Chance-constrained programming with robustness for lot-sizing and scheduling problems under complex uncertainty

Abstract

Purpose The purpose of this paper deals with a capacitated multi-item dynamic lot-sizing problem with the simultaneous sequence-dependent setup scheduling of the parallel resource under complex uncertainty. Design/methodology/approach An improved chance-constrained method is developed, in which confidence level of uncertain parameters is used to process uncertainty, and based on this, the reliability of the constraints is measured. Then, this study proposes a robust reconstruction method to transform the chance-constrained model into a deterministic model that is easy to solve, in which the robust transformation methods are used to deal with constraints with uncertainty on the right/left. Then, experimental studies using a real-world production data set provided by a gearbox synchronizer factory of an automobile supplier is carried out. Findings This study has demonstrated the merits of the proposed approach where the inventory of products tends to increase with the increase of confidence level. Due to a larger confidence level may result in a more strict constraint, which means that the decision-maker becomes more conservative, and thus tends to satisfy more external demands at the cost of an increase of production and stocks. Research limitations/implications Joint decisions of production lot-sizing and scheduling widely applied in industries can effectively avert the infeasibility of lot-size decisions, caused by capacity of lot-sing alone decision and complex uncertainty such as product demand and production cost. is also challenging. Originality/value This study provides more choices for the decision-makers and can also help production planners find bottleneck resources in the production system, thus developing a more feasible and reasonable production plan in a complex uncertain environment.