Scheduling the extended machine interference problem

The extended machine interference problem contains two groups of machines. Group I machines, which are not identical in capability, experience interference because only one operator is working to do loading and unloading. Group II machines have no interference problem because each machine has an operator to serve it. A batch of jobs is ready at the beginning of a scheduling period. Each job has to be processed by one of the group I machines with proper capability and then by a specified group II machine.

The objective is an evaluation of sets of decision rules the operator can use to achieve the best performance of this system. Three performance measures are considered which are makespan, average on duty time of four operators and utilization of the group II machines.

Simulation method is used to collect data and the Friedman test is used to compare the effects of different rules. In the first stage, six rules out of 78 heuristic rules are selected. Then, a search involving these six rules yields 52 combinations of rules. Finally, 13 rules from the set of 52 are identified. Seven of the rules achieve nearoptimal makespan and utilization of group II machines; three of the rules can achieve near-optimal average on duty time, and three of the rules can achieve near-optimal for all three performance measures.

Evaluation of the 13 rules is conducted by comparing the rules to a neutral policy. The results show that the makespan is reduced by about 28%. The average on duty time is reduced by about 15%. The utilization of the group II machines is improved by about 39%.