(737b) Distributed Cooperative Industrial Demand Response

Demand response (DR) refers to the active adjustment of an electricity consumer’s load profile in response to varying electricity prices and power grid conditions, which has proven to be an effective means to improve grid performance. The industrial sector is particularly well suited for providing DR due to the large sizes of individual industrial loads and the high level of flexibility in many manufacturing processes. Significant advances have been made in the area of industrial DR in recent years, and for power-intensive industries, it has become a crucial factor in maintaining profitability [1].

Industrial loads usually use electricity to manufacture products for their customers, who may operate their own manufacturing processes to further transform these products into higher-value goods. Existing works on industrial DR have exclusively considered interactions between the electricity market and the industrial load while assuming fixed customer demands, hence ignoring potential flexibility on the customer side. As a result, such approaches are only effective for processes that produce storable products and have sufficient storage capacities. However, it is reasonable to assume that just like the industrial load has room to adjust its electricity consumption, its customers also have some flexibility to change their consumption of the products supplied by the industrial load. Therefore, by cooperating with the customers to adjust the product demands, the industrial load may be able to substantially increase its DR capacity.

The major challenge in cooperative DR is that the participants are all self-interested entities, which induces two complicating factors: (1) A customer will only agree to changes in its product demand if it entails some cost benefit. (2) For confidentiality reasons, the customers do not share their operational models with the industrial load; hence, we cannot simply solve a holistic model that integrates all participants’ operational problems. In this work, we address these challenges by establishing a distributed optimization framework for cooperative industrial DR. We propose an optimization model that determines the required changes in the customers’ product demand profiles and the payments that the industrial load needs to make to the customers such that all participants benefit. This results in a nonconvex mixed-integer nonlinear program, which we solve heuristically using the alternating direction method of multipliers (ADMM) [2] in a distributed manner. Results from several case studies show that significant cost savings can be achieved for all participants using the proposed cooperative framework. We further present a set of methods to improve the convergence of the proposed ADMM algorithm and show that near-optimal solutions can be obtained within a reasonable number of iterations in most instances.

References:

[1] Zhang, Q., & Grossmann, I. E. (2016). Enterprise-wide optimization for industrial demand side management: Fundamentals, advances, and perspectives. Chemical Engineering Research and Design, 116, 114–131.

[2] Boyd, S., Parikh, N., Chu, E., Peleato, B., & Eckstein, J. (2011). Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers. Foundations and Trends in Machine Learning, 3(1), 1–122.