The semiconductor supply chain has a vertical disintegration feature, and each member of supply chain chases for different goals which leads to the inefficiency of supply chain performance and the difficulty in monitoring it. Facing fierce competition along with the need to upgrade the service levels and improve the performance of the whole supply chain, it is necessary for all members to share their inventory information and synchronize their operations. In a typical supply chain management, the supply chain network is designed and configured in the beginning. Once a supply chain routing is decided, the performance in each routing needs monitoring and control. There are several performance metrics which can be developed for the needs of semiconductor supply chain. These include cycle time, level of WIP, product throughput, delivery performance, capacity utilization, yield rates, and so on. However, from the entire supply chain point of view, WIP inventory is mostly easy to monitor and control.
There are two types of inventory control in supply chain management. In a stage-based inventory control system, inventory is managed at each stage only. In an echelon-based inventory control system, inventory is counted from the current stage to the last tier’s inventory. Echelon inventory takes the supply chain inventory into account and provides a better basis for supply chain coordination. However, most of the literature related to echelon inventory control has been focused on the logistics and distribution systems, in which lead time is usually fixed without considering of production constraint. This assumption is not applicable under a manufacturing-oriented supply chain such as in the semiconductor industry, in which lead time is dependent on the factory loadings. As a result, this research extends the traditional echelon inventory control method and applies it to the semiconductor manufacturing. In addition, the concept of CONWIP is adopted for WIP monitoring and control. By combining the above concepts, this research is aimed to develop an effective supply chain echelon inventory control scheme in semiconductor manufacturing.
The first step in the proposed scheme is to construct a two-echelon model with the goal of minimizing the echelon WIP inventory control limits under target service levels. Next, this two-echelon model is extended to a multi-echelon model through a backward iteration algorithm. The proposed scheme has been validated through simulation study.
Our simulation results show that with a lower supply chain inventory, the proposed echelon-based inventory control scheme can achieve higher service levels than the stage-based inventory control scheme. This result is due to the fact that echelon inventory is derived based on the global information of the whole supply chain while the stage-based scheme considers only the individual situation. Another reason why the echelon-based monitoring and control methods are better than stage-based methods is that the correlation of the demand facing in different stages is also considered in the control and monitoring. Thus, the releasing decisions take situations of different members into account. When making the releasing decision, members in the supply chain has to consider not only their own situation but also situations of the other members.
Unlike the traditional methods which production planning, monitoring, and control are conducted within individual member, echelon-based monitoring and control methods make the production planning according to the global information in the beginning, and then control and monitor by considering the statuses of different stages at once. Based on the validation results, conclusions are drawn as follows: 1) The proposed scheme can derive the multi-echelon WIP inventory limits effectively; 2) Compared to the traditional stage-based inventory monitoring scheme, the proposed echelon-based monitoring scheme can obtain a higher service level with lower inventory levels. This is the result of reallocating the WIP limits and utilizing overall information.
