Optimal signal timing design for urban street networks under user equilibrium based traffic conditions : final report.

Final reportPDFTech ReportNEXTRANS Project No. 019FY02DTRT12-G-UTC05Heuristic methodsOptimizationTraffic assignmentTraffic congestionTraffic delayTraffic flowTraffic signal timingUrban areasUnited StatesIllinoisChicago (Illinois)Purdue UniversityLi, ZongzhiDu, LiliLiu, YiIllinois Institute of TechnologyNEXTRANS Center (U.S.)United States. Dept. of Transportation. Research and Innovative Technology AdministrationUniversity Transportation Centers Program (U.S.)US Transportation CollectionIn the ever-growing travel demand, traffic congestion on freeways and expresswaysrecurs more frequently at a higher number of locations and for longer durations withadded severity. This becomes especially true in large metropolitan areas. Particular to theurban areas, excessive crowdedness caused by inefficient traffic control also results inurban street networks operating in near or over-saturated conditions, leading to unpleasanttravel experience due to long delays at intersections. As a consequence, the recurrenttraffic congestion on roadway segments and vehicle delays at intersections inevitablycompromise energy efficiency, traffic mobility improvement, safety enhancement, andenvironmental impacts mitigation. All too often, neither restraining travel demand norexpanding system capacity is desirable and practical. Conversely, effectively utilizing thecapacity of the existing transportation system has been increasingly thought of as thesolution to congestion relief. With respect to the urban street networks, developingeffective means for urban intersection signal optimization becomes essential to reduceintersection delays.Conventional signal timing optimization methods use historical traffic data andassume that traffic flows will remains unchanged after the implementation of new signaltiming plans. Traffic flows are assumed to be constant, but in fact, when signal timingplans change, travel times for some travel routes will alter, which requires drivers in thenetwork to adjust their choice of travel routes to arrive at the destinations, and result in redistribution of traffic in the network. Therefore, the effects of interactions betweensignal timing plans and traffic flows need to be explicitly taken into consideration. Thisstudy introduces a new methodology that jointly considers signal timing optimization andtraffic assignment in an overall analytical framework that contains model formulationsunder assumptions consistent with real world situations. Such a framework is well suitedfor applications in real world cases. Specifically, the overall optimization framework isformulated as a bi-level optimization problem. In the proposed basic model, at the upperlevel, a traffic signal timing optimization problem for urban network is introduced tominimize system total travel time by optimizing signal green splits. At the lower level, astatic user equilibrium problem is formulated for networkwide traffic assignment. In thevehicle delay estimation, the time-dependent stochastic delay model in the 2010 HighwayCapacity Manual (HCM 2010) is employed and formulated as Variational Inequalityconstraints, what allow the state-of-the-art MPEC solver, GAMS/NLPEC, to solve theproblem for a local optimal effectively and efficiently. The bi-level optimization model isfirst tested using a small network (the test network) and a computational experiment usinga subarea network in the Chicago central district is conducted to assess the practicality ofthe model formulation in real world applications. In order to import more reality to the basic model and also consider the potentialsystem benefit that comes from different signal phasing design, an enhanced model isdeveloped based on the basic model by employing integer and binary variables.Formulating the problem with binary variables allows for the selection of proper phasingdesign. Heuristic solution methods are proposed and tested using the test network