Optimization Methodology for Bus Garage Locations

Abstract

An optimization model is developed for determining both the optimal bus garage locations and optimal assignment of buses to those garages. The model uses a list of candidate garage sites, the transit system’s service and operational characteristics and the characteristics of the regional highway system as input. As output, it produces the optimal number, sizes and locations of bus garage additions or changes. The model considers garage operating and garage construction cost and the costs of deadheading and driver relief under separate operating schedules for morning, all day, and evening route assignments on weekdays, Sundays and Saturdays. In the methodology development, the problem is first formulated as a discrete solution-space fixed-charge facility location problem. The fixed charge problem is formulated as a mixed integer program. However, because of the difficulty involved in solving a large-scale problem with a general purpose mixed-integer program, the problem is solved by a special purpose branch-and-bound process. This process utilizes an efficient sequence of classical transportation problem (a quick solving, special class of linear program). The methodology is demonstrated with a small-scale sample problem.