Evaluation of real time scheduling policies using simulation

The purpose of this research is to investigate the effects of different scheduling operating policies in real- time scheduling. The scheduling policies include due date assignment method, priority rules, process batch method and operator reassignment policy. A specific case of production scheduling-in a hypothetical assembled product manufacturing system was investigated in this research. The simulated production system encompasses fifteen work centres, each containing one to three identical machines and each machine requires one operator with all operations being perfectly efficient on all machine types. The production -system produces finished products as well as components and sub-assemblies. Orders from outside the system arrive for service generated according to the exponential distribution. The orders coming to the system were classified into "priority" and "standard" orders. Processing times at each work centre are statistically independent and uniformly distributed. A computer simulation technique was chosen as the approach method. A computer simulation written in DBASE III PLUS was used to generate the data for analysis. In order to analyse the behaviour of the simulated production system with respect to the different performance criteria, a number of performance measures were selected. These are, mean tardiness, percent tardy, mean work in progress, mean machine utilisation, and mean operator utilisation. The 2x6x2x2 complete factorial is analysed by the analysis of variance (ANOVA) procedure to statistically determine whether due date assignment method, priority rules, process batch method, and operator reassignment policy or their interaction significantly affect the performance criteria considered. Further analysis to identify where significant differences in performance occur is conducted via Duncan multiple comparison test. Based upon the statistical analysis it was found that the relative impact of due date assignment method, priority rule, process batch method, and operator reassignment policy or their interaction for scheduling policies in real time scheduling to be dependent upon the measure of performance considered. In respect of root mean square of tardiness, the scheduling policies involving the slack time remaining (STR) priority rule are the most important of scheduling policies in minimising the tardiness of customer orders produced by the company. In respect to minimising the work in progress, there is no dominant level of due date assignment method, or priority rule or process batch method or operator reassignment policy. However, the scheduling policies involving the variable process batch (VPB) method produce the best result. The scheduling policies involving the variable process batch (VPB) method are the best performers in maximising the utilisation of machine and operator.