Mathematical models for capacitated multi-level production planning problems with linked lot sizes

Various mixed-integer programming models have been proposed for solving the capacitated multi-level lot sizing problem with linked lot sizes. It would be of value for researchers and practitioners to know which of these models is the most efficient under different circumstances. To investigate the comparative efficiencies associated with these models, this paper therefore demonstrates theoretically the relationships between these models when the integrality requirement is relaxed for any subset of binary setup and setup-carryover variables, shows the relative effectiveness of these models in obtaining lower bound solutions associated with linear programming relaxations, and evaluates the relative computational resources and the time needed when using these models through intensive computational tests. These theoretical and numerical results are expected to provide significant guidelines for choosing an effective formulation for different situations.