Flexible preventive maintenance planning for two parallel machines problem to minimize makespan

Abstract

PurposeThe purpose of this study is to determine simultaneously the period of preventive maintenance and the sequence of each job for two parallel machines problem so that the makespan is minimized. With proper planning of preventive maintenance, a shop can avoid lack of flexibility between maintenance planning and production scheduling, and thus concentrate on production efficiency.Design/methodology/approachThis study addresses a scheduling problem, wherein each machine has to be shut down for maintenance during a maintenance interval [C, T] arranged in advance. The start time and end time of the preventive maintenance are the decision variables. A maintenance action is attended by a server who managers only one machine at a time. Three cases are studied in this paper: the unequal lengths of unavailable periods on both machines, the equal lengths of unavailable periods on both machines, and no waiting time is allowed between the two unavailable periods. Each case is solved optimally by an analytical algorithm developed in the study.FindingsAlthough having exponential time complexities, all the proposed algorithms are quite efficient in solving large‐sized problems. Computational results show that it is able to generate the optimal solution for large sizes (up to 10,000 jobs) in a few minutes of computation time.Practical implicationsMost of the papers that dealt with preventive maintenance assumed that the maintenance period is known and specified. However, the flexible preventive maintenance problem, in which the start time and end time of the preventive maintenance are the decision variables, often occurs in the computer center, NC‐machine and IC‐testing machine for job scheduling and repairs arrangement.Originality/valueThe paper presents an efficient approach for each of the three cases of flexible preventive maintenance, which is relatively unexplored in the literature.