NATIONAL UNIVERSITY OF SINGAPORE

DEPARTMENT OF CIVIL & ENVIRONMENTAL ENGINEERING

CE 3121 TRANSPORTATION ENGINEERING

Semester 1, Academic Year 2022/2023

Part I: Traffic Engineering and Transport Planning

Lecture Notes

Lecturer: Dr. Liu Yang

Traffic Engineering and Transportation Planning 2

SIGNALIZED INTERSECTIONS
Recommended readings
1. Mannering, F. L., Washburn, S. S. and Kilareski, W. P. (2013). Chapter 7. Principles of
Highway Engineering and Traffic Analysis, 5th Edition. Wiley.
2. Garber, N. and Hoel, L. (2009). Chapters 8 and 10, Traffic and Highway Engineering, 4th
Edition. Cengage Learning.
3. C. Jotin Khisty and B. Kent Lall (2003) Transportation Engineering: An Introduction, 3rd
Edition. Pearson.

Other references:
1. Roess, R. P., Prassas, E. S. and McShane, W. R. (2004). Chapters 17 and 18. Traffic
Engineering, 3rd Edition. Pearson Prentice Hall.
2. Highway Capacity Manual (HCM2000). Transportation Research Board, 2000.

TYPES OF INTERSECTION OR JUNCTION

 Type of intersections based on geometric (i.e., number of legs) and type of

control.
o no control
o priority junction or yield control (United Kingdom)
o all-stop or stop control (North America)
o rotary / roundabout
o signalized or signal control

WHAT IS TRAFFIC SIGNAL CONTROL?

 Signal control is the most common form of traffic control on the road
network.
o It involves any form of traffic control device operated manually,
mechanically, electrically or electronically by which traffic is
alternately directed to stop and permitted to proceed.
o Institute of Transport Engineers (ITE) definition: All power-operated
devices (except signs) for regulating, directing or warning motorist or
pedestrians.

 For signal control to be useful (efficient and safe):

o Indications must be clearly understood i.e. there must be uniformity
and clarity in design.
o Signal priority must be strictly observed i.e. there must be conspicuity
and suitability of design.

 Design of signals must consider:

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 3

o Whether the signal control is appropriate

o Location of signal posts
o Pattern of signal phases
o Timing of signal sequences
o Progression in signal control
 Traffic signals should be used when the following objectives are defined:
o To reduce certain accident types (right-angle collision in an
unsignalized intersection) i.e. to improve overall safety
o To effect orderly movements and decrease average travel time through
an intersection (increased efficiency and intersection capacity)
o To provide proper coordination along a route
o To equalize the quality or level of service for all or most traffic streams
o To permit right-of-way to pedestrians or minor stream
o Land and cost constraints since roundabouts take up much more
space than signalized intersections.

 Poor design of traffic signals can lead to:

o Increased accident frequencies (rear-end collision more so for fixed
time signals)
o Excessive delays and operating costs to vehicles, especially in
uncongested non-peak periods
o Disregard of signal indications
o Signal installations need to be frequently maintained and their
operations monitored
o Signal failure, although infrequent, can lead to serious and widespread
traffic difficulties especially during peak hours.

Types of Signal Control

Pre-timed operation: Signal timings are fixed but varied by time of day,
usually determined based on historical demands. Vehicle detection is not
necessary with this mode of operation.

Semi-actuated operations: Actuation is based on demand from non-priority

traffic (change of phase, green extensions etc) which is detected by either
video, loop detectors etc.

Fully-actuated operations: All signal phases can be changed subject to

demand.

SCATS/GLIDE System in Singapore

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 4

 SCATS – Sydney Coordinated Adaptive Traffic System. First developed and

deployed in Australia in 1977. Now is Version 6.8. Version 7 in 2012.

 GLIDE – Green LInk Determination. Implemented by LTA (then PWD) in

Singapore in 1988. GLIDE is based on SCATS, but with LTA’s customized
signal offsets.

 Entire Singapore has 2080 intersections, divided into 17 regions. Each

region has a Regional Computer, controlling up to 128 subsystems. A
subsystem consists of one local (site) controller, controlling one or several
nearby intersections. Software in the regional computer determines
whether subsystems’ signals should be coordinated or not: Marriage or
Divorce.

Figure 1: GLIDE’s GUI in LTA’s ITS Centre

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 5

GLIDE’s loop detector at stop line Cabinet housing subsystem’s

controller.
Note the 4-digit subsystem number

Lead-in cable connecting loop Cable connecting pull-boxes or

detectors with roadside pull-box detectors with subsystem controller

Figure 2: Roadside furniture for signalized intersections

PRINCIPLES OF SIGNAL CONTROL

Terminology

Intersection: at-grade crossing of two or more roadways.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 6

Approach consists of a lane or group of lanes through which vehicles enter

the intersection.

Lane group is an individual lane or multiple lanes which are grouped based
on the allowed movements (left, through, right) within each lane and the
sequencing of allowed movements by the traffic signal.

Controller: A device in a traffic signal installation that changes the colour

indicated by the signal lamps according to a fixed or variable plan.

Cycle: One complete colour sequence of signal indication i.e. one cycle.

Cycle length: The time in seconds required for one complete colour
sequence of signal indication i.e. one cycle.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 7

Signal phase (or phase): Part of a cycle allocated to a stream of traffic or a

combination of traffic having the right of way simultaneously during one or
more intervals.

Phase plan: A plan describing the order of the signal phase changes.

Phase diagram: A graphical representation of the phase plan.

Interval: Any part of the cycle length which signal indications do not
change.

Offset: Time lapse in seconds or the percentage of cycle length between the
beginning of a green phase at an intersection and the beginning of a
corresponding green phase at the next intersection. It is the time base of the
system controller.

Change and clearance (intergreen) interval: Total length of time in

seconds of the yellow and the all-red signal indications. This time is
provided for vehicles to clear the intersection after green interval before
conflicting movements are released.

All-red interval: Display time of a red indication for all approaches. It is

sometimes used as a phase exclusively for pedestrian crossing or to allow
vehicles and pedestrians to clear very large intersections before opposing
approaches are given green indication.

Figure 3: Two phase signal system

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 8

SIGNAL DESIGN OF ISOLATED INTERSECTIONS

Fundamental Concept I: Idealized Flow Diagram

 Saturation flow rate (s) is the maximum hourly volume that can pass
through an intersection from a given lane (or group of lanes) if the lane (or
group of lanes) were given constant green over the entire hour. May be
viewed as “capacity” of the lane or lane group. Note that saturation flow
rate corresponds to constant minimum headways.
3600
s
h
where: s = saturation flow rate in vphpl and h = saturation headway in
sec/vehicle.
Typically, the maximum saturation flow rate is about 1900 pcphpl
corresponding to a minimum headway of 1.9 sec.

Figure 4: Concept of saturation headway

 Lost time is the portion of the cycle length (or phase time) that is not being
completely utilized due to vehicle starts or stops at the change of phases.
For each phase, there is start-up lost time (typically 2 seconds) and
clearance lost time (typically amber time + all red interval). Start up and
clearance lost times are added up to arrive at the total lost time for a
phase:
t L  tsl  tcl
where:
tL = total loss time for a movement during a cycle in seconds;
tsl = start-up lost time for a movement during a cycle in seconds; and
tcl = clearance lost time for a movement during a cycle in seconds.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 9

 Idealized flow diagram of vehicle discharge at stop line during a green

phase:
tL

tL

Intergreen, I

 From the idealized flow diagram, we can define saturation flow rate s, lost
time l1 and l2, and effective green g.

 We may define the effective green time out of a cycle:

g

C
where g = effective green time for a movement (sec) and C = cycle time
(sec). Note that lower case symbols refer to a phase or movement while
upper case symbols are the common values in a cycle.

 Effective green time (Time when a movement is effectively utilizing the

intersection):
g  G  Y  AR  t L
where g = effective green time for a movement (sec); G = displayed green
time for a movement (sec); Y = displayed amber (yellow) time for a
movement (sec); AR = displayed all-red time (sec); C = cycle time (sec) and
tL = total loss time for a movement during a cycle (sec).

 Effective red time (Time when a movement is effectively not utilizing the
intersection):
r  R  tL  C  g
where r = effective red time for a movement (sec); R = displayed red time
for a movement (sec); tL = total loss time for a movement during a cycle
(sec); C = cycle time (sec) and g = effective green time for a movement (sec).

 We can also define the approach’s volume/satuaration ratio (y) as

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 10

𝑣
𝑦=
𝑠
where v = highest flow of individual lane flow rate in the signal phase
(pcphpl) and s = saturation flow rate in vphpl. v-s ratio only depends on
demand rate and measures the demand level.

 Capacity c (occurs when x = 1) is the maximum hourly volume that can

pass through an intersection from a lane group under prevailing roadway,
traffic, and control conditions:
g
c s
C
 We can also define the approach’s volume/capacity ratio (y) as
𝑣
𝑥=
𝑐
where v = highest flow of individual lane flow rate in the signal phase
(pcphpl) and c = capacity flow rate in vphpl. The v-c ratio depdents on
both demand rate and signal timing. It measures the level of service of
intersection.

Example: A given movement at a signalized intersection receives a 27-second

green time, and 3 seconds of yellow + all red. The cycle length is 60 seconds.
If the saturation headway is 2.4 second/vehicle and the total lost time is 3
seconds. If the flow rate of the movement is 500 vph, calculate v/s and v/c
ratio of the movement.
Calculate effective green:
g=G + Y – L = 27 + 3 – 3 = 27.
Calculate saturation flow
s = 3600/2.4 =1500
Calculate capacity
c=s*g/C=675vphpl
Calculate v/s and v/c
v/s = 500/1500 =0.33; v/c = 500/675 = 0.74

INTRODUCTION TO QUEUING THEORY

 Queues or waiting line phenomena are everyday occurrences. Examples:

vehicles waiting at intersections, passengers waiting at a bus terminal,
aircrafts waiting for landing, containers waiting to be loaded onto ships,
and many other problems not commonly classified as queuing problems.

 Queueing theory is the mathematical study of waiting lines, or queues. A

queueing model is constructed so that queue lengths and waiting time can

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 11

be predicted. Queueing theory is generally considered a branch of

operations research because the results are often used when making
business decisions about the resources needed to provide a service.

 Queueing theory has its origins in research by Agner Krarup Erlang when
he created models to describe the Copenhagen telephone exchange. The
ideas have since seen applications including telecommunication, traffic
engineering, computing and, particularly in industrial engineering, in the
design of factories, shops, offices and hospitals, as well as in project
management.

 A queuing model is characterized by an arrival pattern, a service facility

and a queue discipline. When all three components are constant, the
system can be analyzed by deterministic methods. However, probabilistic
systems are very common and in some instances, deterministic methods
of analyzing queues may be inadequate.

1. Deterministic model: the arrival rate a(t) and service rate d(t) are
deterministic. We can analyse it using queue diagram.
2. Stochastic queue: Either arrival rate or service rate, or both are
random variables following a known distribution. We can analyse it
using statistics and simulation tools.

 The arrival pattern describes the way in which the objects (or customers)
to be served enter the system (rate and manner of arrival).
o Deterministic arrivals (uniform, constant headway), “D”
o Random arrivals (headway ~ exponential distribution), “M”
o Mean arrival rate = 

 The service facility is characterized by the number and the arrangement of

servers and by a service pattern (rate and manner of service).

A service facility can be a single server or a multi-server facility. Service

counters may be arranged in parallel or in series or any combination of the
two.
o Number of servers is denoted as “N” where N ≥ 1

The service pattern usually measures either the rate at which the
customers are processed or the time required to serve individual
customers.
o Uniform departure or service time, “D”
o Exponential departure or service time, “M”

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 12

o Mean departure or service rate = μ

 The queue discipline refers to the rules by which the next customer to be
served is chosen. Some of the most common rules include the following:
o First-in, First-out (FIFO) scheme which serves customers as they arrive
o Last-in, First-out (LIFO) scheme which serves the last customer in the
queue (Also known as a stack)
Customers may be treated equally or according to some priority rules.
Examples of priority are car-pool or bus-lanes or express lanes in
supermarkets. The priority rules may be pre-emptive or non-pre-emptive
depending on whether or not a higher-priority customer is permitted to
interrupt the processing of a lower-priority customer.

 Type of queuing models commonly used in transportation:

o D/D/1 (Deterministic queuing model with deterministic arrivals,
deterministic departures and one server)
Example: A raw material is entering an automated assembly line exactly
every 5 min and operation takes exactly 4 min.
o M/D/1 (Stochastic queuing model with exponential arrivals,
deterministic departures and one server)
Example: An automated pizza vending machine heats and dispenses a
slice of pizza in exactly 3 minutes. Customers arrive at an average rate
of one every 4 minutes.
o M/M/1 (Stochastic queuing model with exponential arrivals,
exponential departures and one server)
Example: Students arrive at a food stall in Canteen at an (average) rate
of 15/hr. Each spends an average of 3 min. buying food.
o M/M/N (Stochastic queuing model with exponential arrivals,
exponential departures and n server.
Example: Customers arrive at queue in customer service at an
(average) rate of 50/hr. There are N representatives. Each customer
spends an average of 3 min. for service.
o M/G/1 (Stochastic queuing model with exponential arrivals, departures
following a general distribution and one server)

DETERMINISTIC QUEUING MODELS

 The basic principle underlying queuing theory (similar to conservation

principles in fluid mechanics, hydraulics and physics) is that the increase
in the number of customers in the storage area during time t equals the

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 13

number of customers that have arrived in t minus the numbers that have
departed in t.

 Definition: Arrival and service processes are well-defined and not subjected
to uncertainties.

 Relaxed condition: Most processes are not strictly deterministic and some
processes may be approximately deterministic.

Mathematical Representation of D/D/1 Model

Consider the time of arrivals of objects in a queuing system.

 The arrival times can be represented by an arrival function, A(t) which gives
the object index of arrival at time t.
A(t1) = 1, A(t2) = 2, ..., A(ti) = i, ..., A(tn) = n where 𝑛 ∈ ℤ+
Since the index is necessarily consecutive, A(t) is also the cumulative
number of arrivals.

Figure 21: Cumulative arrival

 If the service times can be represented by a departure function, D(t′) which

gives the object index at departure time t
Then, D(t′1) = 1, D(t′2) = 2, ..., D(t′i) = i, ..., D(t′n) = n where 𝑛 ∈ ℤ+

 In a first-in-first-out (FIFO) system, the departure index will be consecutive

(i.e. in increasing order). Then, D(t′) is also the cumulative number of
departures.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 14

Figure 22: Cumulative arrival and departure curves

Consider a particular object (vehicle, person etc) k.

 The arrival of object k is at tk = A-1(k) and t′k = D-1(k). This is given by the
horizontal distance between the arrival curve A(t) and departure curve D(t).

 Since the jump in each step of the arrival and departure curves is 1, the
time between arrival and departure is also the area of the rectangle i.e. D-
1(k) – A-1(k).

Figure 23: Waiting time and queue information

 The total time taken by all objects is then:

𝑇𝐴𝐷 = ∑ 𝑤𝑗 = ∑[𝐷−1 (𝑗) − 𝐴−1 (𝑗)]
𝑗 𝑗

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 15

 If the arrival and departure curves are with reference to the queue, then
the time between arrival and departure will be the waiting time.

∑𝑛 −1 −1
𝑗=1[𝐷 (𝑗)−𝐴 (𝑗)]
 The average waiting time after n objects will be: 𝑤 = 𝑛

Figure 24: Total waiting time

Consider the state of the system at time t.

 Given that ti < t, for all values of i = 1, 2, 3, ..., k: A(ti) ≤ k

Given that t′j < t, for all values of j = 1, 2, 3, ..., m: D(t′j) ≤ m

 Hence the number of objects in the system (or queue) at time t will be:
Q(t) = A(t) – D(t)

1 𝑇
 The average queue length will be: 𝑄 = 𝑇 ∫𝑡=0[𝐴(𝑡) − 𝑄(𝑡)]𝑑𝑡

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 16

Figure 25: Queue storage

 Provided that the boundary conditions are compatible, the area between
the arrival and departure curves is:
𝑇
Area = ∫𝑡=0[𝐴(𝑡) − 𝐷(𝑡)]𝑑𝑡 = ∑𝑛𝑗=1[𝐴−1 (𝑗) − 𝐷−1 (𝑗)]

 Provided that the boundary conditions are compatible, the area between
the arrival and departure curves is:
𝑄𝑇 = 𝑛𝑤
𝑛
from which we get: 𝑄 = (𝑇 ) 𝑤
and given that the flow rate, λ = n/T, we obtain Little’s formula:
𝑄 = 𝜆𝑤
Little's law, or formula is a theorem by John Little which states that the
long-term average number 𝑄 of customers in a stationary system is equal to
the long-term average effective arrival rate λ multiplied by the average time
𝑤 that a customer spends in the system. Although it looks intuitively easy, it
is quite a remarkable result, as the relationship is "not influenced by the
arrival process distribution, the service distribution, the service order, or
practically anything else."

A Realistic Scale for Deterministic Queuing Model

 Suppose that instead of dealing with several objects, we are now dealing
with 1000s or 10000s of objects, the step jump of the arrival and departure
curves will be infinitesimally small to the extent that the step function can
be approximated by smooth lines or curves.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 17

Figure 26: Smoothened arrival and departure curves

 Most transportation servers cannot process customers faster than a certain

rate  which we call service rate (or capacity). Knowledge of the service rate
allows the construction of the departure curve.

 When there is no queue, as long as the arrival rate remains less than the
service rate, the departure curve follows the arrival curve (See Figure 27).

 However, when the arrival rate exceeds the service rate, a queue develops
and the server starts working at full speed. As long as there is a queue, the
departure curve is a straight line with a slope . When the straight line
meets the arrival curve, the queue dissipates.

 Important notes:
o The departure rate cannot exceed the service rate of the server. It may be
less or equal to the service rate.
o Cumulative departures can never exceed cumulative arrivals. D(t) can
never be above A(t) in the queuing diagram.
o When a queue is present, the departure rate will be equal to the service
rate.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 18

Figure 27: Arrival and departure rates

Non-FIFO Queuing

 The derivation above is based on FIFO, since the vertical axis is the index
of the object.


 When the queuing system is non-FIFO, then the departure curve will not
be monotonically non-increasing.
 However, if the index of the object is not maintained, but only the
cumulative numbers of arrivals and departures are used, then the graph
will be exactly the same as the FIFO system.
 Hence, Little’s law is valid for any queuing system.

Example: Vehicles arrive at an entrance to a recreational park. There is a

single gate (at which all vehicles must stop), where a park attendant
distributes a free brochure. The park opens at 8:00 am at which vehicles

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 19

begin to arrive at a rate of 480 vph. After 20 minutes, the arrival flow rate
declines to 120 vph and it continues at that level for the remainder of the
day. If the time required to distribute the brochure is 15 sec, and assuming
D/D/1 queuing, describe the characteristics of the queue.

Solution:
 = 480 vph = 8 veh/min for t ≤ 20 min
 = 120 vph = 2 veh/min for t > 20 min
 = 60/15 = 4 veh/min for all t
In this case:
A(t) = 8t for t ≤ 20 min and A(t) = 160 + 2(t – 20) for t > 20 min
D(t) = 4t for all t

Plot graphically:

The point where queue is dissipated can be determined using:

160 + 2(t – 20) = 4t
Solve and obtain, t = 60 min and n = 240.
Shaded area or total delay = 0.5 × (80 × 20) + 0.5 × (80 × 40) = 2400 veh-min
Average queue = 2400/60 = 40 vehicles
Average waiting time or delay per vehicle = 2400/240 = 10 min

Fundamental Concept II: Signalized Intersection Analysis using

D/D/1 Queuing

 Observed vehicle discharge rate at stop line during a green phase:

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 20

Figure 5: D/D/1 signalized intersection queuing with appraoch capacity

(g) exceeding arrivals (C) for all cycles.
Given the properties of D/D/1 queues, a number of general equations can be
derived by simple inspection:

1. The time to queue clearance after the start of the effective green, tc [note
that (r + tc) =  tc and traffic intensity  = /]:
r
tc 
1   
r  tc
2. The proportion of the cycle with a queue Pq: Pq 
C
  r  tc  r  t c
3. The proportion of vehicles stopped Ps: Ps    Pq
 r  g  C
 r  tc   tc t
Ps    c
 r  g  C C
4. The maximum number of vehicles in the queue Qmax: Qmax  r
r 2
5. The total vehicle delay per cycle Dt: Dt 
2 1   
r 2 1 r2
6. The average delay per vehicle davg: davg   
2 1    C 2C 1   
7. The maximum delay of any vehicle assuming a FIFO queuing discipline
dmax = r

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 21

Example: An approach at a pre-timed signalized intersection has a saturation

flow rate of 2400 vph and is allocated 24 seconds of effective green in an 80-
second signal cycle. If the flow at the approach is 500 vph, provide an analysis
of the intersection assuming D/D/1 queuing.

Solution:
Putting arrival and departure rates into common units of vehicles per second,
 = 500 vph/3600 = 0.139 veh/s,  = 2400 vph/3600 = 0.667 veh/s,  = 0.208

Checking to make certain that capacity exceeds arrivals note that the capacity
(g) is 16 veh/cycle (0.667 × 24), which is greater than (permitting fractions
of vehicles) the 11.12 arrivals (C = 0.139 × 80). So the equations in the
D/D/1 model are valid.

r = C – g = 80 – 24 = 56 sec

1. The time to queue clearance after the start of the effective green, tc [note
that (r + tc) =  tc and traffic intensity  = /

tc 
r

 0.208 56  14.71 sec
1    1  0.208 
2. The proportion of the cycle with a queue Pq
r  tc 56  14.71
Pq    0.884 sec
C 80
3. The proportion of vehicles stopped Ps
t 14.7
Ps  c   0.884
C  0.208  80 
4. The maximum number of vehicles in the queue Qmax
Qmax  r   0.139  56   7.78 veh
5. The total vehicle delay per cycle Dt
 0.13956
2
r 2
Dt    275.19 veh-sec
2 1    2 1  0.208 
6. The average delay per vehicle davg
56
2
r2
davg    24.75 sec/veh
2C 1    2  80 1  0.208 
7. The maximum delay of any vehicle assuming a FIFO queuing discipline
dmax
dmax  r  56 sec

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 22

Optimal Signal Timing Design for Isolated Signalized Intersections

Step 1: Establish Analysis Lane Group

Lane group (or movement): A lane group (movement) consists of one or

more lanes on an intersection approach and having the same green phase.
Lane groups for each approach are established using the following
guidelines:

 Placing similar movements into movement groups (left-turn, through and

right-turn).

 Separate lane groups should be established for exclusive right-turn lanes

(Singapore context) unless the approach also contains a shared right-turn
and through lane. Same guidelines apply for exclusive left-turn lanes.

 When exclusive right-turn lane(s) and/or exclusive left-turn lane(s) are

provided on an approach, all other lanes are generally established as a
single lane group.

Figure 1 shows the typical lane group used for analysis.

Figure 1: Typical lane group for analysis

Step 2: Establish Phase Sequence

a) Determine demand Flow Rates and Saturation Flow Rates
b) Determine projected turns

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 23

c) Phase sequence choice

d) Compare feasible phase sequence and choose the one that
minimize the total demand pressure

Step 2 (a): Determine Analysis Flow Rates and Adjusted Saturation Flow
Rates

Analysis Flow Rate

 To account for this varying arrival rate, the peak 15-minute vehicle arrival
rate within the analysis hour is usually used as the analysis flow rate.
o Highest flow of the four 15-minute periods within an hour
V
 Formula: PHF 
V15  4
where:
PHF = peak-hour factor,
V = hourly volume for hour of analysis,
V15 = maximum 15-min flow rate within peak hour,
4 = number of 15-min periods per hour.

Adjusted Saturation Flow Rate

 When saturation flow rates are not measured, they can be estimated by:
11
s  soN  f i
i 1

s = (adjusted) saturation flow rate for the subject lane group, vph
so = ideal/base saturation flow rate, 1900 vphpl
N = number of lanes in the group
fi = the i-th adjustment factor, each factor accounts for one prevailing
condition that differs from the ideal condition, like lane width, % of
heavy vehicles, approach grade, parking activity/lanes, bus blockage,
area type, lane utilization, right and left turns, pedestrian and bicycle
blockage.

Adjustment factors for saturation flow rates

Factor Formula Definition of variables
Lane width fW  1 
W  12  W = lane width in ft.
30 If W > 16 ft, a two lane analysis
may be considered.
Heavy f HV 
100 %HV = percent of heavy vehicles
vehicles 100  % HV  ET  1 in stream
ET = passenger car equivalents
for heavy vehicles = 2 pc/HV

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 24

Grade fg  1 
%G %G = percent grade on a lane
200 group approach
-6 ≤ %G ≤ +10 (negative means
downhill)
Parking N  0.1 
18N m N = number of lanes in lane
fp  3600 group
N Nm = number of parking
manoeuvres per hour
0 ≤ Nm ≤ 180
fp ≥ 0.5 and fp = 1.0 for no
parking
Bus N
14.4N B N = number of lanes in lane
blockage f bb  3600 group
N NB = number of buses stopping
per hour
0 ≤ NB ≤ 250
fbb ≥ 0.05
Type of area fa = 0.9 for CBD
fa = 1.0 for others
Lane f LU 
vg vg = unadjusted demand flow
utilization v g1N rate for lane group, vph
vg1 = unadjusted demand flow
rate on the highest volume single
lane in the lane group, vphpl
N = number of lanes in lane
group
Left turns Exclusive lane: PLT = proportion of LTs in lane
fLT = 0.85 group
Shared lane: fLT ≥ 0.05
f LT  1  0.15PLT
Single lane:
f LT  1  0.135PLT
Right turns Protected phasing: PRT = proportion of RTs in lane
Exclusive lane: group
fRT = 0.95 (For permitted phasing, see
Shared lane: Highway Capacity Manual for
f RT 
1 details, not covered in CE3121)
1  0.05PRT
Pedestrian- f Lpb  1  PLT 1  A pbT  1  PLTA  PLT = proportion of LTs in lane
bicycle group
blockage ApbT = permitted phase
(Left turns) adjustment
PLTA = proportion of LTs
protected green over total LT
green
Pedestrian- f Rpb  1  PRT 1  A pbT  1  PRTA  PRT = proportion of RTs in lane
bicycle group
blockage ApbT = permitted phase
(Right turns) adjustment

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 25

PRTA = proportion of RTs

protected green over total LT
green

Step 2 (b): Determine Right (Left) -Turn Protection

 When right (left)- turn is not protected, vehicles will have to wait for a gap
in the opposing traffic stream.

 Large right (left)- turn volumes without protection could block other
movements, hence creating a gridlock.

 Protecting right (left)- turns means adding more phases, which leads to
higher lost time, and potentially a longer cycle length and longer average
delays.

 When an approach with more than one lane also has a shared right-turn
lane, the operation of the shared right-turn lane should be evaluated to
determine whether it is effectively operating as an exclusive right-turn
lane because of high volume of turning vehicles.

 Left-turns should only be protected when cross product of right (left)-turn

traffic and the sum of opposing through and left (right)-turn traffic
exceeds the following values.

Number of opposing through lanes Minimum cross product

1 50,000

2 90,000

3 110,000

Step 2 (c): Phase Sequence Choice

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 26

 Non-conflicting movements are grouped into a set and receive a signal

phase for a common right of way.

 Sequentially-linked phases are then considered. Some movements may

overlap into one or more phases.

 Within a cycle, all movements must be catered. Some movements may be

repeated in a cycle.

Example: For the given T-junction and movements, derive the possible sets
and phases.

m1
m2
m3

m4

The possible sets and phases are:

 A:{m1, m2}, B:{m3}, C:{m4}
 A:{m1, m3}, B:{m2}, C:{m4}
 A:{m1, m3}, B:{m1, m2}, C:{m4}

Phase Matrix and Diagram

 A phase matrix is a matrix of movements vs. phases.

 A phase diagram is a diagram to show the progression of movements.

 Properties of phase diagram:

o Non-conflicting movements (except non-protected right turn) are
grouped in a set and receive a signal phase.
o Movements lasting more than one phase are called overlapping
movements.
o All movements are shown progressing forward.
o Nodes represent the start of a phase.
o There should be at least one movement losing right of way and at least
one movement gaining right of way during a phase change.
o Within a cycle, all movements must be catered.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 27

Example: Draw the phase matrix and diagrams for the possible sets derived
in the previous example.

Solution:
m3 m2 m4

A B C A

m1
m2 m3 m4

A B C A

m1
m3 m2 m4

A B C A

m1
Phase matrix:
Movement Start End
1 A C
2 B C
3 A B
4 C A

General rules:

1. Keep the number of phases to a minimum

2. Tolerate a conflict when the demand of one of the movements is low.

3. The plan must be consistent with the intersection geometry (number of

RT lanes, crossing distance, sight distance), lane-use assignments,
volumes, approach speeds, and pedestrian crossing requirements.

Step 2 (d): Compare Feasible Phase Sequence And Choose The One That
Minimize The Total Demand Pressure

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 28

1. Determine Critical Lane Group and Total Cycle Lost Time

Critical lane group (movement):

Critical lane group (movement) requires the longest green time in a phase.
Therefore, critical lane group or movement determines the green time that is
allocated to that phase (explained in Step 3). For a phase i, the critical lane
group (the critical v/s ratio) is the largest v/s ratio of all lane groups j
belonging to phase i.
.

 Adjusted peak volume over saturation flow:

𝑣𝑗
𝑦𝑐𝑖 = max{ } ⁡
𝑗 𝑠𝑗

where
sj = prevailing saturation flow of lane group j
vj = adjusted peak volume of lane group j
yj = flow ratio of lane group j

 Critical lane groups (movements) are the lane groups where their flow
ratios constitute the largest sum of flow ratios in a cycle.

 There is one critical lane group in each phase.

 The flow ratio on the critical/longest path of the movement diagram is

called the critical flow ratio.

 Demand Pressure of a signal phasing is measured by the sum of critical

v/s ratios of its phases:

Y=∑𝑛𝑖=1 𝑦𝑐𝑖

2. Rule of Phase Sequence Choice

We should choose the phase sequence such that the demand pressure
Y is minimized, which generally corresponds to the minimum cycle
length (the minimum average delay).

 The total lost time L is the sum of the lost times in a cycle

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 29

n
L   t L ci
i 1

t L ci  tsl ci  tcl ci

where
(tL)ci = the lost time for critical lane group i
(tsl)ci = the start up lost time for critical lane group i
(tcl)ci = the clearance lost time for critical lane group i
n = the number of critical lane groups, which may not equal the number
of phases

Example: Given the following lane group and phase information for a four-
legged approach, derive the phase and movement diagrams. Determine the
sum of critical flow ratios, critical lane groups, and critical flow ratios. If the
lost time per critical movement is 3 sec, what is the total lost time per cycle?
Phase Ia: NBRT+ SBRT; Phase Ib: SB + SBRT (overlapping phase); Phase Ic:
SB + NB; Phase II: EB + WB
Lane Northbound Southbound Northbound Southbound Eastbound Westbound
Group Right Turn Right Turn
Flow 0.18 0.20 0.28 0.31 0.27 0.29
Ratio

Solution:
Phase diagram:

m1 m1 m5
m4 m3
m3
m2
m6

Phase Ia Phase Ib Phase Ic Phase II

Movement diagram:

m1 (0.2) m4 (0.28) m5 (0.27)

I I I II I
a b c a

m2 (0.18) m3 (0.31) m6 (0.29)

Phase I:
m2 + m3 = 0.18 + 0.31 = 0.49 (Critical movements or lane groups)
m1 + m4 = 0.20 + 0.28 = 0.48
Phase II:
m5 = 0.27

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 30

m6 = 0.29 (Critical movement or lane group)

Sum of critical flows Yc = 0.49 + 0.29 = 0.71
Lost time per cycle = 3 + 3 + 3 = 9 sec

Step 3: Produce Timing Plan

(a) Compute cycle length
(b) Allocate green time
(c) Determine Yellow and All-Red Times
(d) Check minimum green time requirements (for pedestrian cross).
If not satisfied, adjust cycle length.

Step 3 (a): Compute cycle length

 Webster optimum cycle length formula:

1.5  L  5 1.5  L  5
Copt  n
 n
v 
1   1   yci
i 1  s ci i 1

where
copt = optimal cycle length to minimize delay (sec)
L = total loss time (sec)
yci = flow ratio of critical movement lane group i

 Webster developed an empirical cycle length formula that attempts to

minimize the average delay (assuming random arrivals).

 This is an approximate formula. Unfortunately, finding the optimal cycle

length is a very difficult exercise.

 Cycle lengths between 45 and 180 sec are used in the field. All cycle
lengths typically end in 0 or 5; thus, if the cycle length estimate is 52 sec,
a cycle to be equal to 55 sec should be selected.

Step 3(b): Allocate Green Time

 After the cycle length has been calculated, we can determine the amount
of green time allocated to each phase.

 n 
 Cycle time: C    gci   L
 i 1 

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 31

yci
 Effective green time gi: gi  n C  L 
y
i 1
ci

where: yci = flow ratio of critical movement lane group i, C = cycle length
(sec), L = total lost time (sec)

 Effective green time gj for non-critical movements:

g j  gi if lane group i receives right of way in one phase
𝑦𝑗
𝑔𝑗 = ∑𝑀 (𝑔𝑖 − 𝐿𝑖)
𝑚=1 𝑦𝑚
if lane group i receives right of way in more than one phases
where:
yj = maximum flow ratio for all lane groups only receiving right of way in
phase j
Li=lost time within critical movement i due to phase change

 Display green time Gj: G j  g j  t L , j  Y j  AR j

Example: Given the movement diagram and the flow ratio in the figure
below, if the cycle time is 120 sec, lost time is 3 sec/phase, yellow time in
each phase is 3 sec, all red time in each phase is 1 sec, determine the
displayed green in each phase.

m1 (0.35) m4 (0.40)

A B C A

m2 (0.1) m3 (0.2)
Solution:
y 0.35
g1  n c1 C  L   120  6   53.2
45.6 sec
sec
0.35  0.40
 yci
i 1

yc 4 0.40
g4  n C  L   120  6   60.8
68.4 sec
sec
0.35  0.40
y
i 1
ci

𝑦2 0.1
𝑔2 = 𝑀 𝑔1 = (53.2 − 3) = 16.7 sec
∑𝑚=1 𝑦𝑚 0.1 + 0.2

0.2
𝑔3 = (53.2 − 3) = 33.5 sec
0.1 + 0.2

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 32

𝐺𝐴 = 𝑔𝐴 + 𝑡𝐴 − 𝑌𝐴 − 𝐴𝑅𝐴 = 16.7 + 3 − 3 − 1 = 15.7 sec

𝐺𝐵 = 𝑔𝐵 + 𝑡𝐵 − 𝑌𝐵 − 𝐴𝑅𝐵 = 33.5 + 3 − 3 − 1 = 32.5 sec

𝐺𝐵 = 𝑔𝐶 + 𝑡𝐶 − 𝑌𝐶 − 𝐴𝑅𝐶 = 60.8 + 3 − 3 − 1 = 59.8 sec

Step 3 (c): Determine Yellow and All-Red Times

Yellow Time

 According to Institute for Transport Engineers (ITE), yellow time is the

time required for a vehicle to traverse one safe stopping distance at its
approach speed:
v 85
Y  tr 
2 a  Gg 
g = acceleration due to gravity, 32.2 ft/s2 (9.807 m/s2)
G = grade of approach, expressed as a decimal
a = deceleration, a = 10 ft/s2 (3.05 m/s2)
tr = reaction time, 1 sec
v85 = 85th percentile speed of approaching vehicles or the speed limit as
appropriate.

All-Red Time

 The all-red (AR) interval must provide sufficient time for a vehicle that has
just entered the intersection legally on a yellow interval to clear the
intersection before giving the green signal to a conflicting movement.

 Formula:
w l
For no pedestrians: AR 
v15
w  l P 
For some pedestrians: AR  max  , 
 v15 v15 
For significant number of pedestrians or protected by pedestrian signals:
P l
AR 
v15
Where:

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 33

l = length of a standard vehicle

v15 =15th percentile speed of approaching vehicles
P = distance from the departure stop line to the far side of the farthest
conflicting crosswalk
w = distance from the departure stop line to the far side of the farthest
conflicting lane

Step 3 (d): Check for Pedestrian Crossing Time

 The minimum green time required for pedestrian crossing must be

checked against the apportioned green time for the phase.

 Minimum pedestrian green time:

L
G p  3.2   0.27N ped for WE ≤ 3.05 m (10 ft)
Sp
L  N ped 
G p  3.2   2.7   for WE > 3.05 m (10 ft)
Sp  WE 
Where
Gp = minimum pedestrian green time in sec
3.2 = pedestrian start up time in sec
L = crosswalk length in m
Sp = walking speed of pedestrian, assumed to be 1.2 m/s
Nped = number of pedestrian crossing during an interval
WE = effective crosswalk width in m.

Points to Note

 Usual cycle time used by LTA: 80 ≤ C ≤ 120 sec. In large and busy
intersections, C can be up to 180 sec.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 34

 Intergreen time used by LTA:

o Y = 3 sec and more if necessary.
o Flashing arrow for right turn = 3 sec
o AR = 1 or 2 sec per phase

 Minimum displayed green = 5 sec. However, for right-turn phase, it can

be as low as 3 sec.

LEVEL OF SERVICE DETERMINATION FOR SIGNALIZED

INTERSECTIONS

 Use “control” delay as the LOS measurement.

LOS Control delay per vehicle
(sec/veh)
A ≤ 10
B 10 – 20
C 20 – 35
D 35 – 55
E 55 – 80
F > 80

 The control delay is the aggregated (i.e., weighted average) delay of all the
approaches of an intersection.
d v A A
dI  A

v A
A

dI = average delay per vehicle for the intersection in sec

dA = average delay per vehicle for approach A in sec
vA = analysis flow rate for on approach A in sec

 For each approach, the average approach delay is calculated from:

d v i i
dA  i

v i
i

dA = average delay per vehicle for approach A in sec

di = average delay per vehicle for lane group i (on approach A) in sec
vi = analysis flow rate for lane group i (on approach A) in sec

 One limitation of the D/D/1 model in computing delay is that it is based

on uniform arrivals. To overcome this problem, HCM 2000 uses the
following control delay equation:

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 35

𝑑𝑖 = 𝑑1 + 𝑑2 + 𝑑3

d1 = average delay due to uniform arrivals (sec/veh)

d2 = average delay due to random arrivals (sec/veh)
d3 = average delay due to initial queue at intersection (sec/veh)

 Average delay due to uniform arrivals

2
 g
0.5C 1  
 C
d1 
 g
1  min 1, x  
 C
Where:
d1 = uniform control delay (sec/veh)
g = average effective green, sec
C = average cycle length, sec
x = degree of saturation
Note that di ≈ d1 for x < 0.5

 Average delay due to random arrivals

 8kIx 
d2  900T  x  1   x  1 
2

 cT 
where:
T = duration of analysis period (hr)
k = delay adjustment factor due to signal control mode (0.5 for pretimed)
I = upstream filtering/metering adjustment factor (1.0 for isolated
intersection)
s = saturation flow rate of the lane group (vph)

 Adjustment factor for signal control mode (k) may be assumed to be k =

0.5 for pre-timed control. For actuated control, k depends on the unit
extension and degree of saturation. In the later case, 0.04 ≤ k ≤ 0.5 , with
higher k for higher degree of saturation and longer unit extension.

 The upstream filtering/metering factor (I) is to adjust for the effect of

upstream signals on the randomness of arrival pattern at the downstream
signal
variance of the number of arrivals per cycle
I 
mean number of arrivals per cycle
Note that for completely random arrivals, the number of arrivals per
cycle follows a Poisson distribution, in which mean=variance, and I = 1.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 36

If upstream is signalized, vehicles are released in bunches, variance <

mean, which means I < 1.

 Average delay due to initial queue. In CE3121, assume d3 =0.

Example: A lane group at a pre-timed isolated intersection approach has C =

80 sec, g = 24 sec, v = 500 vphpl for the 15-minute peak period, and s = 2400
vphpl. The approach has no initial queue most of the time. Determine the
average delay of the lane group.

Solution:
v   500 
   
x  s    2400   0.694
G   24 
   
C   80 
2 2
 g  24 
0.5C 1   0.5  80  1  
 C  80 
d1    24.75 sec/veh
 g  24 
1  min 1, x   1  min 1,0.694 
 C  80 
 8kIx 
d2  900T  x  1   x  1
2
 
 cT 
 8  0.5 1 0.694  
 900  0.25   0.694  1   0.694  1 
2
  5.45 sec/veh
  2400  0.25  
d3 = 0
di = 24.75 + 5.45 + 0 = 30.2 sec/veh

SIGNAL COORDINATION

 The above calculation (d1) assumes uniform traffic arrivals throughout

the cycles. This assumption is unrealistic. The arrival pattern is affected,
among other factors, by the level of coordination between adjacent
intersections.

 The purpose of signal coordination is to time the sequence of signals such

that the as many vehicles as possible arrive when the signal indication is
green.

 Aim is to provide a progressive traffic flow ("green wave") along an arterial

 Generally speaking, the larger the proportion of vehicles arriving during

green, the lower the signal delay.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 37

 Optimum operations of individual intersection may have to compromise

with system performance, e.g. new cycle length C

 Consider vehicles released at a signalized intersection as platoons and the

platoon progresses through a series of intersections at a constant speed

 Time-space chart (See Figure 2).

 Benefits - reduce stops and delay, regulate travel speed

Figure 2: Time space chart for coordinated signals

Nonuniform Arrivals
 Given signal coordination, the arrival rate at green interval will be larger
than the arrival rate at red interval as shown in the figure below.

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 38

TRAFFIC ACTUATED SIGNALS

 Adjust green times (and cycle length) continuously in accordance with

real-time measure of traffic demand from detectors

 In some cases, include skipping of phases

 Amber and all-red intervals are fixed

 Basic timing parameters are:

o minimum (displayed) green
o vehicle interval
o maximum (displayed) green

 Each phase is given a minimum green, extendable by vehicle interval up

to the maximum green.

 Detectors at a typical intersection approach

B A

o A - sense the presence of a vehicle prior to the begin of its phase

CE3121 Transportation Engineering

Traffic Engineering and Transportation Planning 39

o B - sense the arrivals of vehicles during the green interval to help to

extend the green time
o C - detects the presence of right-turn vehicles

CE3121 Transportation Engineering