Robust job-sequencing with an uncertain flexible maintenance activity

Abstract

In this study, the problem of scheduling a set of jobs and one uncertain maintenance activity on a single machine, with the objective of minimizing the makespan is addressed. The maintenance activity has a given duration and must be executed within a given time window. Furthermore, duration and time window of the maintenance are uncertain, and can take different values which can be described by different scenarios. The problem is to determine a job sequence which performs well, in terms of makespan, independently on the possible variation of the data concerning the maintenance. A robust scheduling approach is used for the problem, in which four different measures of robustness are considered, namely, maximum absolute regret, maximum relative regret, worst-case scenario, and ordered weighted averaging. Complexity and approximation results are presented. In particular, we show that, for all the four robustness criteria, the problem is strongly NP-hard. A number of special cases are explored, and an exact pseudopolynomial algorithm based on dynamic programming is devised when the number of scenarios is fixed. Two Mixed Integer Programming (MIP) models are also presented for the general problem. Several computational experiments have been conducted to evaluate the efficiency and effectiveness of the MIP models and of the dynamic programming approach.