Problem instances found in the literature that are used in computational studies of the resource constrained project scheduling problem, typically include only a few resources. In some practical applications, however, the number of resources may be significantly higher. In this paper, problem instances with a large number of resources are considered and a Benders decomposition approach is followed to deal with the additional computational effort required. Details of a separation routine for deriving the feasibility cuts and a heuristic for computing primal feasible solutions within a branch-and-cut framework, are provided. Randomly generated problem instances are introduced, which are based on existing problem instances from the literature. The empirical results reported in this paper demonstrate the scalability of the Benders decomposition approach when considering problem instances with a large resource set.
Technical Report FABWI-N-RA-2017-530, Centre for Business Mathematics & Informatics, North-West University, South Africa