Comput. Ind. Eng. | 2021
On parallel dedicated machines scheduling under energy consumption limit
Abstract
Abstract This work studies a discrete manufacturing scheduling problem faced by companies with energy-intensive processes. A scheduling environment with parallel dedicated machines is considered, where each machine consumes electric energy while processing a job. Usually, the manufacturing companies have a contract with an electric utility that specifies the peak energy demand for any fixed metering interval of a day. Violating this energy limit leads to significant penalty fees, which the manufacturing companies prefer to avoid. This paper investigates a class of scheduling problems with such energy consumption limits from theoretical and practical standpoints. First, the computational complexity of four different variants of this scheduling problem was analyzed. Second, an adaptive local search heuristic algorithm was proposed for the general problem. The experimental results show that the proposed heuristic outperforms the state-of-the-art constraint programming model and a mixed-integer linear programming model inspired by the existing literature. For the large instances with 150 and 350 jobs on each machine, the proposed heuristic finds the best solution in 71.43% of the cases, whereas the constraint programming model only in 10.32% of the cases (the mixed-integer linear programming model is not able to provide a feasible solution for these instances).