Chance-constrained co-optimization of demand response and Volt/Var under Gaussian mixture model uncertainty

Abstract

Managing voltage and active load in distribution networks is an increasingly challenging task with the integration of volatile distributed energy resources (DERs) and flexible demands. This paper proposes a two-stage chance-constrained co-optimization framework using a Gaussian mixture model (GMM) to address Volt-VAR optimization (VVO) and demand response programs (DRP). The utilization of GMM in chance constrained optimization CCO (GMM-CCO) approach handles non-Gaussian forecast errors, ensuring network resilience with manageable computational demands. In the first stage, flexible demands, inverters’ reactive power, capacitor bank switching, and battery states of charge are co-scheduled, focusing on minimizing energy loss, reducing grid operational costs, and managing voltage deviations over a four-hour ahead schedule with hourly intervals. The second stage involves intra-hour, near-real-time optimization for VVO to respond to real-time disturbances. Simulations on a modified unbalanced three-phase IEEE 37-node system validate the framework’s effectiveness, comparing it to traditional chance-constrained optimization methods. Additionally, the proposed framework is implemented on the IEEE 69-node system to analyze its scalability and robustness under different levels of uncertainty and varying penetration levels.