ANALYSIS OF SLACK TIME FOR A REAL-TIME RIDESHARE SYSTEM

Abstract

Ridesharing and carpooling are effective traffic demand management strategies with many benefits comparable to other highway management methods. As a safety factor can increase the reliability of transit systems, a slack time can be added to the passenger pickup schedule to increase the system stability and reliability. This thesis focuses on a driver-passenger system, modeling three objective cost functions using similar steps and assumptions. These modeled cost functions yield the optimal slack time for passenger and vehicle and reflect the user’s and supplier’s behavior towards changes in different model components. The numerical results of the cost models are presented to show the relations between different model components and to test the behavior of the cost models. The sensitivity analysis of these relationships reveal that factors such as the value of time, maximum waiting time, the penalty of missed pickup, and the standard deviation of the distribution of arrivals, affect the optimum slack time for the driver and passenger. The findings of these cost models can be integrated with matching algorithms to be used in real-time ridesharing applications. The distributions of arrivals for the passenger and vehicle are both assumed to be normal in this study, but other probability distributions can be substituted to investigate the costs associated with any connection among multiple vehicles or modes. The method presented in this study is applicable when passengers take advantage of advanced scheduling for pickups. A rideshare system might provide some incentives for users to schedule their pickup with substantial lead time rather than requesting a ride with minimal advanced notice, on an as-needed basis.