Multicriteria analysis of a real vehicle scheduling problem under travel time uncertainty

This paper addresses a multicriteria analysis for the vehicle scheduling problem under conditions of uncertain travel times. A previously proposed metaheuristic is used to estimate the distribution of trips to vehicles that produce the lowest possible operating cost. During the optimization process, all solutions are stored into a memory structure, and the local minima are extracted for further analysis. These solutions are evaluated in a set of proposed reliability indicators, via Monte Carlo simulations, allowing the design options to be compared more adequately by a decision-making method. The approach used for mono-objective optimization was validated using data from transport companies of the city of Belo Horizonte/Brazil and the results of the multicriteria analysis show superiority in relation to the solutions used in practice. In addition, the results of the simulation show that not always the least cost solutions are the most appropriate to the real situations of uncertainty.