Transport

Plan
Plan de Systems
de Formación
Formación (40h)
(40h)

Lecture 4. Traffic Theory

Fundamental variables
TRAFFIC OPERATIONS
Introduction:
• Study of traffic theory is fundamental in the planning, design and
operation processes of transport systems.
• Traffic Flow characteristics: headway, flow, speed, spacing and density.
• Analytical tools:
– Supply and demand modeling
– Analysis of capacity and level of service
– Traffic flow modeling
– Shock wave and queueing analysis

Demand. Mobility
Analysis and
Infrastructure resources. Capacity Traffic behaviour
characterisation
Traffic control

1
 Variables fundamentales (1)
• Definitions:
Smoothly monotonous approximation of N(t,x)
N

N (t , x) N (t , to , x)
Count section
Initial time of count
t
Volum, count, traffic, intensity, IMD, demand
FLOW def
N (t , x)
q (t , x) 
t  to
def  x ( A)
j
j

Flow q ( A) 
x A
A

xj def  t ( A) j
Density k ( A)  j

A
t
tj
def x ( A)j
j
Average speed v( A) 
 t ( A) j
j
3

Macro and Micro approach

Approach Variables
MACRO flow density average speed Group of vehicles as a fluid.
Flow conservation.
Equation of state.
Shock waves.
MICRO headway spacing instantaneous speed Individual vehicles as particulates.
Kinematics variables of each
vehicle and its interaction with the
vehicles around (leader and
follower).
Car following models.

2
 TRAFFIC CHARACTERISTICS
SOFTWARE

Headways
Headways between vehicles have an impact on:
• Safety (mínimum clearance distance)
• Level of service (LoS)
• Driver behaviour
• Transport system capacity
x

4 3 2 1 t1

4 3 2 t2

4 3 t3

4 t4

(h)1-2=t2-t1

(h)2-3=t3-t2

3
 Headway distribution
Based on headway counts at a
specific location for different flow
conditions, we have different
headway distributions:
q [10, 14]

q [15, 19]

q [20, 24]

LoS
q [25, 29]

Type of flow Low flow Medium flow High flow

Characteristics Vehicles travel Some vehicles travel independently Car-following. Constant headways,
independently and others travel toguether unless drivers make mistakes
Distribution Random With parameters from a random Constant
distribution to constant distribtuion
7

Headways – Random distribution

Random traffic states where headways have random distributions are
represented by negative exponential distributions derived from Poisson arrival
distributions where:
1. any time has the same probability of having the arrival of one vehicle
2. the arrival of one vehicle at a specific time does not impact on the
probability of the arrival of another vehicle.

P(x): probabilidad of x vehicles arrive during time interval t

Px  
lt x e  lt l: average value of arrivals per unit of time during a time interval t
x!
t: chosen time interval

If h is the time interval under study, we have: Ph  t   e  lt

Pt  h  t  t   Ph  t   Ph  t  t 

F t  h  t  t   N [ Ph  t   Ph  t  t ]
N: Total number of analysed time intervals:
Ft  h  t  Δt  :number of forecasted headways in the time interval group
t  h  t  Δt  8

4
 Headways – Normal distribution

For states with short headways, the distribution that represents

better these states is the normal distribution.

s: standard deviation of headway distribution

t a
s t: average headway (seg/veh)
2
a: mínimum headway

a  t  2s Condition of non-negative headways

Comparison theoretical vs measured distributions

Random distribution Normal distribution

• Better adjustment for low flow conditions

• Normal dist. is symmetric with a bell shape
• Bad adjust. for headways below 1 sec., higher values • Bad adjust. for low flows and good for high flows
• Lower values for headways between 1.5 and 4.5 sec. • Adjustment displaced 0.5-1 sec.
• St. dev. is always higher for random distribution • St. dev. is always lower for normal distribution10

5
 Headways - Intermediate state
1 – Generalized model Headways – Pearson type III distribution.

l
f (t )  l (t  a )k 1 e l (t a )
( k )

𝛼 ≥ 0 ; 0 ≤ k ≤ +∞ ; Γ(k) = (k−1)!
K: shape parameter
• Gamma (a=0)
• Erlang (K integer number)
k

4
Gamma

3
Erlang

2
Negative exponential
Shifted negative exponential
1

0,5 1 a 11

Headways - Intermediate state

1 – Generalized model Headways – Pearson type III distribution.

• For the first group of positive headways, the

probability of this group is a half of the probability
of the next group of headways.
• Pearson III distribution and measured distribution
are qualitatively the same.
• From headways longer than 4 sec and shorter than
1 sec, probabilities are lower for the Pearson
distribution than for the measured distribution.
• There are higher inconsistencies for headways
between 1 and 4 sec.
• Results can vary with changes in the used
parameters.

12

6
 Flow variations
Hourly variation
10%

Flow per direction

Heavy veh’s reduce

People flow infrast. capacity, there is
an equivalent number
of light veh’s.
Vehicular flow

13

Flow distributions
Flow conditions Distribution
Low flow Poisson distribution
Intermediate flow Too complex and not work well
Flow close to capacity Uniform distribution

Poisson distribution: For random traffic states

m x  em
P( x) 
x! where:
P(x) = probability of exactly x vehicles arrive in the
interval of time t
m = mean of vehicles that arrive in the interval t
x = # veh arriving in an interval of time
t = chosen interval of time

Uniform distribution:
Gives the same unique value for all headways during the
counting interval. For instance, during one hour the
number of vehicles counted for each minute is the same.
At capacity of 1800 veh/h, 30 veh/min. Then, the variance
tends to zero. 14

7
 Speed
Speed distribution depends on lane and follow a normal or lognormal distribution
DISTRIBUCIÓN DE VELOCIDADES
A-7 PK.11 (BARNA- LA JONQUERA)
DIR. GIRONA (CARRIL 1-2-3) Right lane
% Central lane
PORCENTAJE
100 Left lane
90

80

70

60 • Sonic weapon / radar

50
• Magnetic induction loops
40

30 • Videos (times)
20

10
• Surveys (bias)
0
40 50 60 70 80 90 100 110 120 130 140 150 160 170

SPEED
VELOCIDAD

15

Speed

• Normal probability paper • Logarithm probability paper

16

8
 Speed, microscopic
Effects of intersections and overtakings

Effects of ramps and slopes on the speed of trucks

v
7% -3% -2% 0%
5%
0% 3%

Lose of speed depends on vehicle

P1 power  Slow lanes
P2
P3
d 17

Speed, macroscopic
Speed maps

5:30
5:15
5:00
4:45
4:30
4:15 >70 km/h
4:00
50-70 km/h
S-1 S-2 S-3 S-4 S-5 S-6 S-7

<50 km/h
5:30
5:15
5:00
4:45
4:30
4:15
4:00
S-1 S-2 S-3 S-4 S-5 S-6 S-7

18

9
 Speed, macroscopic
Speed maps

19

Speed, macroscopic
Tools for the analysis of travel times and delays
Isochrones: Locus (“lugar geométrico”) of points located at the same travel
time from an origin of the trip

20

10
 MACROSCOPIC ANALYSIS
SOFTWARE

21

Fundamental Traffic Equation

x
Stationary:
- Constant speed
- Constant headway
L n, # vehicles - Constant spacing
Identical values of variables at any point of
the diagram (x,t)
t
T

Stationary traffic:
Fundamental equation:

q  k  vl (kl /k )  vs k l  1,2
l 22

11
 Conservation equation

• Vehicles are conserved, they cannot disappear.

• Conservation law equivalent to other conservation laws

for mass, energy, etc.

• In a closed system, number of vehicles entering -

number of vehicles exiting = number of vehicles stored.

23

Conservation equation

With generation and dissipation of vehicles g(x,t):

24

12
 Relative flow

25

Traffic diagrams
• 3 fundamental variables (q, k, v), 3 degrees of freedom
• Traffic state in a univocal way, from 3 dof to 1 dof
• We need 2 equations:
• Fundamental equation: q=kv (mathematical result).
• Equation of state: empirical relationship between
fundamental variables.
• Any relationship is only valid for one infrastructure and
drivers/vehicles observed.
u
DATOS
DATA

k
26

13
 Greenshields

• Relationship between 2 variables from observation.

• Density and speed.
• Photographic measurement, 6 points.
• Linear relationship between v-k.

27

Greenshields
u • uf free flow speed, maximum average
k
u  u f (1  )
uf kj speed for safe and comfortable driving,
for low densities. 100-120 km/h

• kj jam density, maximum density when

vehicles are in a jam stopped, 125-150
kj k
veh/km

• qmax maximum flow or capacity, when optimal speed and

density (uo, ko), qmax =2000-2200 veh/h, uo=70-90 km/h,
ko=20-25 veh/km.

• Two parts, free-flowing and congested

• Limitations: uo = uf/2, significantly higher; ko = kj/2, much

smaller.
28

14
 Greenberg
u
uf 𝑢 = 𝑢 ln
𝑘 • uo or um optimal speed, difficult to
𝑘 calibrate

• kj jam density, maximum density.

kj k

• This approximation describes well congetsed traffic

conditions, there is a maximum density, jam density.

• However, this approximation does not describe well light

traffic conditions, for low densities, speed tends to infinite.

• Solution: Truncate the model at the free flow speed.

29

Underwood
u

uf • uf free flow speed

𝑢 =𝑢 ·𝑒

• ko optimal density, difficult to

determine
kj k

• Even when traffic density is the jam value, the speed is not
zero. We can accept that speed is not always zero, vehicles
stop and go even in jam conditions.

30

15
 Edie
u
• uf free flow speed
uf
• kj jam density

• uo optimal speed
ko kj k • ko optimal density
UNDERWOOD GREENBERG

• Two-regime model: Underwood for free-Flow conditions and

Greenberg for congested traffic.

31

Derived diagrams

• p (pace) = 1/v, travel time per unit distance

32

16
 Diagrams in a highway

33

Factors that influence the FD

Infrastructure User and vehicle type

u Highway 2 lanes/direction u
Fhv:0%
Roadway 1+1 Fhv:20%

commuters

q others
q

34

17
 LWR model – Kinematic/shock wave theory
• Lighthill‐Whitham‐Richards (LWR) model
• Objective is to predict the evolution of traffic variables (traffic state)
in time and space… due to flow restriction or moving obstruction.
• Interesting for queued traffic conditions: queue length, queue
growing speed, queue dissipation.
• Assumptions: conservation of vehicles, equation of state (traffic
diagram)

35

Transitions between traffic states

• Acceleration/deceleration

• Traffic wave propagation

• Trajectory of information

• Extension/dissipation of
queue

Shock wave is a discontinuity of

traffic states (flow, density and
speed). Vehicles located at the
shock wave change their speeds
instantaneously.

36

18
Shock waves – Aerial pictures

37

Shock wave speed

38

19
 Wave speed on the fundamental diagram

39

Traffic states due to traffic light and (x, t) diagram

Traffic light with an arrival of vehicles at a constant flow qA

Fundamental diagram q-k

Space-time
diagram

40

20
 Traffic states due to bottleneck and (x, t) diagram

Bottleneck at x=xc

41

Traffic states due to flow variations in a bottleneck

Density map
5:30
5:15 <40 veh/km
5:00
4:45 40-60veh/km
4:30 >60 veh/km
4:15
4:00

S-1 S-2 S-3 S-4 S-5 S-6 S-7

Shock wave
s
Frontal stationary

High
Bottleneck Forward
Backward density recovery
recovery

Rear Flow in number of lanes

stationary
2,0 1,5
t equivalent to capacity
q q: 1,5 2,5

42

21
Classification of shock waves

Shock waves
Frontal
s Forming stationary Recovery

Forward Backward
forming recovery

High density

Backward Forward
forming recovery

Rear
stationary

43

Density maps

44

22
 Fundamental diagram

45

Relative flow for observer

46

23
 Shock waves - acceleration

• All intermediate speeds

• Faster transition for the first

vehicle

47

Shock waves - deceleration

• Slower transition for the first

vehicle
• Progressively, the vehicles
skip states, shocks

48

24
 Shock waves - deceleration

49

Simplifications

• Instantaneous
acceleration/deceleration

• Triangular fundamental
diagram
• Congestion: constant wave
speed w, characteristic
wave speed
• Free-flow: constant wave
vf, free-flow speed
50

25
 Simplifications

• Congested scenario

• Free-flow scenario

51

MICROSCOPIC
SOFTWAREANALYSIS

52

26
 Car-following models
Trajectory between two vehicles n and n+1. s n

SPACING: d n 1 (t )  Ln  g n 1 (t ) g n 1 n+1
Ln
HEADWAY: d n 1 (t )  hn 1 x n 1
1 N
hn 1
DENSITY: k
1 d
N
d n
xn 1
d n 1
x n
t
Car following theory: •Pipes (1953): Californian Motor Vehicle Code. Keep a
length of a car for each 10 mph of speed that vehicle
runs.
Follower Leader •
𝑑 1 1
𝑥
xn1 (t ) xn (t ) 𝑑 = 𝐿 +𝐿 ℎ = •
𝑥
=𝐿 +
10 𝑥•
10
xn 1 (t  t )
2 1
•Forbes (1958,63,68) introduced reaction time. t=1,5”
xn1 (t ) xn (t )  xn 1 (t )

xn (t ) ℎ = Δ𝑡 +
𝐿 𝑑 =ℎ
• •
𝑥 (𝑡) = Δ𝑡𝑥 (𝑡) + 𝐿
•
𝑥
53

Car-following models
Response=f (Sensitivity, Stimuli)

GENERAL MOTORS ••
𝑥 (𝑡 + Δ𝑡) = 𝛼 𝑥 (𝑡) − 𝑥
• •
(𝑡)

• 1st Generation GM: MIN MAX MEAN

Reaction time t 1” 2,2” 1,6”

Sensitivity (s-1) a 0,17 0,74 0,37

• 2nd Generation GM: a a1 as vehicles are far

a2 as vehicles are close
• 3rd Generation GM: •• 𝛼 • •
𝑥 (𝑡 + Δ𝑡) = 𝑥 (𝑡) − 𝑥 (𝑡)
𝑥 (𝑡) − 𝑥 (𝑡)

a o   VELOCIDAD   L 
T  t a0
Connection between this micro model to
Circuit GM 1,5” 44 km/h
Greenberg’s model (Macro)
Tunnel in Holland 1,4” 29 km/h
Tunnel Lincoln 1,2” 32 km/h
54

27
 Car-following models
• 4th Generation GM: 
 a ' x n 1 (t )  

x n 1 (t  t )   x n (t )  x n 1 (t )
  
 x n (t )  x n 1 (t )
Sensitivity parameter is an
adimensional constant a’ and  
the speed of the follower.

a l ,m xn 1 (t  t )m
• 5th Generation GM: xn 1 (t  t )  xn (t )  xn 1 (t )
xn (t )  xn 1 (t )l

1 4ª

1ª
3ª l
2ª
1 2
55

Car-following trajectories

Diagramas de seguimiento entre vehículos

56

28
 Connection between micro and macro models

Microscopic density models are consistent with the fundamental traffic equation (macro)

GREENBERG MODEL:

u = u0 log ( kj / k ) space-mean speed

u0: optimal speed (qmax)
u0  a0 ¿suspicious?

a0
3rd GENERATION GM: xn1 (t  t )  x (t )  x (t )
x (t )  x
n n 1
(t )
n n 1

Integration with regard to t: 𝑥̇ = 𝛼 log 𝑥 − 𝑥 +𝐶

1 𝐶
k=1/d 𝑢 = 𝛼 log + 𝐶 = 𝛼 log 𝑤𝑖𝑡ℎ 𝐶 = 𝑒
𝑘 𝑘

57

Connection micro and macro

 C 
k=kj u=0 log  2   0 C  k
 k  2 j
 j 

 kj 
u  a 0 log 
 k 

Comparing with the GREENBERG macro model:

u0 = a0

Optimal speed (maximum flow)

58

29
 Traffic stability

C LOCAL ASYMPTOTIC
stability Stability
0 Nonoscillatory Damped oscillatory
C= a T INSTABILITY
(0,37) (0,50)
Values higher than C, traffic more instable: 0,5 Damped Increased oscillatory
• High T: slow reaction 1 oscillatory

• High a: Very sensitive response 1,5

(1,57)
2 Increased
oscillatory
STABILITY:
LOCAL: interaction between only 2 vehicles.
ASYMPTOTIC: interaction of a row of infinite
vehicles.
C=1,6
C=0,5

59

Traffic stability

Propagation of effects of speed change causes the crash

60

30
 MACROSCOPIC APPLICATIONS
SOFTWARE

61

Fundamental diagrams
u
DESIGN A

uc C

q
CONTROL q C

A B

k kc kj k
THEORY
B
kc A

uc u 62

31
 Traffic flow rate
Traffic flow rate (or volume): number of vehicles that cross a
fixed point during a given time.
ADV (IMD): Average daily volume (Intensidad media diaria)
HV (IH): Hourly volume (Intensidad horaria)
It is related to time headway between vehicles: 3600
where: q60 hourly flow rate; h average time headways (sec.) q 60 
h
Service flow:: Maximum flow (qmax) for a given Level of Service.
Capacity: Maximum flow (qmax) under prevailed conditions of the roadway, traffic
and control (for high-capcity highways: qmax=2200 veh/hour-lane).
Saturation flow:: Maximum flow (qmax) assuming a green phase in the traffic lights
(qmax=1800 veh/h of green-lane). Examples:
• Ramblas: 1.200 veh/h of green-lane
• Córcega: 1.500 veh-h de green-lane
Variation of q: Monthly/seasonal. ADV (IMD) in April-May/Oct-Nov; Daily

63

Estimating hourly flow rates

The estimation of hourly flows is needed for a correct planning and design of
new infrastructures and improvement of existing ones.
Annual average daily traffic (AADT) cannot be used for these studies due to the
great variations of traffic flow patterns.
We have to use the directional design hourly volumen (DDHV), the volume only
in one direction, calculated from hourly volume in both directions and average
daily traffic.

 DDHV   DHV 
DDHV      AADT
 DHV   AADT 
Where:
DDHV = design hour volume in major direction
DHV = design hour volume combining both directions
AADT = annual average daily traffic combining both
directions
DDHV  D  K  AADT
Where: D = ratio DDHV and DHV
K = ratio DHV and AADT 64

32
 Estimating hourly flow rates

Values for adjustment

factors K and D

Influence of flow variations

on the design flow

65

Capacity or Saturation flow rate

Capacity (C): Maximum flow for passengers or vehicles that pass a point
or section during a time Interval, depending on roadway, traffic and
contorl conditions.
However, capacity only represents one of the possible traffic states.
Grading of Levels of Service A, B, C, D, E and F
u A B
C
D
Continuous flow: E

Capacity is a measure of maximum

productivy, but it does not take into
account quality and level of service for q
users  Service flow rate is associated
to LoS. Cj Flow for level of service j
N Number of lanes

IS  C j  N  f w  f HV  f d fw Width and obstacles factor

f HV Heavy vehicles factor
fd Type of driver

66

33
 Capacity or Saturation flow rate

67

Capacity or Saturation flow rate

• Traffic light (discontinuous)

IS  1800  N  f w  f HV  f g  f p  f pb  f bb  f a  f RT  f LT

Slope factor Bus blocking factor

fg f bb

fp Parking adjustment factor fa Area type factor

Parking blockage factor f RT Right and left turn factors

f pb f LT

68

34
 Capacity or Saturation flow rate

Example of spatial diffusion of flow at a signalized intersection

69

Transport
Plan
Plan de Systems
de Formación
Formación (40h)
(40h)

Lecture 4. Traffic Theory

35