A two-stage stochastic programming project scheduling approach to production planning

Abstract

In engineering-to-order (ETO) or manufacturing-to-order (MTO) systems producing highly customized items, the high level of customization, together with long flow times, forces the production plan to be defined before information on items customization, and details on the manufacturing activities are completely disclosed. Due to the partial available information, the production plan must provide a robust schedule of the activities and of the resources utilization, trying to incorporate a certain degree of anticipation of uncertainty. This paper proposes a two-stage stochastic programming project scheduling approach to support production planning in ETO/MTO system. The approach provides a baseline production plan together with a set of revisions of the plan to react to the occurrence of uncertain events. A scenario-based approach is used to model the changes affecting the characteristics of the activities to be processed. The proposed approach is tested on random-generated instances and on a real manufacturing system producing machining centers.