Evaluation framework for organization-based ridesharing :service design considerations and potential for AUB -

Abstract

Ridesharing services, including Carpooling and Demand Responsive Transit, have been gaining increased interest recently by individuals, employers and institutions alike. From the individual traveler’s perspective, car ownership and increasing fuel costs represent the main drivers to seek ridesharing services, especially in areas lacking reliable and attractive public transport services. On the other hand, the objective of employers and institutions (such as universities, hospitals, etc.) is to reduce car parking demand by their personnel (employees, students, etc.), as well as to reduce traffic congestion in their surroundings, while at the same time maintaining feasible travel alternatives for affiliated individuals. This dissertation presents a framework for the evaluation of ridesharing services in an organization-based context encompassing impacts and evaluation criteria and can be used as a decision support tool by any organization. This framework consists of three main modules: the demand estimation module, the service design module, and the evaluation module. Each of the three modules was analyzed taking into account the essential components, methods, and the needed data. An important aspect of this framework lies in the development of a many-to-one ride matching approach based on the Capacitated Vehicle Routing Problem with Time Windows. The problem is context-related to the ridesharing specifics, including unit demand, asymmetric network, narrow time windows at departure, and common arrival time at destination. New heuristic algorithms are proposed based on different traversals of hierarchical spanning trees derived from the all-pairs shortest path matrix (distance or travel time matrix). A tree type is selected depending on the solution strategy, where the Proximity Clustering Tree (PCT) is used to prioritize matching passengers within proximity clusters while the Minimum Deviation Tree (MDT) is used to match passengers along route with minimum deviations. Results are compared with exact soluti