Models for Predicting Bus Delays

Abstract

The provision of accurate bus arrival time information to users is essential for improving the attractiveness of public transit service. Although bus-tracking technology provides real-time information to the control center about the location of a bus, there is a need to improve the prediction of bus travel time downstream from location of last observation in mixed traffic operations. Methods are required for making predictions in normal traffic conditions, as well as in conditions in which a temporary lane closure is experienced due to a variety of reasons, such as incidents and road improvement activities. The development of models for the estimation of the effect of changes in traffic and lane closures on bus performance is described. A microsimulation approach was used, supplemented by field studies. The models developed meet calibration tests and were verified by field data.