Abstract
Semiconductor is a capital-intensive industry in which equipment costs account for more than seventy percentage of the capital investment in semiconductor test facilities. In an industrial investigation, the machine interference may be 10% of machine time. Hence, there is a need to assign an appropriate number of machines to the operators to minimize machine interference time or labor cost. This paper aims to develop an effective methodology to determine the optimal assignment relationships between the test machines and the operators for different product mixes to enhance utilization for the optimal system performance. In particular, we employed response surface methodology and genetic algorithms to explore alternative assignment rations and thus identify well-performed assignment alternatives for the test machines and operators in various decision contexts with simulation. An empirical study with real data collected was conducted in a semiconductor test facility to validate this approach. The results have shown the validity of the proposed approach in real settings. Indeed, the developed approach has been implemented on line. © 2013 Springer Science+Business Media New York.