Suhandi, Victor (2014) Analysis of Optimality and Development of Priority Dispatching method to Minimize Total Tardiness for Unrelated Parallel Machines Scheduling. In: 6th International Conference on Operations and Supply Chain Management (OSCM), 10-12 December 2014, Bali.
|
Text
OSCM2014.pdf - Published Version Download (2229Kb) | Preview |
Abstract
Research on jobs scheduling on parallel machines has been very diverse. There are problems in which a job has a different processing time when it was done on a different machine. The problem that arises is the assignment of machine which we choose to do the job and also schedule jobs on machines. The performance measure is total tardiness. We use priority dispatching (PD) method. This method is a heuristic method. This method can handle very large problems and is expected to give good results. This method schedules jobs one by one. Criteria used in selecting a couple of job and the machine which are assigned, respectively, are: jobs that will surely experience tardiness; jobs which have least number of preferences that not to be tardy; least ready time; load of machines that count from the sum of the time of all jobs that still may be done on each machines; and the due date. Assessment results of this method are done by comparing the total tardiness produced by the first come first serve (FCFS) method and generate and test method (G & T). FCFS method is a representation of a very simple method that is frequently used in practice and the results are usually far from optimal. Meanwhile, the G & T method is an enumeration method to generate all combinations and test in order to obtain optimal results, but this method is only able to handle a simple problem. The G & T method uses depth first search (DFS) strategy and combine it with backtracking. If the results of the PD are close to G & T, it means PD method is close to optimal, but instead when they are away from G & T and closer to the FCFS method, it means it is far from optimal. Testing data was conducted by using random numbers for a hypothetical 6 jobs and 3 machines that are still feasible using the G & T. The processing time is generated with random numbers which have uniform (15, 25) and the due date of each job is uniform (30, 50). It is generated to make the possibility of tardiness occurrences. Each machine has to process about 2 jobs so that the due date is given in the range of 2 times of the processing time of each job. The results obtained by the PD method have the optimality closeness of 91.75% (excellent) for the case where each job can be done on all machines. And for the case where each job which cannot be processed on all the machines, it gets the optimality closeness value at 50.42% (less good).
| Item Type: | Conference or Workshop Item (Paper) |
|---|---|
| Uncontrolled Keywords: | scheduling, unrelated parallel machines, priority dispatching, optimality closeness |
| Subjects: | T Technology > T Technology (General) |
| Depositing User: | Perpustakaan Maranatha |
| Date Deposited: | 24 Aug 2016 04:13 |
| Last Modified: | 24 Aug 2016 04:13 |
| URI: | http://repository.maranatha.edu/id/eprint/20855 |
Actions (login required)
 |
View Item |
