Highlight 1: Decision-making under uncertainties

Decision behavior of prudent demand under future price uncertainties

We show that risk-aware behaviors in demand response originate from non-quadratic discomfort state cost functions and price uncertainty with skewed distributions. We proposed a novel demand response formulation that combines non-anticipatory multi-stage decision making with non-quadratic cost functions, and showed prudence (the positive third-order derivative of the cost function) is the first principle that causes risk-averse decision behavior despite the formulation having a risk-neutral objective. We further proved that, given a non-quadratic cost function, the change in decision action level due to prudence scales proportionally with the skewness of the price distribution. [Link]

Highlight 2: Strategic behavior analysis for energy management

Game-based framework for coincident peak demand shaving

Coincident peak (CP) demand charges customers based on their individual demand at the system’s overall peak time. We proposed a novel game-based framework to analyze the CP shaving problem, developing a theoretical model to examine the impact of strategic customer behavior on system efficiency. The game structure exhibits different characteristics depending on the extent to which customers can shift their demand. We derived analytical Nash equilibrium solutions and analyzed the price of anarchy by comparing it to a standardized centralized peak-sharing model. [Link]

Modeling customers’ subjective behavior on energy trading

Distributed energy resources (DER) shift customers’ roles from consumers to prosumers, facilitating their participation in the energy market, where their subjective behavior impacts market operations. We modeled subjective prosumers in energy trading using prospect theory and stochastic game models, and we developed a solution algorithm that combines fitting, Markovian processes, and heuristic methods. [Link]. Using a behavior trial dataset from Ireland, we implemented a learning-based method to understand decision-making behavior and then embedded this behavior into the subsequent energy trading game model. [Link]

Highlight 3: Equitable sustainability energy solutions

Demographics-driven equitable electrification path (ongoing work)

New York City is projected to experience nearly double its electricity demand over the next two decades due to the electrification plan, requiring a 60% increase in investment in energy infrastructure. A top priority is balancing efficiency and affordability. We first analyze the distribution pattern of the current energy infrastructure and identify the associated socio-technical impact. Then, we proposed a demographics-driven metric to prioritize investment at the census tract level. This metric offers a novel approach to grid planning and supports an equitable path toward electrification.

Equitable pricing tariff schemes with price response behavior

Time-varying pricing tariffs incentivize customers to shift their electricity demand and reduce costs but may increase the energy burden for those with limited response capabilities. We proposed a joint learning-based identification and optimization method to design equitable time-varying tariffs. This method connects learning and optimization by embedding customers’ price response behaviors, captured by a learning network structure, into the tariff design optimization. The equitable tariffs protect low-income consumers from price surges while motivating peak reduction and ensuring revenue recovery for utility companies. [Link]