Multi-objective parallel machine scheduling problems by considering controllable processing times

Abstract

This study examines parallel machine scheduling problems with controllable processing times. The processing time of each job can be between lower and upper bounds, and a cost is associated with the processing of a job on a machine. The processing time of a job can be decreased, which may lower the cycle time, although doing so would incur additional costs. This study develops two multi-objective mathematical models, which consist of two and three inconsistent objective functions, respectively. The first model minimizes the total manufacturing cost (TMC) and the total weighted tardiness (TWT) simultaneously, while the second uses makespan (C_max) as an additional objective function. In contrast to conventional mathematical models, efficient solutions are attained using the lexicographic weighted Tchebycheff method (LWT). Experimental results indicate that the LWT yields better-spread solutions and obtains more non-dominated solutions than its alternative, that is the weighted-sum method, which is a widely used yet promising approach to achieve multi-objective optimization. Results of this study also demonstrate that in purchasing machines, the variation in the fixed costs associated with the processing of jobs on machines is critical to reducing TWT. Moreover, using C_max as an additional objective function typically improves TWT and worsens TMC.

Keywords:

• multi-objective scheduling
• controllable processing times
• lexicographic weighted Tchebycheff (LWT)
• pareto optimization