Abstract :
This paper looks at the application of queuing decomposition models to study work flow behaviour under various shop routing assumptions. This approach proves to be valuable in helping to understand variance and covariance effects. Results show that when processing time coefficients of variation are <1, it is beneficial to have work flow that is as unidirectional as possible. However, as the pattern becomes less random, the position of a bottleneck machine with respect to the flow pattern becomes important in addition to its utilization and processing time variance. Understanding the relationships involved can help guide process design and improvement efforts.
