Multi-product reconfigurable disassembly line balancing and lot-sizing optimisation under uncertainty

Disassembly of End-of-Life (EOL) products is critical in remanufacturing, as it enables the recovery of valuable components. However, increasing EOL product diversity and uncertainty in product supply and component demand challenge disassembly line design and planning. To address these challenges, we investigate a multi-product reconfigurable disassembly line balancing problem integrated with lot-sizing under uncertainty. A two-stage stochastic programming model is developed to determine line configurations and disassembly plans, aiming to ensure a target profit with high probability. To overcome the stochastic model's reliance on known probability distributions and its sole focus on the likelihood of meeting the target profit, we introduce a novel target-oriented Profit Achievability Index (PAI). The PAI evaluates the achievability of a target profit by jointly considering the average magnitude of profit shortfalls and their tail risk. A new PAI-based distributionally robust optimisation model is then formulated, and an improved bisection search algorithm with a divide-and-conquer strategy is developed to enhance computational efficiency. Experimental results reveal managerial insights into the value of disassembly line reconfigurability and target-oriented modeling. Furthermore, computational experiments on random instances show that the proposed algorithm reduces average solution time by approximately 86% and 72% relative to Gurobi and basic bisection search, respectively.