Optimal work-in-process control for a closed multistage production system with machine preference

Abstract

In this work, we expose a closed-loop production system which has great similarity with the Kanban system as the Work-in-Process (WIP) is constant (CONWIP). This system is a multistage single product and it supports more than one machine at every stage. These machines can have different characteristics and they have limited capacity of buffering. We consider also that there is a preference for a machine usage in each stage. We use a Closed Queueing Network (CQN) with a repetitive blocking mechanism to model this system. The used resolution approach is the Maximum Entropy Method (MEM). In the first place, we look for the optimal WIP to keep in this system with respect to a production rate. In the second place, the routings to machines in the same stage are evaluated regarding the buffering capacity and the machine characteristic. We present some numerical results on a simple two-stage production system which holds the same number of batches looping. Interesting perspectives for the development of the presented model are looked over.