Events

Decarbonization of the power sector with human-in-the-loop

The variability of renewable generation is a barrier to the decarbonization of the power sector. Existing methods for coping with uncertainty in power systems focus predominately on the supply side (e.g using energy storage, power imports, or supplementing with controllable fossil generation). However, many of the emerging technologies that consume power have inherent flexibility, meaning control of the demand-side will be possible. This talk will cover the integration of residential demand flexibility into power systems, with the objective of offsetting variability from renewable generation. First, the classical optimal power flow problem will be introduced and the challenges of scaling it to include controllable generation will be discussed. Then two methods will be introduced which both aim to coordinate flexible resources in a scalable manner that preserves consumer privacy. The first method uses formal optimization to determine price signals, while the second uses multi-agent reinforcement learning.