Peak demand estimation in distribution systems: impacts of plug-in electric vehicle charging on load coincidence

Increasing transport electrification is expected to substantially alter electricity demand patterns in distribution systems (DSs). In particular, large-scale adoption of plug-in electric vehicles (PEVs) may modify the temporal coincidence of customer loads, challenging peak-demand estimation methods calibrated under historical demand conditions, such as Velander’s formula. This study investigates how PEV charging affects load coincidence and peak-demand estimation in a real urban DS in Stockholm County, Sweden. A probabilistic Monte Carlo framework is used to generate stochastic PEV charging demand across penetration scenarios ranging from the 2023 baseline conditions to full fleet electrification. The resulting charging profiles are integrated into a detailed DS model, enabling the analysis of system operation, coincidence factors, and the performance of the Velander formulation across different aggregation levels, customer groups, and seasons. The results show that increasing PEV penetration markedly alters coincidence behaviour, with secondary substations exhibiting a pronounced U-shaped response across the penetration range: the median coincidence factor falls from about 0.96 at the 2023 baseline to 0.61–0.62 near 15–20% penetration, and recovers to about 0.91 at full electrification. Consequently, the relationship between annual energy consumption and coincident peak demand changes substantially, leading to systematic errors in peak-demand estimation. To address these shortcomings, a set of analytical and data-driven formulations that explicitly account for PEV demand characteristics are developed and evaluated. These models improve predictive accuracy by up to 70% relative to the classical approach and provide more robust performance across PEV penetration scenarios. Overall, the findings provide new insights into how PEV charging affects coincidence behaviour across DS aggregation levels, reveal limitations of the classical Velander formulation, and demonstrate how alternative formulations can improve peak-demand estimation under transport electrification.