CONSERVE: Client Side Intelligent Power Scheduling
We present preliminary work supporting a consumer-side smart grid agent designed to operate autonomously in order to maximize economic reward for its client. The agent must function in an environment of real-time electricity pricing and ubiquitous uncertainty (price profiles fluctuate, client behavior patterns shift) by leveraging advanced scheduling technology that we adapt to residential power management. For those resident loads that can be automatically controlled the scheduling engine will produce advance schedules that respect client preferences and constraints.
In a larger context with multiple such agents within a community, we argue that advance scheduling provides distinct advantages over dispatching systems. Such schedules support collaborative power management between largely self-interested grid agents to realize efficiencies in local power production and consumption.
We report here initial efforts to model and schedule a "thin slice"of the smart home power management problem, yet one of the more complex aspects; water heating via a heater with advanced control features. We present a model of household water heating patterns together with real-time pricing profiles and demonstrate that an LP approach can rapidly generate cost-optimal solutions to this static problem. Issues and prospects for scaling LP approaches to the larger dynamic problem of interest, as well as trade-offs in adopting less computationally intensive heuristic scheduling methods are discussed.