An Evaluation of Signalized Intersection

System Analysis Techniques

Walter J. Freeman, P.E., Kien Y. Ho, P.E., Elizabeth A. McChesney, E.I.T.

A system of signalized intersections is a critical element in the smooth operation of both arterial
and urban street facilities. The amount of vehicular and pedestrian traffic which can be
processed by a system of intersections depend on a) characteristics of the traffic and pedestrian
stream, b) traffic control measures, c) various physical and operating characteristics of the
roadway and d) the environmental conditions which have a bearing on the experience and
actions of the driver. Because many such factors influence interrupted flow through a system of
intersections, it is important that the best technique be utilized for analyzing a system of
intersections. A system of signalized intersections is defined as two or more signalized
intersections that are closely spaced (under 1500) together. Intersection systems not only
control, to a large extent, the capability of major and secondary arterial streets to accommodate
the flow of traffic and pedestrians in an urban area, but they also may seriously affect or limit
the ability of nearby freeways to perform at maximum efficiency. The system of signalized
intersections selected for this study provide an ideal representation of this operational scenario.

Today there is a variety of engineering analysis tools/softwares available to analyze and model
an existing or proposed signalized intersection system. While the majority of the analysis tools
provide the engineer with valuable information pertaining to intersection system delays, queue
length, saturation flow, levels of service, etc, there is not a single analysis software that could
accurately predict all this valuable information for an actual or proposed field condition. For
example, when signalized intersections are closely spaced, some software analysis techniques
are good at predicting delays only and weak at predicting other valuable information such as
queue length or queue spill back.

The purpose of this report is to evaluate techniques for the design and analysis of a system of
signalized intersections. As a case study, a grid system of eight signalized intersections located
in downtown Boston was chosen for this report. These eight closely spaced intersections,
provide an ideal condition for testing analysis tools ability to predict all the available
information related to delays, queue length, saturation flow rates and other measures of
effectiveness.

This report compares several analysis techniques and their predicted and observed results for the
eight observed network intersections. Evaluation includes testing how well these techniques
predict delays, queue lengths, queue spillback and levels of service. In addition, sensitivity
analyses are performed to evaluate the limitations of these techniques.

Finally, the report will determine how efficiently each analysis technique performed on the
signalized intersection system. In addition, the report shall spell out the requirements for an
"acceptable technique" to aid in the design and evaluation of signalized intersection systems.
LITERATURE REVIEW

The history of signalized intersection system analysis techniques is addressed in this review. A
detailed literature search revealed that signalized intersections capacity have been subjected to a
great deal of study since the mid 1950s. The procedures/techniques for the design and analysis
of signalized intersections are tabulated as follows:

Software Softwares Function Types of Facility

1. HCS Highway Computes intersection capacity, Developed for analyzing isolated

Capacity Software delays and level of service (LOS) intersections. The intersections
Release 3.0 based on the Highway Capacity may be signalized or controlled
Manual. by two-way or all way stop signs.
Applying arrival adjustment
factors to interconnected lane
groups may approximate
progression.

2. PASSER II Analyzes arterial progression only. Developed for analyzing arterial

Optimize and simulate only the progression.
mainline arterial and ignores the side
streets. Optimize cycle length, splits
and offsets. Provide bandwidths,
time-space diagram, efficiency
attainability, fuel consumption and
speed progression.

3. PASSER III Analyzes signalized diamond Developed for analyzing

interchanges only. Computes signal signalized diamond
timing coordination for the signalized interchanges.
interchange. Optimizes offset and
splits.

4. PASSER IV Analyzes a network of signalized Developed for a network of

intersections and based on more than one arterial where
progression bandwidth optimization bandwidth optimization is the
also based on MAXBAND. main goal . It also computes the
cycle length and green split but
not phasing .

Software Softwares Function Types of Facility

 Analyzes and optimizes a network of Developed for analyzing and
5. SYNCHRO signalized intersections. It provides a simulating a network / grid of
detailed summary report on capacity, signalized intersections.
level of service, lanes, volumes,
timing, queue lengths, blocking
problems, delay, stop and fuel
consumption.

6. SimTraffic Analyzes and simulates a network/ Developed for analyzing and

grid of signalized and unsignalized simulating a network / grid of
intersections. It provides result on the signalized and unsignalized
amount of MOEs such as delay, intersection.
stops, queues, average speed, fuel
consumption, and throughput. Sim
Traffic models intersection traffic
throughput based on car following
formulas, acceleration rates,
deceleration rates, reaction to yellow
light, reaction times, gap acceptance,
cruise speed, turning speed and
vehicle driver performance
characteristics.

7. SIDRA Analyzes signalized and unsignalized Developed for isolated signalized

intersections and roundabouts. It intersections and roundabouts
computes delay, level of saturation ,
and level of service. Its output
consists of queue length, stop rate,
fuel consumption, and reserve
capacity. Calculates queues for
various probabilities of occurance
based on empirical data.

8. SIG/CINEMA Analyzes and optimizes cycle length, Developed for isolated signalized
timing and phasing sequence. intersections.
Equalizes v/c ratios for critical lane
group analysis, delay/vehicle for
critical lane group and all
approaches. Minimizes total delay of
intersection.

Software Softwares Function Types of Facility

9. HCM/CINEMA Computes level of service and lane Developed for isolated signalized
capacity for all approaches. Similar intersection.
to the HCM (1985) procedure for
analyzing isolated signalized
intersection.

10. TRANSYT 7F Evaluate existing timings and Developed for analyzing a

optimize proposed condition to network and grid system of
minimize stops, delay, fuel signalized intersections.
consumption and cost. Provides
delays, average queues, stop, fuel
consumption. & Time space
diagrams .

11. NOSTOP Analyze arterial progression program Developed for analyzing

that optimizes offsets and cycle signalized intersection on
lengths to provide maximum two- arterial street.
way progression. Provide a plot of
the cycle optimization results and a
time-space diagram.

12. TRAF-NETSIM Simulate a network / grid system of Developed for simulating a

signalized intersections. Provide an network / grid system of
animation of the system that display signalized intersections.
effect of progression, bus stop , gaps,
delays , signal phasing and timing
plan, queue for all approach
intersections. It also provides part of
the system in real time.

13. CORSIM Current update for the TRAF- Developed for simulating a
NETSIM and TRAF-FRESIM network / grid system of
products. Corsim is similar to Sim signalized intersections.
Traffic with some limitations and
enhancements. Relative limitations
include longer running time and
capacity to analyze fewer
iintersections. Relative enhancements
include ability to model parking, bus
stops and random interruptions.

Among all the analysis techniques identified, six techniques were selected for consideration in
this study. The five techniques were selected based on each of the techniques ease of use and its
ability to compute important traffic characteristics of a network of signalized intersections. For
comparison purposes, the HCS techniques was also selected for this study. The techniques
considered are:

1. TRANSYT 7F
2. CORSIM
3. SimTraffic
4. Passer
5. Synchro
6. HCS

A detailed description of each of the techniques is described below. Out of these six techniques,
the four techniques that specifically analyze a network / grid system of signalized intersections
were further reviewed and categorized based on techniques input requirements and output
options as shown in Table 1.

Table 1. A Comparison of Signal System Analysis Input Requirements and Output Options.

C T
o r
Signal System Input Requirements r a
and Output Options Synchro Sim- s n
3.2 HCS Traffic i s
m y
t
Input Requirements
Map Links and Nodes X X X X
Coordinates X X X X
Lanes Number and movements permitted X X X X X
Ideal Saturation Flow X X
Lane Width X X
Grade X X
Detector Locations X X
Turning Speed X X X X X
Right Turn on Red X X X X X
Volumes Traffic X X X X X
Conflicting Pedestrians X X X X X
Peak Hour Factors X X X X X
Growth Factors X X X X X
Heavy Vehicles X X Limited (L) L L
Adjacent Parking Lane X X X
Parking Maneuvers X X X
Bus Stops per Period X X X
Control Type Sign L X X X L
Signal X X X X X
Signal Settings X X X X X
Lost Time X X X X X
Phase Sequence X X X X X
Left Turn Type X X X
 Right Turn Type X X
C T
o r
r a
Synchro Sim- s n
3.2 HCS Traffic i s
m y
t
Output Options
Total Delay X X X X
Stop Delay X X X X
Total Stops X X X
Stops/Vehicle X X X
Travel Distance X X X
Travel Time X X X
Fuel Used X X X X
Emissions X X X
Vehicles Entered & Exited X X X X
Vehicles Denied Entry X
Maximum Queue X X
95th Percentile Queue X
50th Percentile Queue X X
Queuing Penalty X X

1. TRANSYT 7F

Transyt 7F is a traffic signal simulation and optimization program that can evaluate an
existing traffic signal timing condition and evaluate and optimize a proposed signal timing
conditions. The optimization involves the minimization of delay, stops, fuel consumption and
fuel cost. The procedure provides output that includes average queues, delays, stops, fuel
consumption, time-space diagrams, flow profile diagrams and platoon progression diagrams.
The main function of this procedure does not provide bandwidths but it does minimize stops,
delay and fuel consumption, which may or may not provide arterial progression. The
disadvantage of this procedure is it provides a poor model for analyzing or simulating arterial
progression if the primary goal is wide bandwidth along the arterial. However, this is an
excellent tool for minimizing stops and delays in a network / grid system of signalized
intersection. This procedure is mainly used for traffic signal network and arterial where
signalized intersections are closely spaced (less than 1500) together. The Transyt 7F
procedure provides a detailed methodology for optimizing signal system, also called MOST.
The methodology for optimizing signal system includes the following: conduct and management
of signal timing projects; traffic signal timing elements as applied to the traffic signal timing and
analysis program; timing performance measures; model calibration; and timing implementation.

2. SimTraffic

SimTraffic is a simulation program developed for analyzing and simulating traffic behavior in a
network / grid of signalized and unsignalized intersections. Its primary function is to check and
fine tune traffic signal operations that are normally difficult to model such as closely spaced
intersections with blocking and lane change problems, operations near unsignalized intersections
and driveways, and operations of heavily congested intersections. SimTraffic is run directly
from Synchro data input and requires data related to mapping, links, geometry, lanes, volume,
timing and actuation. When compared to its predecessor ,CORSIM, SimTraffic generally uses
the same driver vehicle performance characteristics, and gives comparable amounts of MOEs
(measures of effectiveness) such as delay, stops, queues, average speeds, fuel consumption and
throughput. Although SimTraffic has fewer features than CORSIM, it can model larger
networks and is much easier to use.

3. CORSIM

SimTraffic and Corsim are virtually identical in describing existing traffic operations. The
traffic behavior algorithms are the same. This program will simulate bus stops and parking
maneuvers, which SimTraffic will not. However, since virtually no parking activity occurred
during the study, we elected to use the more user friendly SimTraffic for our study.

4. PASSER II

Passer II performs signalized intersection arterial progression analysis. This technique simulates
and optimizes arterial progression. This technique is only used to evaluate arterial progression
and it is not a good tool to use for analyzing a network or grid system of signalized intersections.
It is an excellent model for bandwidth optimization. It calculates level of service, queue,
bandwidth, efficiency, fuel consumption, number of stops, speed of progression, a time-space
diagram and offsets. Furthermore, it optimizes cycle length, splits and offsets.

Since the network studied is a grid rather than an arterial, we did not evaluate this software.

5. Synchro

Synchro uses 1994 Highway Capacity Manual Techniques and considers the same factors as
HCS described below. In addition, traffic signal offsets and random traffic variations are
factored into the computational procedure.

Like Transyt 7F, Synchro uses traffic signal optimization procedures that can evaluate
existing traffic signal timing conditions and evaluates and optimize a proposed signal timing
conditions. .

The procedure provides output that includes average and 95 percentile queues, delays, stops,
fuel consumption, and percent of time that queues exceed available storage.

6. HCS

HCS (Highway Capacity Software) is a program based on the Highway Capacity Manual. Its
primary function is to analyze capacity and provide level of service for isolated intersections.

HCS requires the input of factors related to ideal conditions (12 ft lanes, level grade, no
parking, all passenger vehicles, no pedestrians, etc.) which affect capacity and level of service.
Each intersection required the following traffic inputs: number of lanes per approach, volumes
per lane, lane width, % grade, % heavy vehicles, parking, # bus stops per hour, conflicting
pedestrian crossings per hour, pedestrian button and minimum pedestrian green time, arrival
type, right turns on red and lost time. The required timing inputs for each intersection include
phasing diagrams, whether the signal is actuated, and the green, yellow and red times for each
phase. HCS output the following information: adjusted saturation flows for each approach,
volume adjustments, capacity analysis, delays and levels of service.

For the purpose of this study, HCS was run for each of the eight urban signalized intersections.

DATA COLLECTION AND REDUCTION

Traffic data utilized in this report were obtained in the Boston downtown central business
district. Eight signalized intersections described below were selected for this study. These
intersections are located in the vicinity of Boston City Hall, Haymarket Bus Station, Freedom
Trail and major on and off ramp connection to Interstate 93. Video cameras were set up at each
of the eight locations during the PM peak hour period. Data were collected for the heaviest
traveled weekday peak hour conditions, which were from 5:00 to 6:00 PM. Each of this
locations was videotaped for a minimum of 45 minutes between 5:00pm and 6:00 PM. On
certain intersection approaches where the traffic volumes were heavy and back of queues long,
an additional dedicated camera was assigned to tape the approach.

The traffic volume, delays, vehicle classifications, and maximum back of queue were obtained
for most approach lane groups at each of the eight intersection locations. These data were
collected by cycle for 20 cycles. Twenty cycles were selected since SimTraffic calculates only
the longest back of queues and the longest queues in twenty cycles represent the 95 percentile.
Signal timing did not vary over the study period since all signals analyzed are pretimed. Cycle
lengths, signal phases and offsets were measured before and after each analysis period and they
were unchanged.

STUDY LOCATIONS

Geometry and Lane Use

Sudbury Street at Congress Street

Congress Street is a divided; two-way roadways crossed by Sudbury Street a northbound one
way street. Between intersections, Congress Street has a fence in the median to discourage
midblock pedestrian crossings and no parking except for buses.

A thousand feet upstream of its intersection with Congress Street, Sudbury Street begins with
two travel lanes with parallel parking on the right side and angle parking on the left. Four
hundred and eighty feet from the intersection, general parking is prohibited on the right side and
parking is restricted to bus passenger pickup and discharge. Eighty feet from Congress Street
the left side angle parking is eliminated to permit entry to the Haymarket parking garage. From
this point to the north, the approach is marked for five lanes, with the right curb lane marked for
right turn only and used for bus pickup and discharge. Within 60 feet of the Congress Street
intersection, an illegally parked car occupied the left lane, which is marked as left turn only.

During our PM peak analysis period, the right curb lane is usually occupied with buses lying
over or engaged in picking up or discharging passengers and other buses occasionally use the
second lane from the curb to pickup passengers. Typically, more than 60 feet from the
intersection there are three lanes for moving traffic: a right turns lane, a right/through lane, and a
left/through lane. Additionally, when there is blockage of the right lane, vehicles will turn right
from the second and third lane from the curb. In our study period, 148 vehicles (25% of the
total lane traffic) turned right from the second from curb lane and 9 vehicles turned right from
the third from curb lane.

On the eastbound Congress Street approach there are three through lanes and an exclusive left
turn lane about 140 feet long. Buses stop to pickup or discharge passengers on the right curb
lane. Beyond Sudbury Street, the median lane is signed for left turn only.

On the westbound Congress Street approach there are three through lanes with the curb lane
being used mostly for right turns.

There are crosswalks on all approaches with heavy pedestrian traffic during peak periods, with J
walking prevalent on the northern crosswalk during the PM peak.

Congress Street and Merrimac Street at New Chardon Street

Congress/Merrimac Street is a divided road in the vicinity of the intersection. New Chardon
Street is a four lane, one way southbound street.

Westbound, Congress Street has three lanes of which one is exclusively for left turns. During
our study period, a bus was parked on the westbound curb lane 25 percent of the time.
Eastbound, Merrimac Street has two lanes, with no parking and the curb lane used for through
and right turning traffic.

Southbound approaching Congress Street, New Chardon Street has four lanes: one for lefts, one
for left and through traffic, with the others unmarked. Occasionally, the third lane is used for
left turns. New Chardon Street south of the intersection is one way with three travel lanes and
one parking lane. There is a major parking garage entrance on the left side of the street,
immediately followed by the parking garage exit.

There are crosswalks on all approaches.

New Chardon Street and Ramp RC

Ramp RC intersects New Chardon Street between Blackstone Street and Congress Street. Ramp
RC has one lane that is stop sign controlled. All ramp traffic must turn right at New Chardon
Street. Immediately downstream of Ramp RC New Chardon Street is four-lane wide. The ramp
essentially adds a lane to New Chardon Street, but most ramp traffic turns left at Congress
Street. About 150 feet downstream of the Ramp RC intersection is a very heavily traveled,
uncontrolled crosswalk.
New Chardon Street at Blackstone Street, Relocated Cross Street and North Washington
Street
Blackstone Street is a one way street with two lanes carrying through and left turning traffic.
Many buses use this approach to make U-turns into the Haymarket bus station.

Relocated Cross Street is also a one way street with two lanes at the intersection, carrying
through and right turning traffic.

Blackstone Street is a divided, four lane, two way street with no parking in the vicinity of the
intersection. Some buses from North Washington Street turn into the Haymarket bus station
immediately downstream of the intersection.

New Chardon Street is a one way, three-lane street with no parking south of the intersection.

There are crosswalks on Cross Street Extension and Blackstone Street.

Sudbury Street at Blackstone Street

Sudbury Street is a five lane, one way street. A Jersey Barrier that limits traffic movements to
right turns into and out of a new parking garage separates the right curb lane from the other
lanes. This barrier continues around the northeast corner onto Blackstone Street. Few vehicles
except those entering and leaving the garage use this lane.

Lanes on Sudbury Street to the left of the barrier consist of two for left turns only, one for
through movements, and one for through and right turning traffic.

Blackstone Street is a two lane, one way street leaving the intersection in both directions. It is
one way eastbound east of Sudbury and one way westbound west of Sudbury. East of Sudbury,
Blackstone Street has two travel lanes with parking on the right side. West of Sudbury,
Blackstone Street has two travel lanes with no parking.

The continuation of Sudbury Street north of Blackstone Street is a one way, two lane road
segment joining a U-turn from Cross Street to form a weaving section leading to Ramp RG and
Blackstone Street.

The traffic actuated, signalized exit from the MBTA bus station joins Sudbury Street
immediately south of Blackstone Street. Many buses from this exit make a U-turn onto
Blackstone Street. Frequently the phase is skipped since many buses turn left on red.

There are crosswalks on the Sudbury Street and the western Blackstone Street legs of the
intersection.

Traffic Signal Operations

Five of the Congress Street signalized intersections are interconnected and part of the system
including the Blackstone Street intersections with Sudbury Street and New Chardon Street.
During our period of analysis, all system locations were pretimed, operating with a 100-second
cycle. This system included:
Congress Street at Merrimac and New Chardon Street
 Congress Street at Sudbury Street
Congress Street at Hanover Street
Congress Street at North Street and City Hall garage
Congress Street at City Hall Crosswalk
Blackstone Street at Sudbury Street and MBTA Station
Blackstone Street at New Chardon Street and Cross Street

Two other signalized intersections are adjacent to the system, but are not part of the system and
operate on different cycle lengths. They are Blackstone Street at Hanover Street and Congress
Street at State Street. These locations were included in the study due to their proximity and the
impact their signal operations have on traffic arrivals at the system.

Intersection phasing for the system locations is described below.

Congress Street at New Chardon Street is controlled by a three phase, pretimed traffic signal.
The westbound left turns move only on the leading protected phase. Pedestrians walk with
green.

Sudbury Street at Congress Street is controlled by a three-phase traffic signal. The eastbound
left turns have a lagging protected phase during which these left turns and non-conflicting
crosswalks have right-of-way.

Congress Street at Hanover Street has a two-phase signal with right turns from Hanover Street
sharing a phase with the pedestrian crossing of Congress Street.

Congress Street at North Street has a four-phase signal with the following phase sequence:
Congress Street eastbound
2-way Congress Street (during which left turns are permitted)
exclusive pedestrian walk
North Street and the City Hall garage exit

Congress Street at City Hall crosswalk has the following phase sequence:
exclusive pedestrian walk
Congress Street eastbound
2-way Congress Street

Blackstone Street at Sudbury Street has three phases:

Sudbury Street northbound
exclusive pedestrian walk, when actuated
MBTA parking station, when actuated

Blackstone Street at New Chardon and Cross Street has two phases, with Blackstone Street left
turns permitted.
EXISTING AND COMPUTED OPERATIONAL CONDITIONS

Table 2. shows a comparison of measured and computed delays and queues for each of the lane
groups in the study area. Following Table 2 we have described traffic operations at each study
area intersection as observed or measured and as predicted by the various analysis methods
employed.

Table 2. Analysis Result Comparison

Stopped 50 %-Tile 95%-Tile Back of

Back of Queue (ft)
Delay/Veh Queue (ft.)
HCS Synchro SimTraffic Actual Synchro Actual Synchro SimTraff Actual
ic
New Char. at 21.4 20.4 21.4 23.1
Congress
EB 24.7 24.6 28 24 115 161 141
WBL 27 30.8 30 157 150 208 210 225
WBT 23.4 11.3 17 276 100 348 175
WB 25.1 20.6 22 23 276 150 348 210 225
SBL 17 17.7 203 238 288 280 400
SBTR 19.9 20.4 79 290 115 260 525
SB 18.8 19.2 16.5 23 203 290 288 280 525

Congress at Sudbury 15.1 15.5 54 20.3

EBL 29 31.3 38 119 188 135
EBT 17.8 8.3 19 68 93 425
EB 19.8 12.9 15.5 22.8 119 175 188 425 250
WB 9.5 2.9 5.8 6 15 34 384
NBTL 15 15.2 94 152 51
NBTR 10.9 21.5 349 438 1028 500
NBR 24.4 25.5 288 480 1028
NB 15.1 22 100 25 349 300 480 1028 500

Congress at Hanover 3.3 3.7 3.2

EB 3.1 3.5 95 87 198
WB 2.7 3 22 31 50
SB 19.3 19.7 23 52 30

Congress at North St. 14.4 14.2 12.6

SBL 16.4 38.6 182 352 350
SBTR 18.2 16.8 184 225 384
SB 17.7 22.4 18.5 184 352 384
NBTL 7.1 0 0 1 14
NBR 2.6 1.5 40 98 14
NB 5.4 1 0.1 40 98 14
WBL 36 29.9 95 162 140
WBLTR 29.4 29.9 89 154 105
WB 33.7 29.9 35.9 95 162 140
Congress at Crosswalk 11 11.1 18.8
SB 0 0.7 2.9 18 85
NB 24.2 22 33.4 256 406 304

Blackstone at Hanover 3.9 3.9 3.1

SB 3.9 3.9 3.1 71 80

Blackstone at Sudbury 1.1 2.1 3.1

NBL 0 0.5 50 85
NBTR 0 0.5 20 70
NBR 0 0.5 30 37
NB 0 0.5 0.75 50 85
T-Station EB 23.2 34.2 36.9 91 58 80

Blackstone at New 5.3 5.9 8

Chardon
NB 0 1.1 2.3 0 112 112
SB 3.6 3.7 3.6 90 100
WB 26.2 26.2 31.8 123 122

NEW CHARDON AT CONGRESS STREET operates at a C level of service, with all arrivals
being serviced during the cycle. There are, however, queuing problems during the course of the
cycle caused by downstream blockages. New Chardon Street left turns have to wait nearly every
cycle for the eastbound queue on Congress Street to move forward. Many of these left turners
enter the intersection but are unable to clear the intersection; thereby blocking westbound left
turns on the following phase.

All of these blockage problems are resolved within a few seconds, as the blocking vehicles move
forward, so gridlock does not occur.

None of the software programs correctly indicated blockage problems. Synchro predicted longer
maximum (95 percentile) queue lengths than SimTraffic, but not as long as those observed on
two of the three approaches. On the third approach, Synchro over-predicted maximum delays by
50%, while SimTraffic predicted maximum queues within 10 percent of that observed.

Overall, stopped delay predictions agreed well with those measured mostly within 10 percent.

CONGRESS STREET AT SUDBURY STREET operates at a C level of service, with all

arrivals being serviced within a cycle. During a few cycles per hour, eastbound through traffic
and northbound right turn traffic is blocked by downstream queues on Congress Street.
Frequently, J-walking pedestrians block eastbound left turns for a few seconds, but the vehicles
easily clear the intersection within their phase. Eastbound back of queues extends into the
upstream New Chardon Street intersection almost every cycle, but the last vehicle always clears
during the green.

The Sudbury Street approach has the longest delays and queues of all intersection lane groups.
The back of queue during the cycle on this approach extends 8 to 20 vehicle lengths from the
stop line. When there are buses occupying the right lane, the queue is longer, with no buses the
queue is shorter. Vehicles in the longest queue are often delayed half a minute.

HCS and Synchro predicted average delays for the intersection that were about two thirds of
those measured. SimTraffic predicted double the measured delays.

None of the models predicted the eastbound blockage or high delays that were observed.
Synchro and HCS under-predicted northboound delays by 24 and 17 seconds per vehicle,
respectively, while SimTraffic over-predicted delays observed by more than a minute.

SimTraffic accurately predicted maximum eastbound queues, but significantly over estimated
northbound maximum queues. Synchro, on the other hand, accurately predicted northbound
maximum queues, but significantly underestimated eastbound queues.

CONGRESS STREET AT HANOVER STREET operates at an A level of service. When

long queues occur they are due to backups from North Street.

Predicted average delays were within one second of each other here.

CONGRESS STREET AT NORTH STREET AND THE CROSSWALK should be

considered as one intersection since they are about 100 feet apart. Considered as such, the level
of service is D, with westbound right turning queues to North Street extending back to State
Street almost every cycle and eastbound left turn queues occasionally extending upstream
through Hanover Street to Sudbury Street. Pedestrians crossing North Street often interfere with
Congress Street left and right turning traffic.

The three software packages predicted delays within 10 percent of one another at this location.

The westbound queue that often extends to State Street was predicted by SimTraffic but
underestimated by Synchro.

BLACKSTONE STREET AT SUDBURY STREET operates at an A level of service. The

pedestrian phase is seldom called and the buses from the T-station often turn left on red without
waiting for a signal.

All software also predicted very short delays at this location.

BLACKSTONE STREET AT NEW CHARDON STREET operates at a B level of service.

All vehicles arriving on red were served on the following green and delays are minimal except
on the southbound approach, where they average about half a minute per vehicle.

The software programs all predicted delays within 2 or 3 seconds of each other.
CONCLUSIONS

For most lane groups, Synchro, HCS, and SimTraffic computed stopped delays with the same
level of service. Computations by approach with these methods also yield delays that are within
20 percent of measured delays. One exception was the northbound approach of Sudbury Street
to Congress Street which has various types of parking and curb activity along its length.

SimTraffic underassigns traffic to the right lane on this Sudbury Street approach, resulting in
computations of longer queues and delays than actually occur. HCS ignores the short available
storage and predicts delays that are too short by nearly ten seconds.

Synchro generally predicts slightly shorter queues than observed and less than computed by
SimTraffic.

SimTraffic would be more accurate and useful in urban environments if bus stops and other
user-specified interruptions could be specified.

Using proper arrival type factors HCS yields results comparable to those computed by network
models in predicting delays. Research by the New England ITE Technical Committee indicated
that this close agreement would not occur. We are surprised that this close agreement occurred.
One reason for this may be that there is only one lane in the system that does not have a
protected left turn phase.

Overall, SimTraffic was best at identifying traffic problems when they exist. It did, hovever,
exaggerate these problems.

REFERENCES

1. Highway Capacity Manual , Special Report 209, Third Edition, Transportation Research
Board, Washington, D.C., 1994

2. TRANSYT 7F Users Guide, Volume 4 , U.S. DEPARTMENT OF

TRANSPORTATION Federal Highway Administration, University of Florida, TRB, Dec 1991
AUTHORS INFORMATION

Walter J. Freeman is a Senior Civil Engineer with Sverdrup Civil Inc., Boston, Massachusetts
02108, Member, ITE

Kien Y. Ho is a Senior Traffic Engineer with Sverdrup Civil Inc., Boston, Massachusetts 02108,
Member, ITE

Elizabeth A. McChesney is a Transportation Engineer with Sverdrup Civil Inc., Boston,

Massachusetts 02108, Member, ASCE