A High-Resolution Synthetic EV Charging Dataset for Cold-Climate Distribution Grid Impact Analysis: Trondheim, Norway (2020--2030)

This dataset contains synthetic electric-vehicle (EV) charging data for Trondheim, Norway, covering the period from February 2020 through December 2030. It was developed to support long-term EV charging-demand analysis, distribution-grid impact assessment, transformer-loading studies, charging-demand forecasting, energy-management optimization, charger-capacity planning, and data-driven smart-charging research.

The dataset is based on observed EV charging behaviour from Trondheim collected between December 2018 and January 2020. These historical charging records were used as the behavioural baseline for modelling delivered energy, plug-in duration, arrival and departure patterns, user type, seasonal behaviour, and related charging characteristics. The original observed charging data are not redistributed in this Zenodo record.

Synthetic charging activity was generated using a multi-stage modelling workflow that combines calendar and seasonal features, Norwegian public-holiday information, Trondheim weather information, EV-adoption growth assumptions derived from Statistics Norway (SSB) vehicle-registration data, stochastic daily session-count modelling, Conditional Tabular Generative Adversarial Network (CTGAN) generation, seasonal Kernel Density Estimation (KDE), and post-generation physical-feasibility correction.

Two main files are included. The corrected session-level file contains 256,826 synthetic charging sessions and is intended for analyses involving plug-in and plug-out times, delivered energy, plug-in duration, average charging power, user type, and session-level charging behaviour. The hourly aggregated file contains 76,993 hourly charging-activity records and is intended for grid/load analysis, forecasting, transformer-loading studies, and energy-management applications.

The hourly aggregated file is activity-based rather than a complete continuous hourly time series. Each row represents an hourly interval in which charging energy is delivered. Hours with no allocated EV charging energy are not included. A small number of records extend into early January 2031 because charging sessions beginning on December 31, 2030 may continue past midnight; these spillover records are retained to preserve energy conservation.

The hourly dataset includes temporal and calendar variables, total hourly EV charging energy, equivalent average charging power, active session counts, private/shared charging-energy breakdowns, private/shared active-session counts, Trondheim temperature information, EV-adoption scenario identifiers, and charger-power information.

Weather-related features were derived using Trondheim climate information, including data obtained through the MET Norway Frost API. EV-adoption growth information was based on Norwegian registered electric passenger-car statistics from Statistics Norway (SSB/StatBank). Calendar-related features include weekday/weekend status, seasonal classification, Norwegian public holidays, and other special-period indicators used during synthetic generation.

 The final medium-scenario dataset contains approximately 3.26 GWh of EV charging energy, has a mean hourly charging energy of 42.33 kWh and a maximum hourly aggregated energy of 1559.87 kWh, and contains no violations of the 7.2 kW per-session charger-power constraint after correction.

The dataset is synthetic and should not be interpreted as measured future EV charging demand. Future charging behaviour may differ because of changes in EV battery capacity, charging technology, electricity tariffs, charging infrastructure, travel behaviour, and EV adoption patterns. The hourly load profiles are physically constrained at the individual charger level but are not limited by transformer capacity, feeder capacity, parking-space availability, or site-level grid constraints. Therefore, high aggregated hourly loads should be interpreted as modelled future charging demand rather than guaranteed grid-feasible operation.

This dataset accompanies the related data article, A High-Resolution Synthetic EV Charging Dataset for Cold-Climate Distribution Grid Impact Analysis: Trondheim, Norway (2020–2030). 
https://doi.org/10.48550/arXiv.2608.30199