Modeling user charging behavior in EV networks using M/M/n queueing systems

Abstract

EV charging networks face congestion due to behavior-driven demand. Using k-means segmentation of real sessions, an M/M/n multiserver queueing model, and discrete-event simulation, the study shows that three user segments remain stable with near-zero waiting times, while long daytime AC sessions drive utilization to 0.88–0.94 and increase expected waiting to 70–120 minutes. The results indicate that managing the behavioral mix of demand is more effective than uniform capacity expansion. across charging locations.