The Energy Flexibility group develops the algorithms, models, and market mechanisms needed to make electricity demand an active participant in grid balancing. As grids incorporate more variable renewable energy, the ability to shift, shed, or reshape loads in real time becomes increasingly valuable — and increasingly necessary for reliability and cost-efficiency.
Our work spans three scales: individual assets (EV chargers, HVAC, water heaters, industrial equipment), aggregations of assets managed by utilities or aggregators, and grid-level market clearing and dispatch. Methodologically we draw on stochastic and robust optimization, model predictive control, reinforcement learning, and mechanism design.
Optimal scheduling of EV charging at the distribution level, balancing grid congestion, user preferences, and renewable availability using online convex programming.
Hierarchical control for aggregating rooftop solar, battery storage, and flexible loads to provide ancillary services to the bulk power system.
Hybrid neural network architectures that embed power-flow constraints to improve short-term load forecasts under weather and behavioral uncertainty.
Market mechanisms that incentivize demand response in large commercial buildings while preserving occupant comfort and equipment lifetime.