emissions, 7-14 surface conditions, 7-8
space headway, 2-1, 2-5 suspended particulate , 7-13
space mean speed, 2-6-2-7, 2-9-2-10, 6-15, 7-11
spacing, 2-1, 2-1, 2-26, 4-8, 5-2, 5-17, 5-29, 5-34
specific maneuvers at the guidance level, 3-24 T
speed, 2-3, 2-6, 2-8, 2-11, 2-14, 2-16, 2-18, 2-22,
2-24, 2-28, 2-31, 2-33, 4-1, 4-15 tail end, 5-6-5-8
speed (miles/hour) versus vehicle concentration temperature, 7-15
(vehicles/mi), 4-17 time-dependent delay models, 9-10
speed and acceleration performance, 3-24 time headway, 2-1
speed-concentration relation, 4-13 time mean speed, 2-6-2-7
speed-density models, 2-19 tire pressure, 7-8
speed-density relation, 5-15-5-15, 5-20, 5-22-5-23, tire type, 7-8
5-27, 5-34 total delay, 6-23
speed-flow models,2-13, 2-1,9 6-8 total trip time, 6-17
speed-flow relation, 6-6 TRAF-NETSIM , 6-22, 6-23
speeds from flow and occupancy, 2-8, 2-9 traffic breakdowns, 5-15, 5-42
speed limit changes, 3-28 traffic conditions, 7-9
speed noise, 7-8, 7-12 traffic control devices (TCD), 3-9
speed of the shock wave, 5-4 traffic control system, 6-1
speed-spacing, 4-15 traffic data, 7-15
speed-spacing relation, 4-1 traffic dynamic pressure, 5-23
spillbacks, 5-9 traffic intensity , 6-2, 6-15
stability analysis, 5-8, 5-25, 5-28-5-29, 5-43 traffic network, 6-1
standard deviation of the vehicular speed distribution, traffic performance, 6-1
5-22, 5-39 traffic signal change, 3-9
state equations, 5-9, 5-9 traffic simulation, 10-1-10-2, 10-4, 10-7, 10-15-
State Implementation Plans (SIPs), 7-15 10-17, 10-20, 10-22
stationary sources, 7-13 traffic stream, 4-1
statistical distributions, 10-5, 10-6 trajectories of vehicles, 5-4
steady-state, 7-11 trajectory, 5-4, 5-7-5-9
steady-state delay models, 9-3 transients, 5-15-5-15, 5-17, 5-20
steady-state expected deceleration, percentile estimates transmission type, 7-8
of , 3-21 travel demand levels, 6-1
steady-state flow, 4-15 travel time , 6-1, 6-10
steady-state traffic speed control, 3-24 trip time per unit distance, 7-9
steering response times, 3-9, 3-9 two-fluid model , 6-1, 6-17, 6-22-6-23, 6-25
12-5
two-fluid parameters, 6-18, 6-18, 6-20, 6-23, 6-25 vehicle miles traveled, 6-11
two-fluid studies, 6-20 vehicle shape, 7-8
two-fluid theory, 6-12, 6-16, 6-24 vehicles stopped ,average fraction of the, 6-17
turning lanes, 5-9, 5-9 viscosity, 5-22, 5-24, 5-29, 5-34
visual acuity, 3-11
visual angle, 3-11-3-13, 3-15, 3-16
U visual performance, 3-11
volatile organic compounds, 7-13
undersaturation, 5-8
effect of upstream signals, 9-15
UMTA, 7-15 W
UMTA Model, 7-15
urban driving cycle, 7-11 Wardrop, 2-4, 2-4, 2-6-2-7
urban roadway section, 7-11 Wardrop and Charlesworth, 2-4, 2-4
uncongested flows, 2-12 Weber fraction, 3-13, 3-13
wheel alignment, 7-8
wind, 7-8
V wind conditions, 7-8
wind speed, 7-15
variability among people, 3-16 work zone traffic control devices, 3-17
vehicle ahead, 3-13
vehicle alongside, 3-14
vehicle characteristics, 7-8 Y
vehicle emissions, 7-14
vehicle fleet, 7-8 yield control for secondary roadway, 3-27
vehicle mass, 7-8, 7-9, 7-12
12-6

Traffic Flow Theory 2001

Uploaded by

Copyright:

Available Formats

Traffic Flow Theory 2001

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Traffic Flow Theory 2001

Uploaded by

Copyright:

Available Formats

Traffic Flow Theory

2. TRAFFIC STREAM CHARACTERISTICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1

3. HUMAN FACTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1

4. CAR FOLLOWING MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-1

5. CONTINUUM FLOW MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-1

6. MACROSCOPIC FLOW MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-1

7. TRAFFIC IMPACT MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1

8. UNSIGNALIZED INTERSECTION THEORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1

9. TRAFFIC FLOW AT SIGNALIZED INTERSECTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-1

10. TRAFFIC SIMULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-1

11. KINETIC THEORIES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-1

2. TRAFFIC STREAM CHARACTERISTICS

4. CAR FOLLOWING MODELS

5. CONTINUUM FLOW MODELS

7. TRAFFIC IMPACT MODELS

8. UNSIGNALIZED INTERSECTION THEORY

9. TRAFFIC FLOW AT SIGNALIZED INTERSECTIONS

10. TRAFFIC SIMULATION

11. KINETIC THEORIES

4. CAR FOLLOWING MODELS

8. UNSIGNALIZED INTERSECTION THEORY

9. TRAFFIC FLOW AT SIGNALIZED INTERSECTIONS

10. TRAFFIC SIMULATION

11. KINETIC THEORIES

L length of vehicles of uniform length

ck constant of proportionality between occupancy and

ki the (average) density of vehicles in substream I

qi the average rate of flow of vehicles in substream I

Å average speed of a set of vehicles

A A(x,t) the cumulative vehicle arrival function over

kj jam density, i.e. the density when traffic is so heavy that

uf free-flow speed, i.e. the speed when there are no

Fourth draft of revised chapter 2

Chapter 2: Traffic Stream Characteristics

2.1 Definitions and terms

2.1.1 The Time-Space Diagram.

Figure 2-1. Time-space diagram.

The characteristic geometries of trajectories on a time-space diagram describe the motion of

For observation intervals containing large m,

i.e., flow is the reciprocal of the average headway.

If the L contains large n,

Figure 2-2. Trajectories in time-space region.

mean velocity becomes

U( A) area of the shaded strips | A |

Figure 2-3. Trajectories of vehicle fronts and rears.

2.1.6 Three-Dimensional Representation of Vehicle Streams. It is useful to display flows

Figure 2-4. Three-dimensional representation.

2.2 Measurement issues

2.2.3 Importance of location to the nature of the data

the data observable there will be as shown in Figure 2.5.

2.2.4 Selecting intervals from which to extract data

2.3 Bivariate models

2.3.1 Speed-flow models

It is useful to summarize some of the antecedents of the understanding depicted in Figure

V = (A - eBQ) e-c - K edQ (2.25)

It is instructive to review the history of depictions of speed-flow curves in light of this

2.3.2 Speed-concentration models

2.3.3 Flow-concentration models

Early studies of highway capacity followed two principal approaches. Some

2.4 Three-dimensional models

= parameter of log normal distribution ~ standard deviation

0.17 (no unit) -0.63 (no unit)