A hybrid general lot-sizing and scheduling formulation for a production process with a two-stage product structure

In this paper we describe the model of an actual problem from the chemical industry, for which we developed a planning and lot-sizing tool. The production process is characterized by a two-stage product structure as well as sequence-dependent setup cost and setup times. The develop a model which is a modified version of the General Lot-sizing and Scheduling Problem (GLSP) framework and combines discrete-time as well as continuous-time elements. We furthermore develop two alternative reformulations which are based on the transportation problem and use disaggregated production variables instead of standard production and inventory variables. We show that both alternative formulations yield significant improvements with respect to computational time and integrality gap.