Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Industrial and Management Systems Engineering

Major Professor

Tapas K. Das, Ph.D.

Committee Member

Andrei Barbos, Ph.D.

Committee Member

Changhyun Kwon, Ph.D.

Committee Member

Hadi Charkhgard, Ph.D.

Committee Member

Zhixin Miao, Ph.D.


aggregated demand response, electric vehicles, mathematical program with equilibrium constraints, robust optimization


Grid modernization using advanced metering infrastructure (AMI) will continue to enhance timely communication among the system operator (SO), producers, and consumers. This will further empower the vision of dynamic pricing and demand side management (DSM). The phrase dynamic pricing in this dissertation refers to the practice of disclosing binding prices of electricity just ahead of consumption. As regards DSM, the focus is on collective demand response (DR) by aggregators managing consumers’ loads in smart and connected communities (households, businesses, industries and aggregation of electric vehicle batteries). However, practitioners and researchers alike have expressed the fear that dynamic pricing may cause wild fluctuations in demand, which in turn will adversely affect both the network and market. To dispel this common apprehension and to show that it is possible to treat electricity as any other commodity (where binding prices are declared before consumption), there is a need to develop complementary policies for dynamic pricing decision by SOs and DR actions by load aggregators.

The overarching goal of this dissertation is to examine if it is viable to trade electricity like other commodities, where price is declared in advance and allows the consumers to engage in price responsive demand response actions. To achieve this, two different approaches are developed, one using data driven learning approach and the other using a two-level game theoretical framework. Thereafter, demonstrate both approaches are implemented on a sample interconnected power network and their benefits are highlighted. In the first approach, a comprehensive agent-based methodology guided by data-driven learning model is developed to derive stable and coordinated strategies for dynamic pricing and demand response in smart and connected communities. This methodology is intended to support the policy makers in understanding the joint impact of: 1) the bidding behavior of power producers 2) dynamic pricing by the SO, and 3) DR actions by aggregators managing a variety of consumer loads.

The second approach is based on a robust game-theoretic framework with a two-layer optimization model. The top-layer is a two-stage stochastic model to address day-ahead decisions and the bottom-layer is a robust bilevel model that yields real-time actions comprising hourly dynamic prices by SO and optimal demand response by the aggregators. The two-layer model aims to minimize the cost to consumers while also maintaining SO’s revenue neutral status in the presence of price spikes in the real-time markets.

The final component of this dissertation study focuses on the critical aspects of minimizing disruption in power networks under extreme weather events. An algorithm is presented that allows for optimal islanding of power network to limit the failure propagation during extreme events.