An Optimization Model for the Dynamic Production Scheduling Problem with Machine Failures and Energy Constraint

Abstract

Under uncertain machine-failure disruptions, Flexible Manufacturing Systems (FMSs) must be supported by a strong scheduling optimization model that is able to guarantee satisfaction of due-date requirements by relying on threshold scenarios. This paper proposes a complete robust optimization approach to dynamic production scheduling in the presence of uncertainty in machine failures, minimization of energy consumption, and compliance with production deadlines. The challenge of dynamic scheduling in a flexible manufacturing workshop with machine failure disturbances is to update the production plans quickly, with multiple objectives to be achieved — such as completion time, energy consumption, and schedule deviation — that are often competing, thus greatly complicating the computation. A Mixed-Integer Linear Programming (MILP) model was developed that explicitly considers the degradation in machine reliability, varying energy consumption in machine operational states, and real-time rescheduling triggers. The changing machine failure rate is added to the integrated optimization of job shop production scheduling and predictive maintenance, and the machine state is predicted based on the processing time of the current job. Flexible flow-shop systems scheduling with uncertainties in processing time is optimized using robust optimization and series-parallel production system configurations are taken into consideration in the context of preventive maintenance policies. The proposed framework is based on scenario-based robust optimization, using hybrid dynamic rescheduling triggers based on events and periods. To address the issues of rescheduling problems, namely how to reschedule a job and when to reschedule, a hybrid dynamic rescheduling trigger strategy (with four judgment mechanisms) has been developed. Based on computational experiments on the benchmark instances, the results confirm that the model outperforms the deterministic and traditional robust methods in terms of computational performance. The model reduces the make span by 12%–15% on average and reduces the energy consumption under disruption scenarios by 8%–14%. The findings provide a realistic approach for manufacturing systems to meet production efficiency, equipment reliability, and sustainability goals while operating under uncertainty.