A Reactive Kanban System by Allocation of the Number of Kanbans

Abstract

In order to realize agile control for multi-stage production systems, we have developed a reactive Kanban system. In the previous system, the unstable changes in product demand can be detected by monitoring the time series of demand data and detecting unstable change with the exponentially weighted moving averages chart. For the detected change, the previous system adjusts the number of Kanbans at each inventory point as a response to the change. By the dynamic adjustment of the number of Kanbans for the unstable changes in product demand, the previous system can maintain the required level for the mean waiting time of product demand by minimizing the total of the mean work-in-process inventories. However, in the previous system, the number of Kanbans is allocated equally to each inventory point. As shown in previous literature on determining the number of Kanbans, uniform allocation is not always optimal for minimizing the total of the mean work-in-process inventories. Based on that fact, this paper proposes a reactive Kanban system by allocating the number of Kanbans for each inventory point. Similar to the previous system, the proposed system can detect the unstable change of product demand and adjust the number of Kanbans as a reaction to the unstable change. However, in the proposed system, the optimal number of Kanbans for each inventory point is determined based on simulation results for minimizing the total of the mean work-in-process inventories to satisfy the required level for the mean waiting time of product demand. In addition, the proposed reactive Kanban system is modified by combining the system with the previous system where the number of Kanbans is allocated equally to each inventory point. The effectiveness of the proposed system is investigated with simulation experiments.