A priori reformulations for joint rolling-horizon scheduling of materials processing and lot-sizing problem

• A lot sizing and sequencing model with prepared materials is proposed.
• The prepared materials have sequence-dependent setup costs and times.
• Reformulation and strengthened constraints improve the model.
• Three alternative metaheuristics are used to fix the setup variables.
• The metaheuristics greatly improve performance.

In many production processes, a key material is prepared and then transformed into different final products. The lot sizing decisions concern not only the production of final products, but also that of material preparation in order to take account of their sequence-dependent setup costs and times. The amount of research in recent years indicates the relevance of this problem in various industrial settings. In this paper, facility location reformulation and strengthening constraints are newly applied to a previous lot-sizing model in order to improve solution quality and computing time. Three alternative metaheuristics are used to fix the setup variables, resulting in much improved performance over previous research, especially regarding the use of the metaheuristics for larger instances.