Analysis of Signalized

Intersections
Winter 2006
CEE 320
 Outline

1. Key Definitions
2. Baseline Assumptions
3. Control Delay
4. Signal Analysis
a. D/D/1
b. Random Arrivals
c. LOS Calculation
d. Optimization
Winter 2006
CEE 320
 Key Definitions (1)
• Cycle Length (C)
– The total time for a signal to complete a cycle
• Phase
– The part of the signal cycle allocated to any combination
of traffic movements receiving the ROW simultaneously
during one or more intervals
• Green Time (G)
– The duration of the green indication of a given
movement at a signalized intersection
• Red Time (R)
– The period in the signal cycle during which, for a given
phase or lane group, the signal is red
Winter 2006
CEE 320
 Key Definitions (2)
• Change Interval (Y)
– Yellow time
– The period in the signal cycle during which, for a given
phase or lane group, the signal is yellow
• Clearance Interval (AR)
– All red time
– The period in the signal cycle during which all
approaches have a red indication
Winter 2006
CEE 320
 Key Definitions (3)
• Start-up Lost Time (l1)
– Time used by the first few vehicles in a queue while reacting
to the initiation of the green phase and accelerating.
2 seconds is typical.
• Clearance Lost Time (l2)
– Time between signal phases during which an intersection is
not used by traffic. 2 seconds is typical.
• Lost Time (tL)
– Time when an intersection is not effectively used by any
approach. 4 seconds is typical.
– tL = l1 + l2
• Total Lost Time (L)
– Total lost time per cycle during which the intersection is not
used by any movement.
Winter 2006
CEE 320
 Key Definitions (4)

• Effective Green Time (g)

– Time actually available for movement
– g = G + Y + AR – tL
• Extension of Effective Green Time (e)
– The amount of the change and clearance interval at the
end of a phase that is usable for movement of vehicles
• Effective Red Time (r)
– Time during which a movement is effectively not
permitted to move.
– r = R + tL
– r=C–g
Winter 2006
CEE 320
 Key Definitions (5)

• Saturation Flow Rate (s)

– Maximum flow that could pass through an intersection if
100% green time was allocated to that movement.
– s = 3600/h
• Approach Capacity (c)
– Saturation flow times the proportion of effective green
– c = s × g/C
• Peak Hour Factor (PHF)
– The hourly volume during the maximum-volume hour of
the day divided by the peak 15-minute flow rate within
the peak hour; a measure of traffic demand fluctuation
within the peak hour.
Winter 2006
CEE 320
 Key Definitions (6)

• Flow Ratio
– The ratio of actual flow rate (v) to saturation flow rate (s)
for a lane group at an intersection
• Lane Group
– A set of lanes established at an intersection approach
for separate analysis
• Critical Lane Group
– The lane group that has the highest flow ratio (v/s) for a
given signal phase
• Critical Volume-to-Capacity Ratio (Xc)
– The proportion of available intersection capacity used
by vehicles in critical lane groups
Winter 2006
CEE 320

– In terms of v/c and NOT v/s

from Highway Capacity Manual 2000
 Baseline Assumptions

• D/D/1 queuing
• Approach arrivals < departure capacity
– (no queue exists at the beginning/end of a
cycle)
Winter 2006
CEE 320
 Quantifying Control Delay

• Two approaches
– Deterministic (uniform) arrivals (Use D/D/1)
– Probabilistic (random) arrivals (Use empirical equations)

• Total delay can be expressed as

– Total delay in an hour (vehicle-hours, person-hours)
– Average delay per vehicle (seconds per vehicle)
Winter 2006
CEE 320
 D/D/1 Signal Analysis (Graphical)
Departure
Rate Arrival
Rate
Vehicles

Queue dissipation

Total vehicle delay per cycle

Maximum delay
Maximum queue

Time
Winter 2006
CEE 320

Red Green Red Green Red Green

 D/D/1 Signal Analysis – Numerical
λ
ρ= ρ < 1.0
µ
• Time to queue dissipation after the start of effective green

ρr
(1 − ρ )
t0 =

• Proportion of the cycle with a queue

r + t0
Pq =
c
• Proportion of vehicles stopped

λ (r + t0 ) r + t0 λ (r + t0 ) µt0 t0
λ (r + g )
Ps = = = Pq
λ (r + g ) λc ρc
Ps = = =
c
Winter 2006
CEE 320
 D/D/1 Signal Analysis – Numerical
λ
ρ= ρ < 1.0
µ
• Maximum number of vehicles in a queue

Qm = λr
• Total delay per cycle

λr 2
2(1 − ρ )
Dt =

• Average vehicle delay per cycle

λr 2 1 r2
2(1 − ρ ) λc 2c(1 − ρ )
Dt = × =

• Maximum delay of any vehicle (assume FIFO)

Winter 2006
CEE 320

dm = r
 Signal Analysis – Random Arrivals

• Webster’s Formula (1958) - empirical

 c  2 +5 ( g / c )
d'= d + − 0.65 2  x
2 1/ 3
x
2λ (1 − x ) λ 
d’ = avg. veh. delay assuming random arrivals
d = avg. veh. delay assuming uniform arrivals (D/D/1)
x = ratio of arrivals to departures (λc/µg)
g = effective green time (sec)
c = cycle length (sec)
Winter 2006
CEE 320
 Signal Analysis – Random Arrivals

• Allsop’s Formula (1972) - empirical

9  x2 
d ' = d + 
10  2λ (1 − x ) 

d’ = avg. veh delay assuming random arrivals

d = avg. veh delay assuming uniform arrivals
(D/D/1)
x = ratio of arrivals to departures (λc/µg)
Winter 2006
CEE 320
 Definition – Level of Service (LOS)

• Chief measure of “quality of service”

– Describes operational conditions within a traffic
stream
– Does not include safety
– Different measures for different facilities
• Six levels of service (A through F)
Winter 2006
CEE 320
 Signalized Intersection LOS

• Based on control delay per vehicle

– How long you wait, on average, at the stop light
Winter 2006
CEE 320

from Highway Capacity Manual 2000

 Typical Approach

• Split control delay into three parts

– Part 1: Delay calculated assuming uniform arrivals (d1).
This is essentially a D/D/1 analysis.
– Part 2: Delay due to random arrivals (d2)
– Part 3: Delay due to initial queue at start of analysis time
period (d3). Often assumed zero.

d = d1 (PF ) + d 2 + d 3
d = Average signal delay per vehicle in s/veh
PF = progression adjustment factor
Winter 2006

d1, d2, d3 = as defined above

CEE 320
 Uniform Delay (d1)

 g
0.5C 1 − 
d1 =  C
 g
1 − min (1, X ) 
 C

d1 = delay due to uniform arrivals (s/veh)

C = cycle length (seconds)
g = effective green time for lane group (seconds)
X = v/c ratio for lane group
Winter 2006
CEE 320
 Incremental Delay (d2)
 8kIX 
d 2 = 900T ( X − 1) + ( X − 1) + 
2

 cT 
d2 = delay due to random arrivals (s/veh)

T = duration of analysis period (hours). If the analysis is based on the

peak 15-min. flow then T = 0.25 hrs.
k = delay adjustment factor that is dependent on signal controller mode.
For pretimed intersections k = 0.5. For more efficient intersections k
< 0.5.
I = upstream filtering/metering adjustment factor. Adjusts for the effect of
an upstream signal on the randomness of the arrival pattern. I = 1.0
for completely random. I < 1.0 for reduced variance.
c = lane group capacity (veh/hr)
Winter 2006

X = v/c ratio for lane group

CEE 320
 Initial Queue Delay (d3)

• Applied in cases where X > 1.0 for the

analysis period
– Vehicles arriving during the analysis period
will experience an additional delay because
there is already an existing queue
• When no initial queue…
– d3 = 0
Winter 2006
CEE 320
 Control Optimization

• Conflicting Operational Objectives

– minimize vehicle delay
– minimize vehicle stops
– minimize lost time
– major vs. minor service (progression)
– pedestrian service
– reduce accidents/severity
– reduce fuel consumption
– Air pollution
Winter 2006
CEE 320
 The “Art” of Signal Optimization

• Long Cycle Length

– High capacity (reduced lost time)
– High delay on movements that are not served
– Pedestrian movements? Number of Phases?

• Short Cycle Length

– Reduced capacity (increased lost time)
– Reduced delay for any given movement
Winter 2006
CEE 320
 Minimum Cycle Length

L× Xc
=
v
C min
X c − ∑ 
n

i =1  s  ci

Cmin = estimated minimum cycle length (seconds)

L = total lost time per cycle (seconds), 4 seconds per
phase is typical
(v/s)ci = flow ratio for critical lane group, i (seconds)
Xc = critical v/c ratio for the intersection
Winter 2006
CEE 320
 Optimum Cycle Length Estimation

1.5(L ) + 5
=
v
C opt
1− ∑ 
n

i =1  s  ci

Copt = estimated optimum cycle length (seconds) to

minimize vehicle delay
L = total lost time per cycle (seconds), 4 seconds per
phase is typical
(v/s)ci = flow ratio for critical lane group, i (seconds)
Winter 2006
CEE 320
 Green Time Estimation

 v   C 
gi =    
 s i  X i 

g = effective green time for phase, i (seconds)

(v/s)i = flow ratio for lane group, i (seconds)
C = cycle length (seconds)
Xi = v/c ratio for lane group i
Winter 2006
CEE 320
 Pedestrian Crossing Time
L  
G p = 3.2 + +  2.7  for WE > 10 ft.
N ped
Sp  WE 

+ (0.27 N ped ) for WE ≤ 10 ft.

L
G p = 3.2 +
Sp

Gp = minimum green time required for pedestrians (seconds)

L = crosswalk length (ft)
Sp = average pedestrian speed (ft/s) – often assumed 4 ft/s
WE = effective crosswalk width (ft)
3.2 = pedestrian startup time (seconds)
Nped = number of pedestrians crossing during an interval
Winter 2006
CEE 320
 Effective Width (WE)
Winter 2006
CEE 320

from Highway Capacity Manual 2000

 Example
An intersection operates using a SB
simple 3-phase design as pictured.

WB

EB
Phase Lane Saturation Flows
group
1 SB 3400 veh/hr
2 NB 3400 veh/hr
Winter 2006

3 EB 1400 veh/hr
NB
CEE 320

WB 1400 veh/hr
 Example
What is the sum of the flow ratios for the critical lane groups?
What is the total lost time for a signal cycle assuming 2 seconds of
clearance lost time and 2 seconds of startup lost time per phase?

SB

30 150
400

30
EB 200

300 WB
20

1000
100 50
Winter 2006
CEE 320

NB
 Example
Calculate an optimal signal timing (rounded up to the nearest 5
seconds) using Webster’s formula.

1.5(L ) + 5
Copt =
1 − ∑ (v s )ci
n

i =1
Winter 2006
CEE 320
 Phase Design

• Separate the conflicting movements in an

intersections into various phases
• If design for no conflicts, impossible
• Design phases with minimum conflicts or
less severe conflicts
• The design is based on geometry, flow
pattern especially the turning movements
Winter 2006
CEE 320
 Two Phase Traffic Signal Timing
Winter 2006
CEE 320
 Three Phase Traffic Signal Timing
Winter 2006
CEE 320
 Four Phase Signal Time with
through and right turning only

1. Turning movements are

comparable to through
movement
2. When through traffic and
turning traffic need to share
same lane
3. May be inefficient when
turning are relatively low
Winter 2006
CEE 320
 Four Phase Signal Time with through
and right turning only

Very efficient for one lane one movement Rarely used

Through traffic volume is high
Winter 2006
CEE 320
 Equivalent Straight-Through Passenger Cars

Passenger Car Equivalent (PCE)

Why Equivalent Factor for turning movement?
 Because turning movements and heavy vehicles
require additional green time to complete their
maneuvers.

Conversion to EST passenger cars

Table 6.1 Adjustment Factors For EST Passenger Cars

Vehicle Type and Movement Adjustment Factor

Passenger car (straight through) 1.0
Heavy vehicle 1.5
Left-turning 1.6
Right-turning 1.4
Winter 2006
CEE 320
 Lane Capacity

• Green Ratio = effective green time/ cycle

length
• The capacity of a lane can be computed as
ci = si (g/c)
c = capacity of lane in veh/h
s = saturation flow rate in veh/h
c = cycle time
Winter 2006
CEE 320
 Critical Lane

1. During any green signal phase, several

lanes are allowed to move. One of these
will have the most intense traffic.
2. It needs more time than any other lane
moving at the same time
3. This lane is critical lane and it is only one
in each signal phase
Winter 2006
CEE 320
 Calculate Change Interval

• Change Interval = YT + AR
V
YT = t p +
2a + 2 g r G
w+l
AR =
V
– YT = yellow time (rounded to the nearest 0.5 s)
– tp = perception/reaction time as 1s
– V = speed of the vehicle , taken as 85th percentile speed
– A = deceleration rate for the vehicle, taken as 3.05 m/s2
– G = percent grade
– gr = acceleration due to gravity
– AR = all red time
– w = width of the cross street in meters
– l = length of the vehicle
Winter 2006
CEE 320
 Calculate Minimum Cycle
Length
• Cycle length > 120 seconds are unusual
• The calculation of optimum cycle length was developed by
Webster (1958), who assumed that vehicles arrive at an
intersection on a random basis. The overall total intersection delay
is minimized.

1.5LT + 5
1.0 − ∑i =1 y i
c= n

 c is the cycle length, rounded up to nearest 5 s

 LT is the total lost time, approximated as the total
yellow and all-red times per cycle
 yi is the ratio of the observed lane volume (critical) to
the per-lane saturation flow in each phase
Winter 2006
CEE 320
 Allocate Green Time

The cycle length is the sum of all effective

green times plus the total lost time

Distribute the sum of all effective green
times by
 y 
GTi = TGT ×  n i 
 
 ∑i =1 y i 
• yi = critical lane volume
Winter 2006
CEE 320
 Check Pedestrian Crossing
Time
• Its ability to provide adequate pedestrian crossing time
• The minimum green time required for pedestrian crossing
should be checked against the apportioned green time for
the phase
• For signal timing purposes, the pedestrians walk is
assumed at a rate of 1.2 m/s. 0.9 m/s is used for elderly
pedestrians.
PGT = 7 + w − YT − AR
PWS
PGT − the pedestrian green time in seconds
PWS - the pedestrian - walking speed in m/s
YT - the yellow time in s
AR - the all red - time in s
w - width of the street in meters
Winter 2006
CEE 320