Dynamic Load Scheduling for Energy Efficiency in a Job Shop with On-site Wind Mill for Energy Generation

With growing concern over global warming and increasing fuel costs, manufacturing facilities are seeking strategic and operational changes to achieve energy efficiency in production. On-site energy generation provides a greener and less expensive source of energy but its highly stochastic and intermittent nature makes production planning very complex. Demand Side Management (DSM) for real-time operation has emerged as an effective tool for carbon footprint reduction. In DSM, a local controller installed at the manufacturing facility uses an intelligent algorithm to make dynamic decisions for scheduling production line electrical loads. In this paper, we propose a framework for the intelligent algorithm in a job shop manufacturing facility aimed at minimizing the total electricity cost such that the total order delays are controlled. We consider a manufacturing facility with a wind mill as a source of on-site renewable electricity generation. The intelligent algorithm makes sequential decisions based on weighted objective function involving electricity cost and delay virtual queues while using different information e.g. historical decisions, current state and near future energy expectations from the wind mill. Job shop scheduling problems with traditional objective functions are NP Hard and the dimension of energy efficiency makes the problem more complicated. The intelligent algorithm developed in this paper uses Recurrent Neural Network for predicting electricity from the wind mill for the future time slots. Simulation results confirm that the proposed online algorithm framework for job shop scheduling is an effective DSM approach, outperforming the current industry practices and achieves results comparable to benchmark solutions. (C) 2019 The Authors. Published by Elsevier B.V.