Cooperative Dispatching is a real-time scheduling methodology, which consults downstream machines before making a job dispatching decision on any given machine. This paper proposes such an approach for minimizing the mean tardiness in a dynamic flowshop where new jobs arrive continuously, at random points in time, throughout the production cycle. Cooperative Dispatching is based on the idea that individual machines act self-interestedly, with the objective of optimizing their local performance criteria. A consulted machine attempts to influence upstream dispatching decisions in a manner that promotes its ability to minimize its total local tardiness. A machines influence in the dispatching decision depends on current congestion and due-date tightness levels in the shop. A multiple regression model is proposed to help determine the weight a consulted machines preferences will carry in the dispatching decision. Conflicting demands from the different machines are resolved by a minimum regret decision procedure, which aims to minimize the aggregate deviation from the consulted machines preferences. The winning candidate that ultimately emerges from this procedure is the job that is dispatched. A comparative analysis to evaluate the performance of cooperative dispatching, compared to six other dispatching rules that are commonly favoured for tardiness-based criteria, is performed by means of simulation, using randomly generated test problems. Computational results indicate that Cooperative Dispatching outperforms the other dispatching rules, across a broad range of flowshop congestion and due-date tightness levels.
ASJC Scopus subject areas