Smart Power Systems of the Future: Foundations for Understanding Volatility and Improving Operational Reliability
Abstract
This project addresses the impact of the integration of renewable intermittent generation in a power grid. This includes the consideration of sophisticated sensing, communication, and actuation capabilities on the system's reliability, price volatility, and economic and environmental efficiency. Without careful crafting of its architecture, the future smart grid may suffer from a decrease in reliability, volatility of prices may increase, and the source of high prices may be more difficult to identify because of undetectable strategic policies. This project addresses these challenges by relying on the following components: (a) the development of tractable cross-‐layer models; physical, cyber, and economic, that capture the fundamental tradeoffs between reliability, price volatility, and economic and environmental efficiency, (b) the development of computational tools for quantifying the value of information on decision making at various levels, (c) the development of tools for performing distributed robust control design at the distribution level in the presence of information constraints, (d) the development of dynamic economic models that can address the real-‐time impact of consumer's feedback on future electricity markets, and finally (e) the development of cross-‐ layer design principles and metrics that address critical architectural issues of the future grid.
In this work, we developed an abstract framework for examining how different market architectures induce a trade-‐offs between efficiency and risk in power systems with real-‐time pricing. We modeled a system in which consumers with market power dynamically update their electricity consumption decisions in order to satisfy their demands, which can be shifted in time up to a deadline, at minimum cost. The consumers in this model can represent load-‐aggregators with market power, or micro-‐grids that relay on the main grid for balancing their power demand, or a fleet of electric vehicles managed by a profit-‐seeking entity. We considered to extreme market architectures: (a) all such consumers cooperate to minimize their total cost, and (b) the consumers do not cooperate. In this setup, in both cooperative and non-‐cooperative architectures, the consumers behave strategically in the market in the sense that they anticipate their impact on the cost and exploit their market power to minimize their cost. In the cooperative scenario, the cost minimization problem (of the representative or aggregate consumer) can be cast as an average cost dynamic programming problem. In the non-‐cooperative scenario, the optimal response of each agent can be cast as a Markov Perfect Symmetric Equilibrium strategy. We then compared the statistics of the stationary aggregate electricity demand process induced by non-‐cooperative and cooperative load shifting schemes. We showed that although the non-‐cooperative load-‐shifting scheme leads to an efficiency loss, it has a smaller tail probability of stationary aggregate demand distribution. This tail distribution is important because it corresponds to rare and undesirable demand spikes. On the other hand, the cooperative scheme achieves higher efficiency at the cost of a higher probability of demand spikes.
Award ID: 1135843