Dr. Kazuo Miyashita, a senior researcher of the Intelligent Systems Research Institute of the National Institute of Industrial Science and Technology (AIST) developed a production control scheme for extensively reducing the manufacturing cost of semiconductor products in collaboration with Hitachi East Japan Solutions, Ltd. (Hitachi-TO).

The AIST developed an ultra-fast production simulation technique to find the optimum level of in-process inventory which constitutes a key factor for reducing the manufacturing cost of semiconductor products and also providing robustness against variation in production volume. The Hitachi-TO developed a distributed production control technique to keep always the optimal inventory volume by autonomously adjusting the production in each of process steps, against variations in production due to failures in the production facilities.

It was nearly impossible to set the optimum inventory level on a case-by-case basis, as required for carrying out the production schedule to meet the demand in the large-scale, complicated manufacturing process for semiconductors, because of long time taken by the production simulation and frequent occurrence of defects in the process. Consequently, the management of in-process stock failed and the manufacturing cost rose owing to overstocking.

Under such a circumstance, the present control system computes the optimal inventory level within 20 minutes, on the basis of new simulation technique 20 to 100 times faster than the conventional method, and using the Genetic Algorithm (GA) and the cluster computing, and the distributed production control allows managing the stock in the manufacturing process always at an optimum level, by keeping the most appropriate inventory in real time for each of process steps. As the production control suppresses the stock level in the semiconductor manufacturing process below a half of the conventional one, the cost of semiconductor production is reduced widely, and the international competitiveness of the Japanese semiconductor industry is expected to be boosted in a quantum leap.

It is planned to embed the present production control scheme into the semiconductor manufacturing execution system (MES), aiming at application to the actual semiconductor manufacturing process. Furthermore, the technique is to be extended to the manufacturing process of high-tech device components other than semiconductors, such as liquid crystals and plasma displays.

The details of the present work will be reported in the 2003 Winter Simulation Conference (WSC’03) to be held on December 7 to 10, 2003, at New Orleans, US., and also exhibited at the Open House 2003 of the Intelligent Systems Research Institute, AIST, to be held on November 13, 2003 at Tsukuba, Ibaraki. Some of the results of this work are being filed for the patent application.

Fig. 1. Background of the present study