LAUSR

With the rapid advances of Industry 4.0 technologies, developing rigorous, model-based algorithms for production system performance metrics calculation, control, and optimization has become a critical task to inject intelligence into the smart manufacturing practice, thus facilitating automated decision-making on the factory floor. This paper is intended to contribute to this area. Specifically, we consider serial production lines with finite buffers and machines following the Bernoulli reliability model. In addition, we assume that one can dynamically control the allocation of a shared workforce via real-time production bottleneck identification and mitigation. We first derive formulas to calculate the performance metrics of two-machine systems. Then, we extend the results to multi-machine cases by developing an aggregation-based analytical algorithm, whose accuracy is verified using numerical experiments. Based on this performance evaluation method, we study the optimization of the control policy. In particular, a space reduction technique is applied that decomposes the system into several sub three-machine lines and search algorithms are proposed to identify a good-quality control policy in the reduced policy space. Numerical experiments are used to demonstrate the efficacy of the optimized policy. An illustrative example is given to provide insights into the structure and operation of an effective control policy.