Optimal Real-Time Control Strategies for Rail Transit Operations During Disruptions

Abstract

Rail transit systems are frequently subject to short term disruptions which temporarily block traffic, leading to increased passenger waiting times and overcrowding of trains. This paper focuses on the development of a real-time decision support system for rail transit operations. We develop a deterministic model of a rail system, and mixed integer programming formulations for several holding and short-turning strategies. We apply the formulations to problem instances based on the MBTA Red Line and demonstrate that passenger waiting time can be significantly reduced (on the order of 15–50%) by applying the resulting controls. We show that the great majority of benefits can be realized by applying only limited control actions on a small set of trains. Our formulations determine optimal control strategies, typically in less than 30 seconds.