Dion, François2006-07-282006-07-2819991999http://hdl.handle.net/10012/370While past experiences show that real-time, traffic-responsive signal control has an ability to improve traffic operations in urban areas when compared to traditional fixed-time control, none of the existing real-time systems currently possess the ability to provide true optimal control in all types of networks. Two notable deficiencies exist when these systems are applied to networks in which passenger cars and transit vehicles share the right of way. First, none of these systems considers the effects on the general traffic of transit vehicles stopping in the right of way to board and discharge passengers. Second, priority of passage is often awarded to approaching transit vehicles without considering all the potential effects that such preferential treatment might have on other traffic. The thesis describes a fully distributed signal control model that attempts to solve the above two problems. The model is a network extension of a real-time, traffic responsive signal control model for isolated intersections named SPPORT (Signal Priority Procedure for Optimization in Real-Time) that has been explicitly designed to consider the effects that transit vehicles might have on general traffic progression while stopped in the right of way and to provide priority to these vehicles on a conditional basis. While the network version of the model still attempts to optimize signalized intersections individually, it introduces procedures that allow the operation of adjacent intersections to be coordinated without excessively constraining the ability of the model to respond to changes in local traffic demands. The SPPORT model features the use of the unique heuristic rule-based signal optimization procedure that allows the model to respond only to traffic events that are defined to have some importance for the signal operation. By ignoring all unimportant events, this optimization process allows a significant reduction in the number of signal-switching combinations that need to be considered to find an optimum signal control solution and make the SPPORT model more amenable to real-time control than exhaustive optimization methods. In the model, coordination needs with downstream intersections are considered by adjusting, within the rule-based optimization process, the times at which green signal indications are required on each intersection approach following the identification of important traffic events. For each approach, adjustment are made on the basis of the projected signal timings at the main downstream intersection, the queue dissipation times along the links leading to that intersection, and the degree to which dwelling transit vehicles stopped in the right of way interfere with the progression of general traffic on these links. Coordination with adjacent upstream intersections is achieved by examining the potential for queue spillback across these intersections, the recently implemented and projected local timings, and queuing conditions on the links joining the upstream intersections with the one being optimized. Coordination with upstream intersections is further enhanced by considering projected departures from them when predicting near-future stop line arrival patterns at each controlled intersection. The simulation studies conducted in the thesis demonstrate the ability of the extended SPPORT model to provide efficient real-time, traffic-responsive signal control in coordinated urban networks with mixed-traffic conditions. When compared to an optimal fixed-time operation, the application of the proposed model reduced the delays incurred by all vehicle passengers at an isolated intersection by as much as 35 percent in scenarios considering transit interference on general traffic progression, transit priority treatments and peaking arrival patterns. When applied to a typical five-intersection urban arterial, reductions of up to 50 percent were observed in a performance function evaluating the stops and delays incurred by all vehicle passengers. These results demonstrate the effectiveness of the priority rules defined within the model and of the model structure in which a series of candidate control strategies are evaluated before the best one is selected for implementation. The results also demonstrate the ability of the model to quickly respond to changing traffic conditions and automatically alter its signal control strategy when queues threaten to spill across controlled intersections. While the signal coordination procedures developed in the thesis are also found to be generally beneficial to the signal control performance, it was discovered that considering projected departures from upstream intersections can negatively affect the signal operation if the model is unable to assign different relative importance to projected traffic events in relation to current or imminent events.application/pdf20056752 bytesapplication/pdfenCopyright: 1999, Dion, François. All rights reserved.Harvested from Collections CanadaDoctoral Thesis