Thesis Defense: Market Design Opportunities for an Evolving Power System

Thursday, December 5, 2019 - 2:00pm to 3:00pm

Event Calendar Category

LIDS Thesis Defense

Speaker Name

Ian Schneider

Affiliation

LIDS & IDSS

Building and Room number

E18-304

Abstract

Thesis Committee:
Munther Dahleh (Chair & Supervisor)
Mardavij Roozbehani (Supervisor)
Paul Joskow
 
ABSTRACT
The rapid growth of variable renewable energy is transforming the electric power system. Variable renewable energy, specifically wind and solar power, has grown five-fold since 2009; it constitutes a quarter of all electricity production in several U.S. states. Renewable energy brings benefits but also new challenges. Wind and solar energy production are variable and uncertain, subject to weather changes that are challenging to forecast. These resources are non-dispatchable; their energy output cannot be fully controlled. At scale, this poses technical challenges for the electricity grid and techno-economic challenges for electricity markets.
 
Improvements to electricity markets and to grid operations can help reduce the cost of a reliable low-carbon power system. This thesis covers several topics to help enable well-functioning electricity markets in systems with high levels of renewable energy. First, this thesis develops a game-theoretic model of producer strategy in electricity markets with high levels of renewable energy. It demonstrates how uncertainty, correlation between stochastic resources, and public forecasting impact producer strategy. It uncovers new issues that could impact market power in systems with high levels of renewable energy. Second, the thesis models and explains the effects of retail electricity competition on producer market power and forward contracting. I show that increased retail competition could increase producer market power and decrease forward contracting; these issues could be important considerations for policies that relate to retail competition. Finally, this thesis proposes new methods for improving incentive-based demand response programs. Demand response programs pay customers for reducing consumption when the marginal cost of electricity is high, but they are challenged because utilities have imperfect information regarding customer demand. I show how tools from online learning can be used to improve the sequential decision problem of choosing customer baselines in demand response programs. This could help improve demand participation in energy markets, which could ultimately help reduce the costs of operating a low-carbon power system.