A chance-constraint programming approach for a disassembly line balancing problem under uncertainty

Against the backdrop of escalating global imperatives for resource conservation and environmental sustainability, the strategies of recycling and remanufacturing have risen to prominence as crucial solutions, particularly in tackling the challenges presented by end-of-life (EOL) products. At the heart of these operations lies the intricate process of product disassembly, where the optimal allocation of tasks becomes paramount. This study focuses on the disassembly line balancing (DLB) problem, a critical aspect of streamlining disassembly activities to enhance efficiency and curtail wastage. While extant DLB formulations abound, a prevalent oversight pertains to the neglect of the dynamic uncertainties inherent in disassembly information and insufficient uncertainty handling, as well as the non-standard constraints that often manifest in practical settings. To bring DLB models closer to real-world applicability, this research introduces a stochastic multiobjective DLB (SMDLB) framework, employing an innovative chance-constraint programming methodology. The framework's primary aim is to minimize multiple objectives, including the idle rate and variance in idle times across workstations, alongside energy consumption, all while factoring in specific confidence levels. Navigating the intricacies of the SMDLB model, this study presents a hybrid metaheuristic algorithm that seamlessly integrates the water cycle algorithm (WCA) for efficient exploration, coupled with the adaptive capabilities of simulated annealing (SA) for effective exploitation. Leveraging a sophisticated stochastic simulation method, the proposed algorithm adeptly addresses the uncertainties inherent in the disassembly process. The practical efficacy of the proposed model and algorithm is rigorously demonstrated through their application to real-world case studies involving a worm reducer, transmission, and a liquid crystal display (LCD) monitor housing. Subsequently, a comprehensive performance analysis, encompassing diverse multiobjective metrics and rigorous statistical tests, serves to benchmark the proposed algorithm against established counterparts in the existing literature. The findings underscore the adaptability of the proposed model, empowering decision-makers with an expanded array of options that facilitate informed decisions, enabling an optimal equilibrium between conflicting objectives, particularly within the realms of uncertainty.