Multi-level lot-sizing problem: Evaluation of a simulated annealing heuristic

The multi-level lot-sizing problem (MLP) is the problem of determining production quantities in multi-stage production settings, such that the sum of set-up and holding costs is minimized. This type of problem is hard to solve to optimality thereby compelling one to use heuristic approaches. In this paper the authors investigate heuristics based on a stochastic search method. Experimental results concerning the quality and efficiency of these methods for the MLP are presented and compared to the quality and efficiency of heuristic methods which are based on applying single-level heuristics on a level-by-level basis.