MIT PI: Munther Dahleh, Department of Electrical Engineering and Computer Science
SkT PI: David Pozo, Center for Energy Science and Technology
Power grids are evolving around modern information and communication technologies, leading to a massive proliferation of data amongst different players across the infrastructure. Additionally, the rapid growth of electrification in the transportation sector will bring about new paradigm shifts on how power grids would be operated. If properly managed, electric vehicles (EVs) can provide substantial value to the electric power grid by improving short term imbalance of supply and demand and providing grid ancillary services. However, complex coordination challenges, heterogeneous user preferences, and regulatory constraints can hinder realization of such potentials. In this project we propose to study the foundational theory and design of architectures for coordinating charging of EVs, with the objective of maximizing efficiency subject to the constraints and uncertain dynamics of both the grid and users. Our approach is centered around principles of mechanism design and distributed control—enabled through properly designed markets—to coordinate EV charging via market-based decisions and incentives. As a central part of this project, we aim to address fundamental challenges that naturally arise in creating and operating an efficient and fair model for data and storage technologies. These challenges span the technical areas of statistical machine learning, quantitative economics, and stochastic control and reinforcement learning. The design of data marketplace tools and procedures will allow the development of generalized guides of systemic processes, policies, and incentive design for the grids of the future to host large capacity for EVs and/or renewable generators. In parallel, storage markets would unlock the huge potential from electric vehicles by enabling the provision of ancillary services currently monopolized by conventional generators. At a fundamental level, this project will produce novel mathematical frameworks for data-driven decision-making. At a practical level, this project will show that such frameworks facilitate the best use of a large amount and wide variety of data to efficiently and seamlessly operate future power networks.