Fleet Management

Fleet management addresses the problem of managing fleets of trailers, containers, boxcars, taxicabs, locomotives and business jets. In many applications, requests to be moved from one location to another arrive randomly over time. Since it may take days to move from one location to the next, it is often useful (even necessary) to reposition equipment before a customer demand is known. We use the techniques of stochastic optimization to solve two-stage and multistage fleet management problems, using the problem of rail car distribution as the motivating application. Rail car distribution is characterized by a high degree of randomness: in the orders being placed by the customers, in the empty cars becoming available, and in the transit times between locations. Car distribution problems also introduce other issues that are not typically addressed in the classical stochastic programming literature, such as lagged information processes, the need for discrete solutions, and the challenges posed when the relevant attributes of a car are allowed to grow (a common progression when solving real problems). This chapter is intended to introduce the reader to a real problem, providing a bridge between the classical literature of stochastic programming and the modeling and algorithmic challenges that would arise while solving a real problem.