Evaluating InSync Performance in Microsimulation

1. Evaluating InSync Performance in Microsimulation Aleksandar Stevanovic, PhD, PE Florida Atlantic University Transpo 2012 Bonita Springs, FL October 29, 2012

2. Why Simulating InSync? • Comprehensive evaluation of its performance • Controlled experiments • Abundance of outputs • Better understanding of pros & cons • Repeatability and comparability

3. Originality of this Study • No comprehensive evaluation of InSync in microsimulation • Regular operations modeled with no attention to special conditions/situations • Evaluation of InSync for regular operations (AM, MD, and PM) + • Freeway Incident • Rail Preemption • Extra Demand • Inclement Weather

4. About SR 421 - Dunlawton Ave • 12 signalized intersections (11 analyzed) • Divided 4-6 lane facility • Left turns • RR crossing & school zones • 35-50 mph speed limits

5. Why SR421 in Volusia County? • A common (sub)urban corridor • Difficult to retime • Seasonal and annual variations in traffic flows • Evacuation route

6. Data Collection for VISSIM Model • Turning movement counts • Travel times along mainline (GPS & video) • Signal timing sheets & Synchro files • Other data – RR gate timings, school zone hours, etc.

7. Google Map as Background

8. NEMA-like Ring-Barrier Controllers

9. Calibration • Making sure that modeled and simulated volumes are equal • Volume balancing • Sources and sinks • Saturation flow rates • Manual process

10. Calibration Details – Nova Rd

11. Calibration Results

12. Validation • Verifying that segment travel times from model and field are comparable • Speed distributions • Acceleration & deceleration • Offsets & detectors • Understand limitations

13. Validation Results AM Peak MD Peak PM Peak

14. TOD Signal Timings

15. InSync – Controllers Run Free • 1) Disable all coordination (set controller into “free/uncoordinated” mode) • 2) Enable Detector Diagnostic Failure Mode • a. Set On failure to 250 minutes • b. Set Off failure to 5 minutes • i) Diagnostic failure mode on a 170 will allow itself to come out of failure. • 3) Set all “Minimum Green” times to 5 sec., or as client feels comfortable • 4) Leave “Maximum Green” times as before • 5) Set “Passage Gap” or “Observed Gap” to 1 sec. • 6) For protected/permitted left turns, omit the left turn call when the opposing thru • movement is green. • • Notes: • o This will prevent a yellow trap if the controller receives calls on • permitted left turn phases when the opposing thru is green, but may • not bring on the adjacent green thru movement to allow the left turn • a permitted movement. • 7) Enable “Soft Recall” on the mainline phases • 8) Disable “Yellow Lock” and “Red Lock” detector locking • 9) Set all “Detector Delays” to 0 sec. • 10) Disable all recalls: Max, Min, Hard, Vehicle, Phase, etc. • 11) Remove “Extensions” • 12) Disable “Anti-Backup” or “Left Turn Trap” • 13) Enable “Max Recall Inhibit”

16. InSync Detectors in VISSIM

17. InSync-VISSIM Interface

18. Field-like InSync Dialog Box If a picture is used, use this layout. Pics are much preferred over bullets or other text.

19. Simulations • Four signal timing scenarios - Identical traffic flows and geometric conditions • Field TOD • Optimal TOD – single section • Optimal TOD – multiple sections • InSyncAdaptive – no additional adjustments • 10 random seeds • Three peak periods (AM, MD, and PM) • 2.5 hours (15 min for warm up and 15 min for cool off) for each peak period

20. Operational Scenarios • Regular operations - 3 peak periods (AM, MD, and PM) • Special operations (only for PM peak) • Freeway Incident - Traffic diversion due to a freeway incident • Rail Preemption - Impact of disruption caused by freight train operations • Extra demand – 20% traffic growth (flat) • Inclement Weather - Reduction in speed and sat. flows due to a heavy rain

21. Freeway Incident • An 1-hour incident on I-95 freeway between SR 421 and SR 400 • One lane closed – traffic diverts to SR 421 • AADT ~ 50,000 veh/day; k factor ~ 15%; d factor ~ 55%; 4 lanes • 1,000 veh – diverted to SR 421 during one hour

22. Diversion Routes 300 300 Incident 200 200 1000 vph

23. Rail Preemption • 4 trains modeled in 2 PM peak hours (every 30 mins) • Anywhere between 20 – 100 cars in a train (modeled 20, 30, 70, and 100) • Train’s speed distributed around 45 mph • Gates take about 7 seconds to go up/down

24. SR 421 & FEC Railway FEC RR

25. Extra Demand • Flat increase of 20% for all of the traffic demand generators • Turning movement proportions remain the same • Impact on intersections – almost all intersections worsen LOS by one grade • Oversaturated network (8 ints. at LOS D or worse; 4 ints. at LOS E or worse)

26. Inclement Weather • Inclement weather conditions (heavy rain) in Florida • PM Peak traffic demand unaltered (worst-case scenario) • Travelling speeds reduced by ~ 15% (from - 20% to -10%) • Saturation flows reduced by ~ 20% (~ ~ 1500 vph)

27. What was Evaluated? • Intersection performance • Main-corridor travel times • Main street vs. Side street • Network performance

28. Results Examples

29. Intersection Performance

30. AM Peak – Intersection Delay

31. MD Peak – Number of Stops

32. PM Peak – Average Queue

33. ED – Level of Service

34. Main-Corridor Travel Times

35. IW Travel Times EB

36. IW Travel Times WB

37. IW Travel Times All

38. IW Travel Times - Summary Percent improvement

39. Main Street vs. Side Street

40. RP Main vs. Side Street T & L

41. RP Main vs. Side Street

42. Network Performance

43. FI Total Network Delay

44. FI Total Number of Stops

45. FI Total Travel Time

46. FI Network Summary

47. Summary of Results

48. Reduction [%] of Delays & Stops Average improvements: Delay ~ 22%; Stops ~ 15 %

49. Summary – All Scenarios InSync better than a respective TOD signal timing Results are not definite or TOD is better

50. Questions, feedback, comments? astevano@fau.edu