A systematic study is described to address various design and implementation issues associated with the problem of real-time bus holding control. Two holding control models have been investigated. The first model follows the conventional threshold-based control logic that determines holding times on the basis of headway to the preceding bus. The second model makes use of both preceding and following headways in identifying optimal control decisions with the assumption that real-time bus location information is available for estimating future bus arrivals at the control stop. An extensive simulation analysis is performed using a real-life bus route operated by the Grand River Transit of the region of Waterloo, Ontario. The simulation results have substantiated several conclusions and yielded new findings on various issues such as where to set the control point, how many control points should be used, what is the optimal control strength, and what is the value of real-time location information.
Abstract