Adaptive input control for job-shop type production systems with varying demands using inverse queueing network analysis

This paper describes the application of inverse queueing network analysis (IQNA) to a work-in-process (WIP) controlling problem of the job-shop type production system where the product mix varies periodically. The authors represent the continuously changing production system by a series of steady-state models which reflect its conditions for the respective periods. The WIP level which suboptimizes multiple evaluation functions defined for each period is obtained successively by using IQNA. The computational results of a sample problem on a hypothetical semiconductor manufacturing line are shown and compared with the simulation results under optimized conditions. The main cause of making the difference between the results of analysis and simulation is indicated by comparing the results of two kinds of simulation experiments, i.e. ‘ordinary simulation’ and ‘compulsory simulation’ the latter of which is closer by nature to the analysis. Finally, the method for reducing the difference is provided based on the indication.