Type-2 integrated process-planning and scheduling problem: Reformulation and solution algorithms. (June 2022)

• Record Type: Journal Article
• Authors: Naderi, Bahman; Begen, Mehmet A.; Zaric, Gregory S.
• Is Part Of: Computers & operations research. Volume 142(2022)
• Publication Date: 2022-06
• Subjects: Type-2 integrated process planning and scheduling problem -- Benders decomposition -- Constraint programming -- Optimality cut; Operations research -- Periodicals; Electronic digital computers -- Periodicals; 004.05
• DOI: 10.1016/j.cor.2022.105728
• Languages: English
• ISSNs: 0305-0548

Abstract

Abstract: We study the type-2 integrated process-planning and scheduling (IPPS) problem where each job is represented by a directed network graph. To the best of our knowledge, there is only one mathematical model in the literature implementing the type-2 IPPS partially, and the solution methods available for this problem are all based on heuristics and metaheuristics. We introduce three properties that enable us to fully formulate all aspects of the type-2 IPPS problem with a mathematical programming model for the first time. To solve our model, we develop a logic-based Benders decomposition method hybridized with constraint programming. We decompose the problem into two smaller ones such that we can use the best solution technique for each one, master problem and subproblem. To enhance our solution approach, we incorporate a combinatorial relaxation of subproblem into the master problem. We evaluate our method using a well-known benchmark including 24 instances and compare its performance with six existing solution methods solving the same benchmark. We solve all the 24 instances of this benchmark to optimality where seven of these 24 instances are solved to optimality for the first time. We also generate a new set of 144 larger instances to further evaluate our solution methods and provide insights on when each method performs better. Highlights: A full mathematical model for the type-2 IPPS problem. Logic-based Benders decomposition method with constraint programming. 7 of 24 existing benchmark instances are solved to optimality for the first time. A new set of 144 larger instances to further evaluate our solution methods. Insights based on numerical experiments on problem size and solution methods.