Reacting JIT ordering systems to the unstable changes in demand

Abstract

In order to realize an agile control for multi-stage production inventory systems, this paper proposes reactive JIT (Just-in-Time) ordering systems for the unstable changes in demand. Two JIT ordering systems, the Kanban system and the concurrent ordering system, are dealt with. For the JIT ordering systems, a reactive controller of bu er size is proposed in this paper. In the proposed systems, the unstable changes in demand are detected with the exponentially weighted moving average charts, and the bu er size at each inventory point is controlled for reacting to the detected unstable changes. The performances of the proposed JIT ordering systems are analysed and compared with each other by simulation experiments. IntroductionThe performances of multi-stage production inventory systems depend signi® -cantly on the ordering system which determines the order release for each process. Various ordering systems have been developed, and the JIT (Just-in-Time) ordering systems, such as the Kanban system (Kimura and Terada 1981) and the concurrent ordering system (Izumi and Takahashi 1993), are the typical ordering systems. JIT ordering systems are ordering systems based on the JIT production concept, that is to produce a necessary item with a necessary volume at the necessary point in time. In order to realize the JIT production concept, research on the JIT production from various view points is carried out all over the world. One of the categories is research on JIT ordering systems, and other categories include research on determining the number of Kanbans in the Kanban system (reviewed by Joo and Wilhelm (1993) and Price et al. (1994)) , that on determining the production sequence at the ® nal stage (reviewed by Joo and Wilhelm (1993) and Kubiak (1993)), and the JIT philosophy (reviewed by Sohal et al. (1988) and Golhar and Stamm (1991)) .For research on JIT ordering systems, the Kanban system (Kimura and Terada 1981) and the concurrent ordering system (Izumi and Takahashi 1993) are proposed. Both of the JIT ordering systems are based on the JIT production concept, but the information utilized in releasing orders is di erent for each system. The Kanban system utilizes three kinds of information, demand arrival, parts supply, and process completion, whereas the concurrent ordering system utilizes only one kind of information, demand arrival. As the di erence in the information utilized makes the performances of the systems di erent from each other, the performances of the JIT ordering systems are analysed and compared with each other . Furthermore, the performances are compared with the other ordering systems, such as MRP (Material Requirements Planning) system. The results show

Asynchronous action-reward learning for nonstationary serial supply chain inventory control

“…Stochastic learning automata methods [2] were used to model the learning behavior of the system. Using a statistical control chart, Takahashi and Nakamura [14] proposed a reactive Kanban system that can detect unstable changes in demand timely. The number of Kanbans as a reaction to detect unstable situation is a control parameter.…”

Managing Preventive Maintenance on a Disassembly Line Using Multi-Kanban Mechanism

“…They demonstrated that in such environments, FKS outperforms the traditional kanban system. Other adaptive and reactive techniques to adjust the number of kanbans were proposed by Tardif and Maaseidvaag [23], Takahashi and Nakamura [24,25,26] and Takahashi et al [27]. Kizilkaya and Gupta [28], Udomsawat and Gupta [29,30], Nakashima, Kojima and Gupta [31].…”

Integrating Lean and Agile Strategies into the Production Control System for Mixed-model Production Lines

“…Therefore, in order to minimize twip while maintaining wt at less than a certain level, the allocation of work elements and/or the buffer size should be altered in response to the changed r. In responding to the changed r, altering only the buffer size is insufficient, but the allocation of work elements should be also altered. For example, in order to minimize twip while maintaining w/<0.20, the buffer size should be altered from (2,5) to (2,6) in response to changes in r from [0,1.4) to [1.4, 0.2] under allocation 1. In order to respond to changes in r as r>0.2, it is difficult to maintain >v/<0.2 even if the buffer size is altered.…”

“…Various control charts have been developed, and the EWMA chart [5] is utilized to detect small changes quickly. In this paper, the EWMA chart is utilized to detect unstable changes in demand like Takahashi and Nakamura [2]. At first, the monitored time series data dij are filtered by exponential smoothing, a kind of lowpass filter, and the EWMA ///y are calculated as follows;…”

“…In the previous literature on production control systems for single-model production lines, systems based on lean and agile strategies have been proposed (Hopp and Roof [1], Takahashi and Nakamura [2]). In the systems, just-in-time (JIT) ordering based on lean strategy is included, and the function to detect unstable changes in demand and adjust the buffer size, that is based on agile strategy, is also included.…”