Model and control for cooperative energy management

Abstract

Proto/Amorphous Cooperative Energy Management (PACEM) aims to build and deploy a highly scalable system for smart power grids that will enable efficient demand shaping for small-user networks. Two key problems are to provide distributed control algorithm for efficient demand shaping and to provide an incentive structure to encourage both users and the electric power sector to opt-in to PACEM. In this thesis, I address the first problem by designing ColoredPower, a probabilistic control algorithm. I implemented and tested ColoredPower in MIT Proto, building on previous work in spatial computing. Simulations in Proto show that ColoredPower operates within 3% error and provides a stable dynamic response time on the order of minutes. To address the second problem, I provide a model for user and power company incentives in PACEM, in the form of the Colored Procurement Mechanism, which enables further work in optimal algorithmic mechanism design.