Improving Paratransit Scheduling by Accounting for Dynamic and Stochastic Variations in Travel Time

Abstract

Travel times in an urban traffic environment often are subject to dynamic and stochastic variations due to random fluctuations in travel demands, frequent interruptions of traffic controls, and unpredictable occurrences of traffic incidents. Although these variations inevitably affect the real-life performance of a paratransit system, they have not been taken into account in the routing and scheduling process by most existing paratransit scheduling systems. The potential effects of these variations on the operational characteristics of a paratransit system such as vehicle productivity and schedule reliability are examined. A dial-a-ride routing and scheduling system capable of modeling dynamic and stochastic travel times was used in the analysis. A series of numerical experiments was performed on a practical problem from the city of Edmonton, Alberta, under hypothetical travel time variation patterns. It was found that both dynamic and stochastic variations in travel times had important effects on the quality of the schedules, and an appropriate consideration of these variations in the scheduling process could substantially improve the reliability and productivity of the schedules.