Multi-objective hybrid genetic algorithm for energy adaptive production scheduling in job shops

Abstract The energy supply of modern economies is becoming more volatile due to increasing integration of renewable energy sources. This necessitates the implementation of demand response programs for the industrial sector to make it financially worthwhile for companies to make optimum use of renewable electricity. Optimal production scheduling, considering operational objectives as well as energy prices will therefore become more relevant. There are multiple options to create flexibility in production differing in their industrial relevance; in this paper, changes in production sequence, load-shifting by rescheduling production pauses, and automatic adjustments of machine energy modes for additional energy savings are considered. An algorithm for energy adaptive production scheduling which can be used for scheduling job shops in the metalworking industry is presented. The algorithm considers two objectives: makespan is used as the operational objective function and a time-of-use based pricing scheme for energy prices and therefore indirectly greenhouse gas emissions is the energetic objective function. Both functions are optimized simultaneously using the non-dominated sorting genetic algorithm-II (NSGA-II). To achieve near optimal results, two methods are used in conjunction to initialize the population. Part of the initial solutions are created by an adjusted priority dispatch rule while the rest is initialized randomly. Due to the pareto optimization, potential users of the developed algorithm can weigh energy cost and makespan to suit their specific needs. Results depend heavily on the modeled production system and energy pricing; however, in the model factory ETA-Fabrik potential savings of about 6 % of energy cost, resulting in a greenhouse gas emissions reduction have been demonstrated with acceptable increases in makespan.