A Decomposition Approach for Sequencing Mixed-Model Two-Sided Assembly Line with Stochastic Processing Time

Abstract

This study tackles the challenge of optimizing mixed-model two-sided assembly lines, where task processing times are uncertain. The objective is to reduce the expected cycle time the average time to complete one product across the assembly line. Given the complexity of assessing objectives amidst stochastic conditions, we formulate the problem as a simulation optimization problem. We introduce a strategic decomposition method that breaks down the core problem into two discrete sub-tasks: allocating tasks to respective work-stations, and determining the sequence of tasks at each station. The decomposition framework systematically partitions the solution space, though it does not ensure a global optimum, it can efficiently guide the search towards a high-quality near-optimal solution with a practical time frame. Based on this framework, we develop a novel simulation-optimization algorithm, termed the Decomposition Approach with Harmony Search (DAHS), which incorporates a harmony search heuristic to effectively navigate the partitioned solution space. Additionally, we implement two innovative strategies to improve the search and simulation procedures. Numerical experiments reveal that our DAHS algorithm outperforms benchmark algorithms in terms of solution quality and computational efficiency.