Capturing diversity in electric vehicle charging behaviour for network capacity estimation

This paper proposes a stochastic data-driven model for uncontrolled charging that accurately captures diversity in individual consumer behaviour. This is important because understanding the diversity between consumers is necessary to accurately estimate the number of electric vehicles? charging a distribution network could support without reinforcements. The model combines readily available travel survey data with high resolution data from an electric vehicle trial, using clustering and conditional probabilities. We demonstrate through a case study of UK residential charging that existing approaches may overestimate the increase in peak distribution network demand by 50%, which has implications for assessing the cost of network investments required. We also show that the peak charging demand varies regionally from 0.2?1.4 kW per household, demonstrating the importance of using locally representative vehicle usage data.

Automated summary

This paper proposes a stochastic data-driven model that accurately captures diversity in individual consumer behavior. Through a case study of UK residential charging, it demonstrates that existing approaches may overestimate the increase in peak distribution network demand by 50%, highlighting the importance of using locally representative vehicle usage data.