A parallel hybrid heuristic for the multicommodity capacitated location problem with balancing requirements

In this paper, a parallel hybrid heuristic is developed for the multicommodity capacitated location problem with balancing requirements. The hybrid involves variable neighborhood descent (VND) and slope scaling (SS). Both methods evolve in parallel within a master–slave architecture where the slave processes communicate through adaptive memories. Numerical results are reported on different types of randomly generated instances, using an increasing number of processors and different distributions of processes between SS and VND.