A decision support algorithm for assessing the engagement of a demand response program in the industrial sector of the smart grid

Abstract

In the industrial sector of the smart grid (SG), a demand response program (DRP) is offered to consumers to motivate them to shift their demand for electricity to the off-peak period. DRP can cause a dilemma for industrial consumers when energy load is decreased since it may disrupt the production process and they may consequently incur losses. Hence, industrial units may choose to accept or reject a DRP. If they choose to engage in a DRP, they may use the available back-up on-site energy resources to access the required amount of energy. Hence, any decision about load curtailment requires a comprehensive assessment of all layers of production and operational management. This paper utilises several methodologies to evaluate the effects of DRP engagement on operational management. Firstly, the Delphi method is employed for extracting and identifying twenty-six criteria embedded in ten operational and production management factors. Secondly, based on these criteria, the production equipment is ranked using the TOPSIS method. This ranking shows which equipment will have less impact on the organisation's profit as a result of participating in a DRP; but, it will not support production and energy planning which is affected by DRP engagement. So, thirdly, a linear programming (LP) model in a discrete scheduling time horizon is proposed which considers the TOPSIS method output and all the constraints imposed by the DRP and the production resources. Finally, based on the proposed methodology, a decision-making algorithm is designed to assist the operation and energy managers to decide whether to accept or reject the offer to engage in a DRP and if they decide to participate, how to best utilize the available distributed energy resources to regain the energy lost. The main contribution of this paper is the proposed methodology which combines the outcome of the Delphi and TOPSIS methods with a linear optimisation model, the effectiveness of which is clearly demonstrated by the sensitivity analysis.