Metro Toronto currently employs fixed time signal control in the Queen Street Corridor. However, the traffic varies significantly among cycles, due mainly to the presence of streetcars in some cycles. Streetcars hold up other traffic while loading, causing the flow profiles to vary from cycle to cycle. This transience is difficult to model with fixed time techniques. Also, a fixed time plan, such as provided by the TRANSYT model may not be optimal. The evolutionary types of real-time control that are provided by the SCOOT model are also not responsive to streetcars in real time. Since Metro Toronto and the Toronto Transit Commission are planning to implement real-time transit priority, mputer model was developed for developing and evaluating real-time signal plans. While the first version of this model treats isolated intersections, a long-term goal is to develop an integrated system of isolated intersections sharing advance information for co-ordinated real-time network control. For now the isolated intersection is considered a reasonable approximation at intersections with large uncoordinated transit volumes in competing approaches. The model was developed to evaluate the effects of various levels of streetcar priority. Blind priority to streetcars, without due consideration for private vehicles can create a wake through which later transit vehicles are less easily accommodated. Therefore, an evaluative decision model was developed to generate alternative real-time phases, selecting the one with the best total delay to people in cars and streetcars. The paper describes the model and an initial application to a busy Toronto intersection to evaluate the feasibility of real-time control.
Abstract