A fuzzy simulation model for production control in a semiconductor wafer manufacturing

Batch processing is the mode of production used in semiconductor manufacturing. The key feature of a batch production facility is that a wide range of products have to be elaborated by sharing the same equipment and resources. Although the production system in a semiconductor manufacturing environment is technologically sophisticated and automated, it still requires labor-intensive operations (loading/unloading operations, operator intervention time within a step...). In particular, the production equipment is subjected to preventive maintenance and calibration as well as to unpredictable failures. In this context, the batch production system is submitted to a high level of uncertainty. Fuzzy set theory is therefore an attractive framework for dealing with the "fuzzy" information and the use of fuzzy numbers seems a good alternative to represent the imprecision of this data, better at any case than affecting more or less arbitrary values. A discrete-event model (MELISSA) for simulating semiconductor manufacturing developed in the authors' laboratory was adapted to take into account the fuzziness of operating times. Some typical simulation results are presented and show how the fuzziness of processing times affects the system performance as evaluated by job lateness, job completion times, job waiting times, job flowtimes.<
>