4-Adaptive_Signal_Control_-_How_Does_It_Work.pdf
- 1. Adaptive Traffic Signals Overview Kevin Fehon, P.E., PTOE Principal DKS Associates
- 2. 2 What is “adaptive” anyway? Signal timing is adapted to the measured traffic in real time Traffic is measured and/or predicted, calculations are made and timing is implemented “Adaptive” systems do not pick the best stored plan to match the measured traffic – that is “Traffic Responsive” (TRPS)
- 3. Why Bother with Adaptive Control? Pre-stored plans accommodate variations in traffic, therefore never optimal Time of day selection also sub-optimal Pre-stored plans age quickly TRPS difficult to optimize, most successful on arterials, often abandoned 1.5 Generation updates plans but suffers from shortcomings of TRPS
- 4. Performance Index Performance of fixed time plans deteriorates
- 5. Performance of adaptive systems more stable 5 Time in YEARS -15% -10% -5% 0% +5% +10% 1 2 4 5 3 Adaptive Effectiveness of Fixed Timing Plans Fixed timing plan Performance Level
- 6. CAN ADAPTIVE SYSTEMS REALLY IMPROVE TRAFFIC OPERATIONS?
- 7. Gresham PM Peak TT (Eastbound Burnside) 405 368 424 393 349 373 314 0 100 200 300 400 500 Travel Time (sec) 1997 Free 1998TOD 2004 Free 2004 (1998TOD) 2004TOD 2007 (2004TOD) 2007 SCATS Legend
- 8. 8 Adaptive Systems Respond to Significant Changes in Demand
- 10. Types of algorithms Sequence based Use a cycle length, like most coordinated signal systems Have a pre-defined sequence of phases (some with flexibility) Non sequence based Do not use a cycle length Do not use a pre-defined sequence of phases
- 11. Types of systems Complete, stand-alone, with full management system capabilities Module within proprietary signal management system External to proprietary signal management system
- 12. Examples of current systems Complete, stand-alone SCATS, SCOOT, LA ATCS Modules for proprietary system Synchro Green (TrafficWare/Naztec), Centracs Adaptive, McCain QuicTrac, Intelight, OPAC (MIST) External ACS-Lite, InSync, Rhodes
- 13. Various architectures Centralized All strategic and tactical decisions at central Distributed Strategic at central, tactical at local Peer-to-peer No central supervisor
- 14. How do they work? Implement one or more operational strategy (See FHWA guidebook) Pipeline to maximize throughput (appropriate with oversaturation) Pipeline to provide smooth flow (appropriate with undersaturation) Equitably distribute green times (appropriate with many turning movements) Manage queues
- 15. Fixed or variable objective function?
- 16. A FEW EXAMPLES…
- 17. InSync Theory Seeks to minimize a weighted delay objective function at one critical intersection Picks next phase (state) that will minimize short term delay function Inserts non-coordinated phases at other intersections outside platoon bands (tunnels) Practical limitations Maximum wait times, overlapping tunnels, driver expectations Effectively runs fixed cycles at critical intersection Best with short sections of arterial
- 19. Success depends on spacing & speed
- 20. ACS-Lite Theory Start with a standardTOD pattern Modify offsets based on detection of arrivals during green Modify phase splits (based on maxouts?) ModifyTOD schedule Practical limitations Cannot adjust cycle length Cannot accommodate traffic markedly different from pattern basis
- 22. Synchro Green Calculation engine like Synchro Like Gen 1.5, traffic responsive with regular recalculation of background pattern
- 23. QuicTrac Estimate platoon speed with system detectors and calculate offsets at master Calculate cycle length at master based on local volumes Calculate splits at local
- 24. SCOOT Calculation engine similar toTRANSYT/7F Based on signal timing theory Measures volumes entering a link Calculates cycle length, splits and offsets Makes regular small steps to follow changing traffic patterns Coordinated groups fixed. Usually someVA
- 26. SCOOT cycle by cycle changes
- 27. SCATS Measures degree of saturation at stop line to calculate cycle length and splits Measures volumes entering links to select best direction of offset Practical implementation of how traffic engineers set up coordination patterns Determines which intersections need coordination – flexible grouping Selects appropriate objective function based on V/C regime
- 28. Sunnyvale: Extension of SCATS 28
- 30. LA ATCS Over 3500 intersections under control System being expanded Apparently operates well, but not really as adaptive as it sounds Not well documented Better supported in last few years 30
- 31. Other systems Intelight adaptive (new) Surtrac (experimental – Carnegie Mellon) OPAC Rhodes SPOT
- 32. 32 What is good? What is not? SCOOT – well proven in many countries, variable in USA SCATS – well proven in many countries, generally good in USA InSync – rapidly growing, variable results Synchro Green – steadily growing, satisfactory ACS-Lite – slow progress, several vendors Centracs Adaptive – integrated and improved ACS-Lite QuicTrac – several successful deployments RHODES – development continues, practical results patchy OPAC – variable results, not growing LA ATCS – apparently good, little published data, expanding beyond City of LA.
- 33. Is an adaptive system right for you Do you have a corridor or closed network Does demand exceed the theoretical and practical capacity of your corridor
- 34. Should I consider adaptive control?
- 35. What do the opponents say? I know exactly what will happen on my street on Monday morning. The transitions cause too much disruption and are counter productive The detection costs too much and is not reliable The communication costs too much or is unreliable I don’t believe the claims I have too much investment in my current system
- 36. How should I get an adaptive system? If any Federal funds involved, must provide a systems engineering analysis, commensurate with the scale of the project Selection must be competitive, but recommend AGAINST low-bid process You are buying technology, with every product different, not buying pavement complying with a physical specification
- 37. Procurement options BestValue Low Bid with SE Support “Consumer Reports"
- 38. Would you do it again? 38 Source: Adaptive Traffic Control Systems in the United States, Matt Sellinger, HDR Engineering, Inc.