Optimal Demand Response for Distribution Feeders with Existing Smart Loads

Load characteristics play an important role in distribution systems, which are traditionally designed to supply peak load; hence, decreasing this peak can considerably reduce overall grid costs. Basic components of smart grids such as smart meters allow two-way communication between the utilities and customers; in this context, controllable smart loads are being introduced, which allow developing and implementing Energy Management Systems (EMSs) for customers and distribution feeders. Therefore, this paper studies the impact of existing smart loads, in particular Peaksaver PLUSTM (PS+) loads in Ontario, Canada, to reduce summer peak loads for distribution feeders. A Neural Network (NN) model of controllable loads is developed and integrated into an unbalanced Distribution Optimal Power Flow (DOPF) model to optimally control tap changers and switched capacitors, as well as sent signals to programmable thermostats of air conditioners in residential buildings, in particular those associated with the PS+ program. The developed integrated DOPF is tested and validated using a practical system, demonstrating the benefits of using existing controllable loads to optimally operate distribution feeders. Keywords—Demand response, distribution system optimal power flow, energy management system, load modeling, neural networks, real-time application, smart grid.