A Mixed Integer Nonlinear Programming Model and Heuristic Solutions for an Automated Demand Response System for Large Facilities

Demand Response is utilized around the globe to alleviate the peak demand economically and to manage reliability-compromising emergencies in power systems. Sri Lanka requires an effective Demand Response system to cater the peak demand more economically than dispatching expensive thermal power plants, while minimizing sub-optimal consumption patterns exhibited by consumers during peak demand periods. Therefore, this paper is focused on the development of an algorithm for an Automated Demand Response system for large facilities, which is customized to suit the requirements of the Sri Lankan power system. Under this system, both the utility organization and the consumers are expected to be mutually benefited. This algorithm consists of three levels: deciding on whether or not to execute an Automated Demand Response event for a particular time interval, determining the optimum facility-level demand reductions, and determining the optimum appliance- level demand reductions. Mixed integer nonlinear programming and a heuristic method are used to solve the optimization problems in this algorithm. Results of this algorithm are analysed using a miniature model of the Automated Demand Response system, consisting of fifteen power plants and five industrial and general-purpose facilities.