Analysis of partitioning problems of installations in job-shops systems through models of queuing networks

This paper studies the focused factory design involving the partition of the facility (plant) into sub-plants and the allocation of capacity in each workstation of the sub-plants. The job-shop system is represented by a generic open queuing network and approximations based on decomposition methods are used to evaluate and optimize the performance of the system. The aim is to reduce the system complexity from the point of view of the product or workstation management, for example, limiting the variance of the product leadtimes in the network. An integer non-linear programming model and a heuristic algorithm are presented. Applying the algorithm to some testing problems, it is shown that the partition of the facility into sub-plants can reduce the variance of the product leadtimes without requiring additional investments of capacity. Moreover, sometimes it is possible to maintain (or even improve) the network performance by partitioning the plant into sub-plants which need less capacity than the original plant.