Three problem areas exist in designing and implementing a kanban controlled JIT system: the identification of flow lines problem, the flow line loading problem and the operational control problem. This paper addresses the operational control problem. A general N-stage serial production system is modeled as a discrete time Markov process. Capacity constraints, stochastic machine reliability and demand variability are included. The model is illustrated by a 3-stage system, describing the effects of the number of kanbans, the machine reliability, the demand variability and safety stock requirements on the performance of a kanban controlled pull system.kanban system, production/inventory, Markov process

No abstract