Empirical Evaluation of Transit Signal Priority through Fusion of Heterogeneous Transit and Traffic Signal Data and Novel Performance Measures

Portland State University

PDXScholar

TREC Friday Seminar Series Transportation Research and Education Center (TREC)

3-6-2015

Empirical Evaluation of Transit Signal Priority through Fusion of

Heterogeneous Transit and Traffic Signal Data and Novel

Performance Measures

Wei Feng

Chicago Transit Authority, weifengpdx@gmail.com

Let us know how access to this document benefits you.

Follow this and additional works at:http://pdxscholar.library.pdx.edu/trec_seminar

Part of theTransportation Commons,Urban Studies Commons, and theUrban Studies and Planning Commons

This Book is brought to you for free and open access. It has been accepted for inclusion in TREC Friday Seminar Series by an authorized administrator of PDXScholar. For more information, please contactpdxscholar@pdx.edu.

Recommended Citation

Feng, Wei, "Empirical Evaluation of Transit Signal Priority through Fusion of Heterogeneous Transit and Traffic Signal Data and Novel Performance Measures" (2015).TREC Friday Seminar Series.Book 26.

Empirical Evaluation of Transit Signal Priority

Transportation Seminar March 6, 2015

Wei Feng,

Chicago Transit Authority

Miguel Figliozzi,

Portland State University

Robert Bertini,

Cal Poly State Univ., San Luis Obispo

Background—Transit Signal Priority

2

Passive

Active

Real-time optimal

Green extension

Early Green

Kamila Widulinski and Matthew Lapointe (2013)

Background—Transit Signal Priority

Evaluation

methods

•

Analytic:

Lin (2002); Abdy & Hellinga (2011)

•

Simulation:

Furth & Muller (2000); Dion et al. (2004)

•

Empirical:

Kimpel et al. (2005); Albright & Figliozzi (2012)

Performance

measures

•

Bus travel time

•

Schedule adherence

•

Headway variability

•

Delay for other vehicles

•

Lack of effectiveness and efficiency measures/evaluation

Pre-install

Before / after

Motivation

Unique set of complementary data sources

TriMet

Bus AVL/APC data

City of Portland

SCATS signal phase log data

Intersection vehicle count data

SCATS: Sydney Coordinated Adaptive Traffic System

AVL: Automatic Vehicle Location

APC: Automatic Passenger Count

Research Questions

Current TSP system in Portland:

–

Effectiveness and efficiency?

–

Time savings for buses vs. delay to cross street vehicles

–

Green extension vs. early green phases?

–

Near-side vs. far-side bus stops?

–

Any problems and improvement opportunities?

Study Corridor

21st 26th 33rd 39th 42nd 50th 52nd 65th 69th 71st, 72nd 82nd Milwaukie 12 SCATS signals

SE Powell Blvd.

Bus Route 9

Bus Route 66

Near-side: 26th EB_{26th WB} 33rd EB 42nd EB_{43rd WB} 72nd EB Far-side: 50th EB_{50th WB} 33rd WB 39th EB 39th WB 52nd EB52nd WB 65th EB65th WB 69th EB 71st EB_{72nd WB}

Far-side (12)

Near-side (6)

Stop-to-stop segment

6

Bus stop-to-stop segments

6 near-side segments

12 far-side segments

SCATS Signals

0 20 40 60 80 100 120 140 E W E W E W E W E W E W E W E W E W E W E W E W MKE 21st 26th 33rd 39th 42nd 50th 52nd 65th 69th 71st 72nd Sec onds Intersections / Directions

Median Cycle Length, Green phase and red phase duration

Red Green

Data Integration

Bus ALV/APC

Database

SCATS Vehicle

Count Database

SCATS Signal

Phase Log

Database

Bus Stop-to-Stop

Trip Database

TSP

Performance

Evaluation

9

Bus Stop-to-Stop Trip Attributes

•

Bus departure/arrival time

•

Passenger activities

•

Signal phase start/end time

•

Priority request

•

Upstream/downstream distance

Input data

•

Probability of arriving at

intersection in:

– Green – Red – Green extension – Early green

•

Signal delay

•

Time savings

Output variables

10

Bus Time Saving (Early Green)

𝑑𝑑

₁

Arrival time

Departure time

𝑅𝑅

_𝑗𝑗𝑠𝑠

𝑅𝑅

_𝑗𝑗𝑒𝑒

=

𝐸𝐸𝐸𝐸

_𝑗𝑗𝑠𝑠

𝑎𝑎𝑎𝑎

_𝑖𝑖

𝑑𝑑𝑎𝑎

_𝑖𝑖

𝑅𝑅

_𝑗𝑗+1𝑠𝑠

𝑅𝑅

_𝑗𝑗+1𝑒𝑒

𝑣𝑣

_{𝑚𝑚𝑚𝑚𝑚𝑚}

𝑑𝑑

₂

_𝑣𝑣

𝑚𝑚𝑚𝑚𝑚𝑚

𝑎𝑎

_𝑙𝑙

𝑎𝑎

_𝑟𝑟

𝐸𝐸𝐸𝐸

_𝑗𝑗𝑒𝑒 EB 39th speed (mph) D ens it y 0 5 15 25 35 0. 00 0. 10 0. 20

am

11

Bus Time Saving (Green Extension)

𝑑𝑑

₁

Arrival time

Departure time

𝑅𝑅

_𝑗𝑗𝑠𝑠

𝑅𝑅

_𝑗𝑗𝑒𝑒

𝑎𝑎𝑎𝑎

_𝑖𝑖

𝑑𝑑𝑎𝑎

_𝑖𝑖

𝐸𝐸𝐸𝐸

_𝑗𝑗𝑒𝑒

=

𝑅𝑅

_𝑗𝑗+1𝑠𝑠

=

𝑎𝑎

_𝑟𝑟

𝑅𝑅

_𝑗𝑗+1𝑒𝑒

𝑣𝑣

_{𝑚𝑚𝑚𝑚𝑚𝑚}

𝑑𝑑

₂

𝑎𝑎

_𝑙𝑙

𝐸𝐸𝐸𝐸

_𝑗𝑗𝑠𝑠 12

Key Performance Measures

–

TSP Frequency

–

TSP Effectiveness (for each TSP request)



_{Probability of benefiting from a TSP phase}



_{Expected time saving}

–

TSP Efficiency (for each TSP phase)



_{Probability of being beneficial to a TSP request}



_{Expected time saving per second of TSP phase duration}

TSP Frequency

0 10 20 30 40 50 60 70 80 90 26th 33rd 39th 42nd 50th 52nd 65th 69th 71st 72nd Average number of bus trips per day

requested TSP did not request TSP

0 10 20 30 40 50 60 70 80 90 26th 33rd 39th 42nd 50th 52nd 65th 69th 71st 72nd Average number of TSP phases per day

Green Extension (GE) Early Green (EG)

When A TSP Request Will Benefit from GE/EG

Benefit from

Green Extension

Benefit from

Early Green

Green GE Red-GE

Cycle

Green Red-EG EG

Cycle

15

Potential Results of A TSP Request

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd

on-time EG = Red/Cycle on-time EG = Red/Cycle

on-time GE = GE/Cycle on-time GE = GE/Cycle

ne

ar near near

Actual Outcomes of TSP Requests

TSP

request

No TSP phase

within the cycle

TSP phase

within the cycle

GE

EG

Both GE and EG

Neither GE or EG

GE: G

reen

E

xtension

EG: E

arly

G

reen

Actual Outcomes of TSP Requests

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd

Neither GE nor EG EG Both GE and EG GE

ne

ar near near

TSP Effectiveness

Bus trips that request TSP

Green extension

TSP

request

No TSP phase

within the cycle

TSP phase

within the cycle

a b c

1

3

Early

On-time

Late

3

1

a

b

c

Time

saving

Effectiveness

Benefit

GE or/and EG

2

19

TSP Request Outcomes for GE

a b c

d

ne

ar

ne

ar

ne

ar

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd

d: no GE a: late GE b: on time GE c: early GE

TSP Request Outcomes for EG

a

b c

d

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd

d: no EG a: late EG b: on time EG c: early EG

ne

ar

ne

ar

ne

ar

Actual TSP Effectiveness

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd

on-time EG on-time EG on-time GE on-time GE

ne

ar near near

Ideal TSP Effectiveness

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd

on-time EG = Red/Cycle on-time EG = Red/Cycle

on-time GE = GE/Cycle on-time GE = GE/Cycle

ne

ar near near

Passenger Time Saving per TSP Request

∑

_𝑖𝑖

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑠𝑠𝑎𝑎𝑣𝑣𝑇𝑇𝑠𝑠𝑠𝑠 𝑓𝑓𝑓𝑓𝑓𝑓𝑇𝑇 𝐸𝐸𝐸𝐸

_𝑖𝑖

∑

_𝑖𝑖

𝑇𝑇𝑇𝑇𝑇𝑇 𝑓𝑓𝑇𝑇𝑟𝑟𝑟𝑟𝑇𝑇𝑠𝑠𝑎𝑎

_𝑖𝑖

∑

_𝑖𝑖

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑠𝑠𝑎𝑎𝑣𝑣𝑇𝑇𝑠𝑠𝑠𝑠 𝑓𝑓𝑓𝑓𝑓𝑓𝑇𝑇 𝐸𝐸𝐸𝐸

_𝑖𝑖

∑

_𝑖𝑖

𝑇𝑇𝑇𝑇𝑇𝑇 𝑓𝑓𝑇𝑇𝑟𝑟𝑟𝑟𝑇𝑇𝑠𝑠𝑎𝑎

_𝑖𝑖 0 5 10 15 20 25 39th 42nd 50th 52nd 65th 69th 71st 72nd se co nds from GE from EG 24

TSP Phase Triggered by TSP Requests

TSP

phase

No TSP request

within the cycle

TSP request

within the cycle

EB

WB

Both EB and WB

Neither EB or WB

GE or EG 25

% of GEs Associated to TSP Requests From

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 39th 42nd 50th 52nd 65th 69th 71st 72nd

Neither EB nor WB EB EB and WB WB

% of EGs Associated to TSP Requests From

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 39th 42nd 50th 52nd 65th 69th 71st 72nd

Neither EB nor WB EB EB and WB WB

TSP Efficiency

TSP

phase

No TSP request

within the cycle

TSP request

within the cycle

a b c

1

2

3

Early

On-time

Late

3

2

a

b

c

Time

saving

Efficiency

Benefit

GE or EG

Bus trips that request TSP

Green extension

Actual Green Extension Efficiency

a b c

d

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 39th 42nd 50th 52nd 65th 69th 71st 72nd

out of cycle c: early b: on-time a: late

Actual Early Green Efficiency

a

b c

d

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 39th 42nd 50th 52nd 65th 69th 71st 72nd

out of cycle c: early b: on-time a: late

TSP Efficiency (Time Saving vs. Delay)

TSP

phase

Major street bus and

other vehicles time saving

Minor street other

vehicles delay

Major street other

vehicles time saving

GE or EG

Bus Passenger Time Saving per EG

0 20 40 60 80 100 120 140 EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd se co nds near-side far-side

∑

_𝑗𝑗

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑠𝑠𝑎𝑎𝑣𝑣𝑇𝑇𝑠𝑠𝑠𝑠 𝑓𝑓𝑓𝑓 𝐸𝐸𝐸𝐸

_𝑗𝑗

∑

_𝑗𝑗

𝐸𝐸𝐸𝐸

_𝑗𝑗 32

Bus Passenger Time Saving per GE

0 20 40 60 80 100 120 140 EB WB EB WB EB WB EB WB EB WB EB WB EB WB EB WB 39th 42nd 50th 52nd 65th 69th 71st 72nd se co nds near-side far-side

∑

_𝑗𝑗

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑠𝑠𝑎𝑎𝑣𝑣𝑇𝑇𝑠𝑠𝑠𝑠 𝑓𝑓𝑓𝑓 𝐸𝐸𝐸𝐸

_𝑗𝑗

∑

_𝑗𝑗

𝐸𝐸𝐸𝐸

_𝑗𝑗 33

Vehicle Time Savings and Delay

Green Extension Efficiency

Assume single occupancy vehicles 0 10 20 30 40 50 60 39th 42nd 50th 52nd 65th 69th 71st 72nd se co nds

Minor street vehicle delay Major street vehicle time saving

Major street passenger time saving

Early Green Efficiency

Assume single occupancy vehicles 0 10 20 30 40 50 60 39th 42nd 50th 52nd 65th 69th 71st 72nd se co nds

Minor street vehicle delay Major street vehicle time saving

Major street passenger time saving

Summary of Findings

Green extension

•

_•

Too many late green extension phases

_{Time savings}

_≈

_Delay

Early green

•

Time savings > Delay

TSP performance

•

_•

Vary significantly across intersections

_{Big gap between actual and ideal performance}

Conclusions

•

Proposed TSP performance measures can help identify

problems/improvement opportunities and support

planning decisions

•

Findings from this study may be site-specific, but the

methodologies are transferable to other corridors/cities

•

TSP effectiveness and efficiency can be greatly affected by

control logic, parameter calibration and signal

detection/communication reliability

Future Work

•

Consider vehicle queuing effect when estimating bus

arrival time probabilities at intersections

•

Utilize new and higher resolution data such as:

–

5-second bus AVL data (finer bus trajectory between bus stops)

–

TSP Optical detector log data (priority log in/out records)

Acknowledgements

Steve Callas

David Crout

Peter Koonce

Willie Rotich

40

Questions?

On Average

TSP request

On-time

No TSP phase

within a cycle

Within a cycle

but early

Within a cycle

but late

GE

EG

1.5%

10%

2.5%

5%

25%

1%

55%

Bus time saving

0.3s

0.5s

Passenger time saving

7.5s

10s

= 100%

Actual

Ideal

GE

EG

6%

27%

0%

0%

0%

0%

67%

= 100%

42

On Average

TSP phase

On-time

No TSP request within a cycle

Early

Late

GE

EG

5%

40%

3%

30%

64%

8%

Bus passenger time savings

20s

90s

28%

22%

Duration

7s

11s

Major street vehicle time savings

Minor street vehicle delay

60s

300s

80s

200s

=100%

=100%

Actual

Ideal

100%

0%

0%

0%

GE or EG

=100%

43