Traffic Signal Automatic Performance Measures Jamie Mackey, P.E. Statewide Signal Engineer UDOT Traffic Operations Center

July 22, 2013 TRB Traffic Signal Systems Committee Mid-year meeting

Implementing Performance Metrics   

Working closely with Professor Darcy Bullock (Purdue) and Jim Sturdevant (INDOT) Written in-house by UDOT programmers (C#) Using ASC/3 high-resolution data logger  



Uploaded every 15 minutes or on demand Storing 10-12 GB per day

Worked with local radar vendor to enhance detection for counting

Performance Metrics Goals 

Transparency and Unrestricted Access 



Access for everyone      



No Special Software – No Passwords – No Firewalls All of UDOT: (TMD, T&S, Region Traffic, Senior Leaders, Planning) Consultants Academia MPO’s Other jurisdictions Public (via dashboard)

All signal partners are invited to add their signals

http://udottraffic.utah.gov/signalperformancemetrics/

UDOT Performance Metrics Website

Map of Signals with Performance Metrics Salt Lake Valley

Performance Metrics Requirements  

Communication to traffic signals Signal controllers with a high-resolution (0.1 sec) data logger    



Server for data storage and processing 



Econolite: ASC/3 (App 2.50+), Cobalt Intelight PEEK & Siemens – not yet Retrieve data with FTP connection Microsoft SQL ($15K, 5TB, 8GB RAM, 24 3GHz processors)

Count detector (optional)

Setback Count Detectors    



Wavetronix Advance Used to timestamp vehicle arrivals 10’ count zone placed ~350’ behind stop bar No additional expense if already in place for dilemma zones Count zone May undercount dense traffic

Stop Bar Count Detectors     

Wavetronix Matrix Used for turning movement counts Lane-by-lane detection zones in front of stop bar Requires detection rack card for every two zones ($$$$$$) Wavetronix is expected to release a new high-capacity detector BIU (fall 2013)

Count zones

Current Metrics & Required Detection Metric

Detection Needed

Purdue Coordination Diagram

Setback count

Approach Volume

Setback count

Approach Speed

Setback count w/ speed

Phase Termination Split Monitor Turning Movement Counts

Stop bar count

Approach Delay (new)

Setback count

Arrivals on Red (new)

Setback count

Current Metrics 800 700

Number of Signals

600 500 400 300 200 100 0 PCD Approach Delay

Speed

Phase Termination Split Monitor

TMC

Performance Metrics Uses 

Daily Operations      



Reporting (future)  



Basic parameters Coordination Detection problems Complaint response/troubleshooting Events, Incidents, Weather, & Construction Alerts (future) Prioritize signal needs Communicate system status to region/senior leaders and public

Modeling/planning

Complaints: Light is Red Too Long – Before Metric: Phase Termination Minor street through & left max out at night only

Gapout Max out

Pedestrian activation (shown above phase line)

Force off

Skip

Complaints: Light is Red Too Long – Before Metric: Split Monitor Northbound Left (Minor)

Minor side street left turn maxes out during the middle of the night. Green for about 30 seconds.

Westbound Through (Major)

Major movement only green for 20 seconds.

Complaints: Light is Red Too Long – After Metric: Phase Termination

Complaints: Light is Red Too Long – After Metric: Split Monitor Northbound Left (Minor)

Phase 7 rarely comes up.

Westbound Through (Major)

Phase 2 rests in green.

Weather: Snow Metric: Approach Speed

Progression speed for snow plans?

Snow storm starts

Coordination: Poor Progression Metric: Purdue Coordination Diagram

Platoon arrives on red

Platoon arrives on green

Coordination: Queue Clearance Metric: Purdue Coordination Diagram

Queue clearing End of platoon cut off

Queue clears before platoon arrives.Vehicles maintain speed through lane drop. Queuing reduced because platoon not cut off

Basic Parameters: Gap/Passage Time Metric: Split Monitor BEFORE: Phase usually gaps out (87% of cycles) CHANGE: Increase passage time AFTER: Phase uses all its time (65% of cycles)

Coordination: Modeling Metric: Turning Movement Counts

Future of UDOT Performance Measures 

High-level performance measures for Senior Leaders & public 

  

Are things getting better, worse, or staying the same?

Travel Time Frequency (INRIX data) Alerts for detector health and operational problems Automatic offset optimization (link pivot)

Vendors: High-resolution Data Logger Enumerations will be a requirement for signal controllers

Future of UDOT Performance Measures 

Pooled Fund Study: "Traffic Signal Systems Operations and Management"   

 

Led by Darcy Bullock and Indiana DOT 3 years (2012-2014); 9 States & Chicago Develop standards for data logger & metrics

UDOT Research will be funding a university study to determine the accuracy of and calibrate count detectors UDOT wins AASHTO Technology Implementation Group (TIG) award – June 2013 

“Automated Traffic Signal Performance Measures” becomes a Focus Technology

Questions Jamie Mackey Statewide Signal Engineer [email protected]

www.udottraffic.utah.gov/ signalperformancemetrics

Utah Traffic Signals with Communication 1400

Number of Traffic Signals

1200

1966 Traffic Signals statewide 60% are operated by UDOT

UDOT

76% have communication

1000 800 600 400 200 0

Cities & Counties

All signals statewide are connected to the same central system

Utah Traffic Signals with Performance Metrics 1400

Number of Traffic Signals

1200

67% of UDOT’s connected signals have performance metrics

UDOT

1000 800 600 400 200 0

Cities & Counties

Purdue Coordination Diagram Bangerter & 5400 S (3/7/2013) Good Offset

Bad Offset

Coordination: Progression Type Metric: Approach Volume One-way Semi-balanced

Balanced

Coordination: Progression Type Metric: Approach Volume One-way Semi-balanced

Balanced

Coordination: Actuated Coordinated Metric: Split Monitor

Events: Concert Metric: Split Monitor



Existing timing plan working well



Event plan start & stop times OK



Added minimum recall for EB

Event Pattern

Basic Parameters: Yellow & Red Intervals Metric: Approach Speed

Posted speed

Use 85th percentile to set yellow & red clearance intervals

Coordination: Modeling Metric: Turning Movement Counts

Detection Problems: Constant Call Metric: Phase Termination Minor street maxes out all night and day

Gapout Max out

Pedestrian activation (shown above phase line)

Force off

Skip

Purdue Coordination Diagram Bangerter & 5400 S, NB (3/21/2013)

One cycle length

Time in cycle

Signal turns yellow, then red Signal turns green

Vehicle arrival Time of day

Purdue Coordination Diagram Bangerter & 5400 S, NB (3/21/2013) This movement receives 45% of the green time in pattern 7 and 77% of vehicles arrive on green.

Vehicles arriving on green

Vehicles arriving on red (upstream left turns?)