Investigation of workload smoothing in the performance of a dual resource constrained job shop
This study examined the effect of workload smoothing on the performance of a dual resource constrained (DRC) job shop utilizing an integrated system. A simulation study was conducted in two stages.
The primary experiments evaluated the effect of four smoothing rules, three order review/release (ORR) rules, and six dispatching rules at a 90% shop utilization level on six performance measures. The second set of experiments provided a sensitivity analysis using the four smoothing rules, three ORR rules, and the four best-performing dispatching rules from the primary experiments at the 85% shop utilization level.
Analysis of Variance, ranked data, and interaction plots were examined to determine which policies provided the best results for both sets of experiments.
Results of this research indicate the following. (1) Workload smoothing had no significant effect on the performance of the DRC job shop. At the 90% shop utilization level, the ORR provided the necessary control using the Maximum Shopload (MSL) release mechanism, except for the performance measures of standard deviation of flowtime and percent jobs tardy where the Immediate Release (IMR) rule performed better. At the 85% shop utilization level, IMR performed better and the primary decision became the dispatching rule to select, except for the performance measure of standard deviation of flowtime. With standard deviation of flowtime, extreme smoothing and MSL performed better with either the Earliest Due Date (EDD) or Modified Due Date (MDD) dispatching rules. It appears more control is required at the lower utilization level to reduce the variation of individual jobs about the mean flowtime. (2) The Maximum Jobload (MJL) release rule designed for this study performed as well as, but no better than, the MSL rule. (3) The "crossover phenomenon" was not an issue as those rules that did exhibit evidence of crossing over were either rules which were not candidates for consideration in policy decisions, or members of a group of rules that were not significantly different from each other and were considered as a group.