A genetic algorithm with an earliest due date encoding for scheduling automotive stamping operations. (March 2017)

• Record Type: Journal Article
• Authors: Roychowdhury, Sayak; Allen, Theodore T.; Allen, Nicholas B.
• Is Part Of: Computers & industrial engineering. Volume 105(2017)
• Page Start: 201
• Page End: 209
• Publication Date: 2017-03
• Subjects: Scheduling -- Meta-heuristic -- Tardiness -- Due date -- Genetic algorithm -- Manufacturing; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2017.01.007
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: Methods are proposed for a real-world stampings scheduling at an automotive company. A comparison is provided with 5 alternatives on 6 test problems for 4 metrics. The proposed genetic algorithm generalized earliest due date method is viable. Conditions for the global optimality of earliest due date scheduling are clarified. Abstract: This article considers a manufacturing scheduling problem related to automotive stamping operations. A mathematical program of the associated single machine problem is formulated with known demand, production constraints involving stamping dies, and limited storage space availability. It is demonstrated that a generalized version of the standard earliest due-date heuristic efficiently generates optimal solutions for specific problem instances (relatively high initial inventory cases and no tardiness) but poor solutions for cases involving relatively low initial inventories and/or longer time horizons. Branch and bound is shown to be inefficient in terms of computational time for relevant problem sizes. To build a viable decision support tool, we propose a meta-heuristic, "genetic algorithms with generalized earliest due dates" (GAGEDD), which builds on earliest due date scheduling. Alternative methods are illustrated and compared using a real-world case study of stamping press scheduling by an automotive manufacturer.