A stochastic game-theoretic optimization approach for managing local electricity markets with electric vehicles and renewable sources

In response to the growing need for sustainable and environmentally friendly energy ecosystems as well as the rapid adoption of electric vehicles (EVs) and renewable energy sources (RESs), local electricity markets (LEMs) offer a distinct potential to improve grid stability, promote clean energy, and reduce costs. However, harnessing the full potential of LEMs requires addressing multifaceted challenges associated with market dynamics, and the integration of EVs and RESs. Our study presents a stochastic bi-level model for managing LEMs, where the local transaction center (LTC) and other market participants (agents)—load aggregator (LA), charging station (CS), and their lower-level prosumers—aim to maximize their profits in a competitive game-theoretic structure. To ensure grid stability and profitability, LTC as the leader seeks to design a robust dynamic pricing strategy under uncertainty, whereas LA and CS, as followers, decide on the energy transaction amounts under uncertainty in their generation and consumption. We address several novel aspects of LEMs, such as heterogeneous behaviors of entities, uncertainties in renewable generation and load profiles, and the consumption profile of the EV fleet. We propose a centralized solution approach based on Karush–Kuhn–Tucker optimization and a decentralized approach based on a hybrid Genetic algorithm for solving the LEM management problem. The centralized approach provides optimal solutions whereas the decentralized approach provides near-optimal solutions much faster while maintaining the privacy of market participants (agents). Numerical results based on a realistic case study demonstrate that increasing the capacity of LTC's energy storage and its initial state of charge increases LTC's profit by 255%. Results also demonstrate that modeling the LEM under uncertainty as a two-stage stochastic program provides a robust pricing decision resulting in a 433% higher profit compared to the deterministic modeling that ignores the uncertainties in the key parameters.