Multi-dimensional labelling approaches to solve the linear fractional elementary shortest path problem with time windows

This paper investigates the linear fractional shortest path problem with time windows. For the specific problem, an elementary path with a minimum cost/time ratio is sought in a directed graph, where two parameters (i.e. cost and time) are associated with each arc and a time window is associated with each node. Indeed, a valid path must satisfy the time window constraints, which are assumed to be of the hard type. Multi-dimensional labelling algorithms are proposed to solve this variant of the classical shortest path problem. Extensive computational tests are carried out on a meaningful number of test problems, with the goal of assessing the behaviour of the proposed approaches. The computational study shows that the introduction of dominance rules and the adoption of a bi-directional search strategy allow the definition of solution approaches that turn out to be very effective in solving the problem under consideration. © 2011 Taylor & Francis.