A genetic algorithm based heuristic for dynamic lot sizing problem with returns and hybrid products. (May 2018)

• Record Type: Journal Article
• Authors: Koken, Pakayse; Arasanipalai Raghavan, Venkatesh; Yoon, Sang Won
• Is Part Of: Computers & industrial engineering. Volume 119(2018)
• Page Start: 453
• Page End: 464
• Publication Date: 2018-05
• Subjects: Inventory -- Dynamic lot sizing -- Hybrid products -- Metaheuristics -- Constraint handling; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2018.03.040
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: Studied dynamic lot sizing problem with returns and hybrid products. Developed a metaheuristic algorithm to find near optimal solutions efficiently. Explored the profitability conditions for producing hybrid products. The performance of the developed metaheuristic is good. The proposed model performs well at medium to high holding cost environments. Abstract: For a hybrid system with manufacturing and remanufacturing, a variant of dynamic lot sizing problem is addressed in this study. In the system, manufactured and remanufactured products are produced on separate lines and sold in segmented markets. In addition to these two types of products, there are also hybrid products produced in the system. Hybrids are used to meet the excess manufactured product demand and integrate the two distinct lines. Therefore, this study investigates the profitability conditions for producing the hybrid products. Using a variant of dynamic lot sizing problem, called dynamic lot sizing problem with returns and hybrids (DLSPRH), which is a constrained mixed-integer nonlinear programming problem, the performance of the system with hybrids is compared to the same system with no hybrids. The DLSPRH is a NP-hard problem. A Genetic Algorithm based heuristic (GA_H) has been proposed to solve the DLSPRH and its capacitated version from the literature. The performance of the algorithm is tested by comparing its results with Simulated Annealing (SA), Variable Neighborhood Search (VNS) and Simulated Annealing with Neighborhood List (SA_NL). Numerical experiments show that GA_H significantly outperforms the other metaheuristic algorithms. On average, GA_H performs 2.51%, 2.24% and 2.06% better than SA, VNS and SA_NL algorithms, respectively. Another finding is that the system with hybrids performs well at medium–high holding cost environments especially when remanufacturing demand is low. Additional managerial insights are also presented.