Informed Prosumer Aggregator Bidding Strategy via Incorporating Distribution Grid Outage Risk Predictions

FERC Order 2222 paves the way for aggregated distributed energy resources (DERs) participation in the wholesale electricity market. A particular DER assumed to be widely available in the future is the distributed prosumer (DP), also called virtual power plant (VPP), which may host PV generation, and stationary and mobile battery energy storage systems, in addition to the on- site passive load. DP aggregation and participation in the day-ahead market ancillary service products (ASPs) require managing uncertainties associated with load consumption and photovoltaic generation, electric vehicle (EV) scheduling, market-clearance prices, etc. Outages in the distribution grid may distort these energy-limited resources from their optimal operating point, potentially impacting their ability to deliver the committed ASPs in real-time. To address these challenges, first, we develop a machine learning algorithm to predict the risk of outages in distribution feeders. Next, we incorporate the distribution feeder State of Risk (SoR) predictions with the bidding model of the DP aggregator to provide an informed decision-making tool for optimal participation in the energy and ASP markets. The simulation results demonstrate the efficacy and scalability of the proposed model in improving the aggregator profitability and preventing penalties for the inability to deliver ASPs due to unexpected energy capacity limits of DP assets.

Automated summary

FERC Order 2222 enables the involvement of distributed energy resources (DERs) in the wholesale electricity market. A specific DER of interest is the distributed prosumer (DP), also known as a virtual power plant (VPP), which can include PV generation, battery energy storage systems, and on-site passive load. In order to effectively participate in the energy and ancillary service markets, DP aggregators need to address uncertainties related to load consumption, photovoltaic generation, EV scheduling, and market-clearance prices. This study proposes a machine learning algorithm to predict outage risks in distribution feeders, and integrates these predictions into a bidding model to improve the aggregator's profitability and prevent penalties for failing to deliver ancillary service products (ASPs) due to unexpected capacity limits of DP assets.