A stochastic network design of bulky waste recycling – a hybrid harmony search approach based on sample approximation

Facing supply uncertainty of bulky wastes, the capacitated multi-product stochastic network design model for bulky waste recycling is proposed in this paper. The objective of this model is to minimize the first-stage total fixed costs and the expected value of the second-stage variable costs. The possibility of operation costs and transportation costs for bulky waste recycling is considered while making the first-stage decisions about the network configuration. Integrating harmony search with sample approximation approach, a meta-heuristic is elaborated to solve the stochastic network design problem of bulky waste recycling, i.e. to decide where bulky waste recycling centers should be opened, and to determine which disposition procedure should be introduced. Several numerical examples are utilized to demonstrate the validness of the harmony-search-based sample approximation algorithm (HS-SAA). In addition, the HS-SAA is compared to a genetic algorithm previously applied to the same problem. HS-SAA has advantages in the solution quality.