Tabu assisted guided local search approaches for freight service network design

Service network design involves determination of the most cost-effective transportation network and service characteristics subject to various constraints. Good progress has been made in developing metaheuristic approaches that can compete or even outperform some commercial software packages (Ghamlouche et al 2004; Pedersen et al 2009). However, since most of these metaheuristic methods involve solving many capacitated multi-commodity minimum cost flow problems, computational time tends to be a bottleneck. In this research, we intend to build on the success of a guided local search metaheuristic (Bai et al 2010) in reducing computational time, without compromising solution quality, and carry out a set of experiments and analyses in an attempt to discover elements and mechanisms that could improve the algorithmic performance further.