Abstract

The aim of any transit system is to provide the best level of service (LOS) to its passengers within the total available resources, like fleet size, vehicle capacity, screw strength etc. one of the ways of achieving this goal is to optimally allocate the available resources on different routes of the system, as well as, operate the system on an optimal schedule. The problem has been formulated as an optimization problem for simultaneously obtaining route-wise optimal resource allocation (fleet size) and minimum waiting time for passengers (which is assumed as a measure of the LOS). Two types of waiting times are considered, waiting time of passengers at their station of origin and time spent by passengers at transfer station waiting to transfer from one route to another. Classical techniques of optimization fail to give a solution to even simpler versions of the present problem (for example, where fleet size distribution amongst routes are assumed to be known).

In this thesis a Genetic Algorithm (GA) based approach to the optimization problem at hand is developed. GAs are chosen as a tool because it works well in a discrete search space and allows for external procedure based declarations during the optimization process. The approach is implemented for a three-route, single transfer station problem. The approach is tested under various problem scenarios. The results show that the proposed approach works extremely well.