A simulation-based assessment of alternative assembly line configurations

The problem of optimally assigning a set of timed tasks (work elements) that compose a production process to workstations, while satisfying precedence constraints, is known as the assembly line balancing (ALB) problem. The ALB problem (and its variants) is an NP-hard combinatorial optimization problem and for large-scale lines the objective is often reduced to finding feasible solutions. The situation is further exasperated by the time variability inherent in manufacturing tasks. Recent surveys on this field reveal that research efforts are directed primarily towards algorithmic approaches that provide incrementally better deterministic solutions to the ALB problem. Concurrently, the same surveys identify a considerable and widening gap between theoretical developments and actual implementations for assembly line configuration. In this paper, the focus is shifted towards developing a concise framework to assess actual performance under realistic conditions and it is proposed that simulation experiments may better predict the performance and robustness of balancing solutions before they are implemented in practice.