An algorithmic framework for the scheduling of construction projects based on ant colony optimization and expert knowledge

The conduction of construction projects in a road network can result in a reduction of the capacity of the lanes that are under construction or maintenance. In order to mitigate the impact of construction projects, it is critical to find the best schedule in such a way that the effects of road capacity reductions, e.g. traffic delay caused by their presence, is minimized. This article propose a novel formulation for the construction projects scheduling problem using bi-level programming. A solution method is then introduced to solve this challenging scheduling problem with a focus on reducing the number of calculations in the optimization. This is achieved by integrating expert knowledge, which can be used to characterize a good schedule, in the algorithmic framework in charge of determining a near-optimal schedule. Such heuristic information can come from relevant experts or practitioners, or from the literature in this field. A solution algorithm named CoANT based on ant colony optimization is presented and implemented on realistic transportation networks. Our results show that CoANT works fast and is able to provide competitive schedules. As an extensible and modular framework, CoANT can be used by relevant transportation agencies as a decision-aid tool for the coordination of construction projects in road networks.