Signalized Intersection Performance Measurement

1. Signalized Intersection Performance Measurement Vendor Perspectives Regarding INDOT/Purdue/Vendor Collaborative Project Eric Raamot Vice President, Engineering Econolite eraamot@econolite.com Presented by: Gary Duncan Sr. Vice President and Chief Technology Officer Econolite gduncan@econolite.com

2. Discussion Points Brief summary of historic intersection performance measurement Overview of the successful Academic/DOT/Industry collaboration used to research enhancements on performance measurement Outline of Product enhancements derived from research

3. Historical Challenges • Typical field data collection has been limited to: • Volume • Occupancy • Average speed • Data averaged over relatively large aggregation periods • Typically 5 - 60 minutes • Controllers have provided some basic MOE logging: • Phase split duration • Reason for termination (Max/Gap/Force Off) • Pedestrian actuation • Transition begin/end • Preemption begin/end

4. Example of Current MOE’s

5. Historical Challenges Gap exists between collection of “data” and generation of “information” Translation from data into information has typically been left up to the traffic engineer. Limitations within controllers based upon size of controller memory and low-bandwidth communications infrastructure have limited data collection in the past. Current technology opens the door for more extensive logging as well a conversion of data points into useful traffic engineering information.

6. Agency Resources Many agencies do not have the resources to retime all intersections every 1-3 years. Little system-level information exists to identify intersections operating with outdated or problem timing. Challenge is to use capabilities of today’s controllers and systems to create tools that reveal opportunities for performance improvement.

7. Research Chronology INDOT /Purdue Instrumented Intersections (2003) Econolite ASC/3 Data Logger (2006) …

8. ASC/3 Data Logger Byte # N Byte # N+1 Byte #’s N+2,N+3 Event Type Assignment TimeStamp . 0 = Begin Phase Off (1-16) Phase #1-16 Deciseconds past hour (0-35999) 1 = Begin Phase Green (1-16) Phase #1-16 Deciseconds past hour (0-35999) 2 = Begin Phase Yellow (1-16) Phase #1-16 Deciseconds past hour (0-35999) 3 = Begin Phase Red Clear (1-16) Phase #1-16 Deciseconds past hour (0-35999) 4 = Begin Ped Off (1-16) Phase #1-16 Deciseconds past hour (0-35999) 5 = Begin Ped Walk (1-16) Phase #1-16 Deciseconds past hour (0-35999) 6 = Begin Ped Clear (1-16) Phase #1-16 Deciseconds past hour (0-35999) 7 = Reserved (Ped State) (1-16) Phase #1-16 Deciseconds past hour (0-35999) 8 = Detector Transition Off (1-64) Detector #1-64 Deciseconds past hour (0-35999) 9 = Detector Transition On (1-64) Detector #1-64 Deciseconds past hour (0-35999) 10 = Detector RFU (1-64) Detector #1-64 Deciseconds past hour (0-35999) 11 = Detector RFU (1-64) Detector #1-64 Deciseconds past hour (0-35999) 12 = Begin Overlap Off (1-16) Overlap #A-P Deciseconds past hour (0-35999) 13 = Begin Overlap Green (1-16) Overlap #A-P Deciseconds past hour (0-35999) 14 = Begin Ovl Grn Extension (1-16) Overlap #A-P Deciseconds past hour (0-35999) 15 = Begin Overlap Yellow (1-16) Overlap #A-P Deciseconds past hour (0-35999) 16 = Begin Overlap Red Clr (1-16) Overlap #A-P Deciseconds past hour (0-35999) 20 = Begin Preemptor (1-10) Preemption #1-10 Deciseconds past hour (0-35999) 21 = End Preemptor (1-10) Preemption #1-10 Deciseconds past hour (0-35999) 24 = Phase Hold Applied (1-16) Phase #1-16 Deciseconds past hour (0-35999) 25 = Phase Hold Released (1-16) Phase #1-16 Deciseconds past hour (0-35999) 26 = Ped Call on Phase (1-16) Phase #1-16 Deciseconds past hour (0-35999) 27 = Ped Call Cleared (1-16) Phase #1-16 Deciseconds past hour (0-35999) 28-31 = Phase Control RFU (1-16) Phase #1-16 Deciseconds past hour (0-35999) 32 = Phase Min Complete (1-16) Phase #1-16 Deciseconds past hour (0-35999) 33 = Phase Term Gap Out (1-16) Phase #1-16 Deciseconds past hour (0-35999) 34 = Phase Term Max Out (1-16) Phase #1-16 Deciseconds past hour (0-35999) 35 = Phase Term Force Off (1-16) Phase #1-16 Deciseconds past hour (0-35999) 36-39 = Phase State RFU (1-16) Phase #1-16 Deciseconds past hour (0-35999) 40 = Coord Pattern Change (0-253,Free,Flash) Deciseconds past hour (0-35999) 41 = Cycle Length Change (0-255,seconds) Deciseconds past hour (0-35999) 42 = Offset Length Change (0-255,seconds) Deciseconds past hour (0-35999) 43-59 = Split (1-16) Change (0-255) Split Value Deciseconds past hour (0-35999) 60-62 = Coord Data RFU (0-255) Deciseconds past hour (0-35999) 100msec time-stamping of all relevant controller events. Event data stored as binary-formatted files, hourly on controller. Controller storage allows for > 24 hours of raw data. Data uploaded via scheduled FTP and uncompressed into SQL database.

9. Research Opportunities Elkhart County(local agency) Noblesville(Corridor Performance Measures) West Lafayette(Pedestrians) McCordsville(Railroad pmt)

10. Research Direction Research intent is to define a set of signalized intersection operating parameters that may reveal inefficiencies in cycle, offset and/or split timing. Identification of inefficiencies will help agencies focus retiming efforts on those intersections with the most observed need. INDOT and Industry intent to is to generalize these parameters such that commercialized systems can provide traffic engineers with a common toolset. Project focus upon measurement and reporting of performance data. Control level changes at intersection not addressed within the scope of project.

11. DATA West Lafayette Ped Int. Indianapolis RR Preempt 2 Int. in ElkhartCounty 4 Int. Indianapolis Arterial ~ 160 mil.Events in SQL DB

12. P1 P2 P3 P4 1500 0 Equivalent Hourly Flow Rate (veh/h) P6 P5 P7 P8 1500 0 0:00 12:00 24:00 0:00 12:00 24:00 0:00 12:00 24:00 0:00 12:00 24:00 Time of Day 750 750 Volume INFORMATION 24 Flow Rates by phase/by cycle

13. P1 P2 P3 P4 1.0 0.0 Volume-to-Capacity Ratio P6 P5 P7 P8 1.0 0.0 0:00 12:00 24:00 0:00 12:00 24:00 0:00 12:00 24:00 0:00 12:00 24:00 Time of Day 0.5 0.5 Capacity INFORMATION V/C Ratios by Phase, 24 Hours

14. Progression INFORMATION

15. Interim Accomplishments • Expansion and consistency of traffic controller raw data logging format. • Presented as CSV format at system level, controller encoding remains proprietary. • PEEK and Econolite have currently provided this CSV interface. • Interim format used by partner vendors for research interoperability. • Development of recommendations for performance based data aggregation and metric logging within controller. • NTCIP 1202 object definitions are an anticipated output from this effort. • Reuse of ACS-Lite status objects as log file being considered.

16. Research Toolset Econolite has supported research through integration of research tools into Centracs system:

17. Evolution of Researchto Product INDOT /Purdue Instrumented Intersections (2003) ~2009Automated SQL Ingestion and Purdue Parameterized Excel Sheets

18. Product Enhancements Difference between what has been developed as research tools for the various corridor studies to date with a generally applicable toolset that can be used daily by the practicing traffic engineer. Econolite has started software development in both Centracs and the ASC/3 controller to generalize the results of the research to date into core product enhancements.

19. “Productization” Requirements: • Aggregation of raw controller events into a cycle-cycle logging of arrival distributions using ACS Lite data objects. • Aggregation reduces data storage/bandwidth requirements to allow broad deployment. • Periodic upload of arrival distribution log into Centracs SQL database. • Supports query of current datasets and historic trends across filterable displays and reports. • Centracs generation of graphical tools that present this data as information in a manner usable by traffic engineers to make retiming decisions. • Enhanced PCD, Split analysis (V/C), Plan Performance graphs.

20. Concluding Observations This research has been a successful collaboration of academia, agency, and vendors to standardize conceptual performance measures that assess the need for COS timing adjustments. Vendors are taking results and enhancing controller and system software to support the performance measurements and informational displays developed. Collaborative project efforts expected to continue for at least an additional year with definition of further performance measures. Vendors carry forward the goal for systems to automatically determine the degradation of performance and suggest remedial action to traffic engineers.

21. Contact Information: Gary Duncan gduncan@econolite.com Eric Raamot eraamot@econolite.com Reference: Smaglik E.J., A. Sharma, D.M. Bullock, J.R. Sturdevant, and G. Duncan, “Event-Based Data Collection for Generating Actuated Controller Performance Measures," Transportation Research Record, #2035, TRB, National Research Council, Washington, DC, pp.97-106, 2007. Brennan, T.M., C.M. Day, J.R. Sturdevant, E. Raamot, and D.M. Bullock, “Railroad Preempted Intersection Track Clearance Performance Measures,” Transportation Research Board Paper ID:10-0118, in press.

22. Research Opportunities Elkhart County(local agency) Noblesville(Corridor Performance Measures) West Lafayette(Pedestrians) McCordsville(Railroad pmt)