|M.Sc Student||Genez Adar|
|Subject||Formulation and Application of a Demand Responsive|
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Shlomo Bekhor|
Public transportation (PT) is an efficient and sustainable solution to congestion, emissions and other problems related to the growing use of private cars. PT shortcomings, such as temporal and spatial incompatibility to ones needs, discomfort and unreliability are the main reasons for the unwillingness of some to give up their cars, as private car patronage remains high. Demand responsive transit (DRT) is an emerging sector of PT aiming to tackle one of these shortcomings by offering the public with an adaptive service. DRT’s development in the 21st century is based on technological progress, especially regarding widespread internet accessibility.
The DRT literature is extensive, and many models has been studied and analyzed. The scheme to be modeled in this research has yet to be studied to its extent and is designed to facilitate travelers ride requests by ensuring a continuous trip, starting at a pick-up location in spatial and temporal proximity to traveler’s request and ending at a drop-off location near-by one’s destination. The service enables ride sharing, as routes and schedule determined according to current demand. These service flexibilities are the key elements enabling low trip rate, with just a minor addition to passengers’ inconvenience compared to a taxi ride.
Methodology of the research was to formulate the model as an optimization problem, and to examine its applicability using sample datasets. The optimization problem is based on the vehicle routing problem with pickup and delivery with time windows (VRPPDTW). The basic VRPPDTW was added with constraints enabling pickup and drop-off locations to differ from traveler origin and destination and enabling service vehicle to idle, at cases, during a shift. The objective function is comprised of three features - minimizing vehicles’ operational costs and travelers discomfort, while maximizing the number of passengers. The outcome was a mixed integer linear programming problem, with decision variables setting the routes of all agents, fleet usage and scheduling, and travelers-to-vehicle allocation. Solution process has two parts: (a) adjusting the road network to the model and finding all shortest paths prematurely, and (b) finding the global optimum. Each part was then solved by generic algorithm.
Results using a toy-size network and a sample travel requests dataset demonstrated model’s advantages. Scenarios produced solutions demonstrating passengers’ walks, vehicle idle time when scenario allows or idle time prevention when situation dictates so. Problem parameters sensitivity analysis examples are shown to demonstrate the use of this tool to find optimality in service planning parameters values.
This work has reaffirmed the conclusion that DRT schemes holds the potential to tip the scale towards a sustainable transportation mode split. This proposed model main innovative element - the enablement of travelers walking segments - has produced another important flexibility, while increasing model complexity. Nevertheless, this research could set the foundation for further studies assessing the proposed model, its performance and its impacts on traffic assignment and travelers’ mode choice. Further research is required to develop an efficient algorithm for solving real world instances, so the true effectiveness of the scheme could be assessed.