An experimental investigation of scheduling non-identical, parallel processors with sequence-dependent set-up times and due dates

An experimental investigation of factors affecting the scheduling of a system of non-identical, parallel processors using a series of experimental designs was carried out. System variables included were processor capacity relationships, sequencing and assignment rules, job size, and product demand distributions. The effect of the variables was measured by comparing mean flow times, proportion of jobs tardy, and processor utilization spread. Results showed that system loading and set-up times play a major role in system performance. Grouping jobs by product will minimize set-up times and hence mean flow time and tardiness at the expense of controlling individual processor usage. Factors involving processor capacities and assignment rules tend to have no effect on any of the system performance measures. Variability in job size and product demand tended to give flexibility in controlling individual processor utilization.