RT-TRACS A daptive Control Algorithms VFC-OPAC Farhad Pooran PB Farradyne Inc.

1. RT-TRACS Adaptive Control Algorithms VFC-OPAC Farhad Pooran PB Farradyne Inc. TRB A3A18 Mid-Year Meeting and Adaptive Control Workshop July 12-14, 1998 Pacific Grove, CA

2. VFC-OPAC (Virtual Fixed Cycle OPAC) • Real-time, traffic adaptive control of signals in a network • Distributed optimization based on the OPAC (Optimization Policies for Adaptive Control) smart controller • Multi-layer network control architecture • Variable cycle in time and in Space

3. VFC-OPAC Development History • OPAC I: Dynamic Programming optimization • infinite horizon (single intersection) • OPAC II: optimal sequential constrained search procedure • finite projection horizon length • OPAC III: rolling horizon approach • real-time implementation • OPAC IV (VFC-OPAC): network model for real-time • traffic-adaptive control

4. Control Layers in VFC-OPAC

5. Control Layers in VFC-OPAC • Layer 1: Local Intersection Control Layer (phase length) - Optimal switching sequences for projection horizon, subject to virtual fixed cycle constraints • Layer 2: Coordination Layer - Real-time optimization of offsets at each intersection • Layer 3: Signal synchronization - network wide calculation of virtual fixed cycle

6. VFC-OPAC Network Module

7. Data Requirements • Ideal detector location is about 10 seconds upstream of stop line (at free flow speed) or upstream of the worst queue on each lane of all through phases. • One count detector in each lane of left turn pockets as far upstream as possible • Automatic compensation for ‘bad’ detectors • Volume, occupancy, and speed measured in the field

8. Control Variables • OPAC optimizes (minimizes) a weighted performance function of total intersection stopped delay and stops subject to minimum and maximum green times • Under coordination, signal timings are also constrained by the current cycle length • Current Counts, Occupancy, and Speed (measured or calculated)

9. Decision Variables • Terminate the current phase in ring 1 (Yes or No) • Terminate the current phase in ring 2 (Yes or No)

10. State Variables • Signal status • Elapsed time since last signal status change • Standing queues • Cumulative delay • Cumulative stops

11. Constraints on Decision Variables • Phase interval timings (minimum green, maximum green, yellow, all red, walk and don’t walk) • Opposing demand (vehicle and pedestrian calls) • Cycle length constraints • Offset adjustments

12. How Are Flow Profiles Developed? • Upstream detectors can provide an actual history for a short portion of the profile. • Smoothed volume can be used for uniform profiles. • Platoon identification and smoothing can be used for cyclic profiles. • Adjustments can be made to eliminate double counting (left turn phases).

13. How Are Flow Profiles Developed? • Upstream detectors can provide an actual history for a short portion of the profile.

14. Cycle Length Optimization • Meet phase switching timing determined by local conditions, while maintaining a capability for coordination with adjacent intersections • Using a cycle length constraint, the cycle length can start or terminate only within a prescribed range • All VFC-OPAC controlled intersections can oscillate with a common frequency

15. Offset Optimization Options: • Leave current offset ( zero change) • Move right one interval (+2 sec) • Move left one interval (-2 sec)

16. Data Sampling • Develops a flow profile for each phase using a user-specified time interval • Head of the profile is actual counts from the recent past. • The tail of the profile is projected for the future using smoothed volume • Smoothed data: volume, occupancy, speed, platoon headways, flow profiles, and phase duration

17. MOE’s • Volume, occupancy, speed by detector and phase. • Estimated measure of queue, delay, and stops by phase.

18. Phasing Flexibility • Supports 8 phases in a dual ring configuration • Does not explicitly control phase sequence • Can recognize and adapt to changes in sequence immediately

19. System Architecture • Isolated intersection control - fully distributed • Coordinated system control - basically distributed except for the following tasks: • cycle length determination is made at central and communicated periodically to the intersection controller • peer-to-peer information is communicated through central on a periodic basis (if adjacent intersection controllers are not linked physically)

20. Hardware Requirements • Local Controller: • a computer board with a floating point processor and 4 MB memory (e.g., 68040 or 68060 boards) • Central: • 3 to 4 PC’s for OI, Server, dB, Device Drivers and Communications with at least 2 GB of HD and 64 MB RAM

21. Communication Requirements • Communications with Central: OPAC status is polled • Communications with Signal Control Software • Peer-to-peer communications

22. Network Type • For coordinated signal control, cycle lengths are calculated for user-specified groups (sections) of signals (arterials or networks) • The cycle is calculated using the critical v/c ratios of the critical intersection in the section • The field computer optimizes offsets with its neighbors, not the entire section.

23. Special Features • Preemption: • Preemption will always take priority over OPAC. • Prioritizes transit and emergency vehicles if they are restricted to particular lanes • Recovers from a preemption immediately

24. Special Features • Oversaturated Conditions: • Isolated intersection control - OPAC will provide maximum green to the affected phase(s) if occupancy on the OPAC detectors exceeds a user-specified threshold. • Coordinated control • Provide maximum green to congested phases, subject to the current cycle length • Adjust cycle lengths in response to increasing congestion

25. Field Installations • 1986 - Isolated OPAC in Arlington, Virginia and Tucson, Arizona (Single intersection control) • 1996 - Isolated OPAC at a Route 18 site in New Jersey (15-intersection arterial) • 1998 - Coordinated OPAC at Reston Pkwy site, in Reston, Virginia (16-intersection arterial )