When and where it counts: enhancing demand response in electric vehicle charging

Abstract

Electric vehicles (EVs) offer a promising solution for mitigating the intermittency of renewable energy through flexible charging. Demand Response (DR) has been tested as one of the key demand-side solutions to capture the flexibility potential of EVs in Korea. This study evaluates the effects of DR interventions focusing on temporal factors and station-level characteristics that are often overlooked in existing literature. Using panel data from 558 EV charging stations (EVCSs) in Korea that participated in the DR program (November 9, 2022-April 30, 2023), we develop a CatBoost-based predictive model to estimate counterfactual consumption and isolate DR impacts at the station level. Results show that EVCSs with automatic controls achieve an average reduction of 11.8 % during event hours, while manual adjustments in charging patterns yield only a 0.4 % reduction, underscoring the limitations of voluntary user compliance. Moderately visited EVCSs exhibit the largest reductions in load, suggesting that station-level characteristics such as occupancy rate play a crucial role in DR effectiveness. Analysis reveals that stations with occupancy rates between 25 % and 63 % demonstrate the most substantial consumption reductions, indicating an optimal operational range for DR program effectiveness. DR interventions were the most effective during evening hours for EVCSs with automatic controls, whereas manual adjustments showed no significant variation by time. In addition, intervention effects during the evening hours differ across seasons. These findings provide insights for the development of DR programs that consider temporal variations and imply the need for automation of EVCSs to enhance grid flexibility.