1. 3rd Street Light Rail Process and Challenges of Developing Transit Signal Priority Javad Mirabdal, Jack Fleck & Britt Thesen Department of Parking and Traffic City and County of San Francisco October 22, 2003

2. Outline  Project Background –Existing Conditions –General Project features –Timeline and Goals  Transit Priority –Software selection process –Vissim Modeling –Results  Lessons Learned

3. Existing Conditions  5.4 mile corridor  35 signalized intersections  3 traffic lanes in each direction with parking  Serviced by MUNI’s 15 Bus line – 7 to 15 minute headways –Current ridership: 25,000

4. General Project Features – Phase 1  65 signalized intersections  19 stations  2 traffic lanes in each direction with left turn pockets at some locations  Most blocks have parking, with parking removed near stations and left turn pockets  Exclusive Right of Way except Bayview Commercial Core and 4 th St. Bridge  Transit Signal Priority & Preemption  Funding: $520 million, Local Sales Tax

5. Intial Phase  New Central Subway to Chinatown  Funded by Federal Match  Currently in Conceptual Engineering  4 Underground Stations General Project Features – Phase 2  New Central Subway to Chinatown  Funded by Federal Match  Currently in Conceptual Engineering  4 Underground Stations

6. Project Goals  Improve Transit Service along Corridor  Reduce travel time and delay  Increase ridership  Connect Southeast SF to Downtown  Transit Oriented Development  Economic Revitalization

7.  Existing  2005  3 rd Street LR  2015 3 rd St. LR & Central Subway Daily Ridership  25,000 71,000  95,000 Travel time to 3 rd /Market from: Visitation Valley (Arleta/Bayshore) Bayview Core (Third/Palou) Mission Bay 36 min 26 min  (N/A) 31 min  23 min 14 min  27 min 19 min 10 min Transit Ridership & Travel Times

8. Construction Schedule Segment B: Mission BayFebruary 2004 Segment C: Central WaterfrontMarch 2005 Segment D: Bayview/Hunters PointFebruary 2005 Segment E: Bayview South & US101 OverpassFebruary 2005 Segment F: Visitacion ValleyMay 2004 Segment J: Platform Finishes & Traffic Signal Priority System/VETAG Controllers May 2005

9. TSP Elements  2070 (TS2) Signal Controllers with Transit Preemption/Priority capabilities  Fiber Optic Cable for Communication  Linked to Central Traffic Management Center (SFgo)  Vehicle Detection System: Video (Autoscope)  LRV Detection System: VETAG

10. Transit Priority Vision  Trains travel station to station without stopping!  Progression on 3 rd Street essentially maintained  No phase waits more than 1 cycle (90 sec)  Side Streets, Left turns and Pedestrians are actuated  Pedestrians receive enough time to cross street at 4 ft/sec, or 2.5 ft/sec if phase is actuated

11. Transit Priority Strategies  Skip Phases  Extend Phases  Shorten Side Street Phases  Lead/Lag (phase rotation)  Early Green  Get Back in Step within 1 Cycle

12. Software Selection Challenges  Meet technical requirements city-wide implementation (more than 2 rings, communication over ethernet, etc.)  Transit Priority and Preemption Features  Easy to learn, implement and maintain  Good technical support

13. Software Wish List  General Features  Pedestrian Features  Transit Priority  General Preemption  Cable Car Preemption  Coordination  Communications (full list available upon request)

14. Evaluated options Narrowed field down to –NextPhase (4-ring software) –VS-Plus (Matrix software, no rings or barriers) Modeled softwares in VISSIM for comparison Software Selection Process

15. Software Selection Process: VISSIM Model  Micro-level simulation  Runs virtual signal controllers for each intersection  Very time consuming to build model but…  Powerful modeling capabilities - pedestrians, bikes, right turn conflicts, trains, passenger loading, etc.

16. VISSIM Modeling Process  Test section of 7 intersections  Nextphase, VS-PLUS, and fixed time conditions  Modeled different train headways  Multiple runs of each model with random arrivals, driver behavior, etc. to get more realistic results Cesar Chavez Evans Cargo

17. Video Clip Video clip

18. Early Green for SB Train NB Train NB Check-Out Call SB Train SB Check-In Call SB Check-Out Call NB Advance Call NB Check-In Call SB Advance Call Virtual signal timing and vehicle detection Return to Main Street to Accommodate NB Train Left Turn Phase Main Street GreenCross Street Phase

19. Model MOE’s Collected data on Measures of Effectiveness (MOE’s) –Travel time through test section –Delay through test section –Travel time for each movement (including cross streets and left turns) –Delay for each movement Queue length –Priority Strategies

20. Modeling Results – Travel Time Through Test Section Movement Fixed time optimized for Cars Fixed time optimized for TrainsNextphaseVS-Plus NB Cars210237172165 NB Trains247243211155 SB Cars319330181155 SB Trains350324197157

21. Modeling Results  Both Systems are very complicated!  Results showed that VS-Plus was better for TSP, but Nextphase was better at reducing delay to vehicles overall.  However, level of TSP/vehicle delay could be adjusted for either software.  We initially selected Nextphase because we already had trained technicians in software.

22. Where are we now?  Unsatisfied with user interface for transit priority and program complexity of Nextphase  Reconsidering software options VS-Plus 2-ring software programs –Naztec –Sepac –City of Los Angeles –Econolite Newly developed software

23. Lessons Learned  We have a vision, but getting there isn’t easy…  Modeling cannot capture complexity of programming  Must thoroughly learn the TPS software before selecting  Need to find balance between competing elements  Transit Priority  Pedestrians  Traffic Circulation  General Safety  Complexity of Programming