| Summary: | Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Civil and Environmental Engineering, 2000. === Includes bibliographical references (leaves 125-126). === Rail transit systems are subject to frequent disruptions caused by a variety of random disturbances, signal problems and door problems, for example. Such disruptions usually last for 10 to 20 minutes, which degrades the level of service significantly. To improve service reliability, transit agencies employ various real time control strategies, such as holding, expressing and short turning, to deal with these disruptions. The effectiveness of these control strategies relies upon the bird's-eye-view of the whole system. Unfortunately, it is difficult for human dispatchers to assess the situation and make good decisions in real time, even with the aid of advanced information technologies such as automatic vehicle location systems. This thesis focuses upon the development of a real-time disruption control model for rail transit systems during disruptions. A deterministic model to representing the rail transit system is first introduced. In the model, the passenger flow rates and running time between stations are constant but station-specific. Assuming that the disruption duration is known, a formulation is developed that makes use of real time vehicle location information and considers holding, expressing and short turning strategies to reduce the impact of the disruption. The objective is to minimize the sum of total platform waiting time and weighted in-vehicle delay. The original formulation is transformed into a linear mixed integer problem, which can be solved by any linear optimizer. The formulation is applied to a disruption scenario on a simplified system based on the Massachusetts Bay Transportation Authority Red Line. The sensitivity of different control strategies to the disruption duration assumption is investigated. The results showed that holding strategies combined with short turning strategies can reduce the weighted waiting time (the sum of platform waiting time and weighted in vehicle delay) by about 10-60%, compared with not applying any control strategies. Expressing only provided modest additional benefits. For the deterministic disruption duration assumption, sensitivity analysis showed that holding and expressing strategies are fairly robust, but the effectiveness of short turning strategies is quite sensitive to the accuracy of the disruption duration estimate. Most problem instances of the formulation can be solved in real-time with the proposed branching sequence used in the branch-and bound algorithm to solve this mixed integer problem. === by Su Shen. === S.M.
|
|---|
