Scaling the largest VPP made of EVs

Project Information

VPPofEVs

Grant agreement ID: 101145536

DOI

10.3030/101145536

Project closed

EC signature date 13 March 2024

Start date 1 April 2024

End date 31 March 2026

Funded under

• The European Innovation Council (EIC)

Total cost

€ 3 764 750,00

EU contribution

€ 2 499 999,00

Coordinated by

DEFTPOWER BV
Netherlands

Periodic Reporting for period 2 - VPPofEVs (Scaling the largest VPP made of EVs)

Reporting period: 2025-04-01 to 2026-03-31

Summary of the context and overall objectives of the project

As the adoption of electric vehicles (EVs) accelerates, concerns about the strain on the electric grid mount. The influx of EVs charging simultaneously poses challenges for grid operators in managing demand spikes and maintaining stability. Without solutions, this could lead to increased grid instability, higher costs for grid management and hinder the transition to cleaner transportation. In this context, the EIC-funded VPPofEVs project develops a plug-and-play virtual power plant. It offers a platform for smart charging, connecting EVs to energy markets. This platform enables demand response and peak-shaving, optimising grid management for distribution system operators, transmission system operators and Balance Response Parties. Using machine learning forecasting, project coordinator Deftpower extracts flexibility from EVs. EV owners benefit by monetising their vehicles for grid balancing.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

Deftpower developed and validated an AI based smart charging and Virtual Power Plant (VPP) platform capable of aggregating electric vehicles and providing flexibility services to electricity markets and grid operators. By March 2026, the platform scaled to 1.84 million fully onboarded EV drivers, maintained a stable average app rating of 4.5 out of 5, and connected 123,791 vehicles, representing approximately 8.47 GWh of battery capacity and around 70,000 charging sessions per day and has processed over 29 million charging sessions cumulatively

A production ready data infrastructure was established integrating charging session data, vehicle telemetry, charging infrastructure data and electricity market prices. On this basis, two AI models were developed: a departure time prediction model for individual charging sessions and a pool level forecasting model predicting charging demand and flexibility with 91–96% accuracy, varying among pool-level sizes. These models were integrated into an operational AI based Digital Twin that continuously estimates EV charging flexibility and enables forecasting, activation and settlement of flexibility services through partner APIs.

The technology was validated through real world demonstrations. The initial pilot in Norway involved 72 EV drivers and successfully responded to six flexibility signals from a DSO aggregator. After the Volue partnership ended, the demonstration was subsequently expanded through collaborations with Charge Point Operators, Balance Response Parties and Distribution System Operators onboarding more than 940 additional users, of whom 446 completed optimized charging sessions. This includes CPO partnerships in the Netherlands and Germany, a signed pilot with Dutch DSO Alliander and a Forecasting API partnership with ANWB. Operational improvements increased the smart charging session quality score.

Analysis of operational data shows that approximately 67% of public AC charging consumption can be shifted from peak to off peak periods, corresponding to an average flexibility potential of about 14 kWh per EV per session.

These developments enabled the first operational smart charging sessions in the Netherlands, Germany and Norway, where charging is optimised based on energy market signals and driver specific rewards are calculated and applied. By M24, the rewards ledger had paid €11,530 to 4,756 EV drivers. The technical maturity of the platform has resulted in multiple commercial deployments and partnerships with major market actors, including Essent (E.ON) Allego, TotalEnergies, Ubitricity (Shell), Equans, Kia Europe, Fastned, the Dutch DSO Alliander and an exclusive smart charging agreement with ANWB, confirming the platform’s readiness for scale up and its position as a leading provider of EV based flexibility.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The project goes beyond the state of the art by delivering a fully operational, large scale EV based virtual power plant (VPP) that combines AI based forecasting, real time optimisation and direct interaction with energy markets across public and private charging infrastructure. Unlike existing smart charging solutions, the platform enables end to end, interoperable flexibility activation, demonstrated in real market conditions in the Netherlands, Germany and Norway. By the end of the project the platform aggregates 1.84 million EV drivers, 123,791 connected vehicles representing approximately 8,470 MWh of battery capacity, and processed around 60,000 charging sessions per day.

Key advances include production ready machine learning models for departure time prediction and short term pool-level flexibility forecasting, achieving 91–96% accuracy, and an AI powered digital twin capable of aggregating flexibility at fleet, geographical and pool level and continuously estimates available flexibility for energy markets and grid operators. This enables reliable intraday optimisation and portfolio level use cases at a scale currently not available in comparable solutions. Demonstrations showed that approximately 67% of public AC charging consumption can be shifted to off peak periods, representing an average flexibility potential of around 14 kWh per EV per session.

The results demonstrate a scalable, driver centric, incentive based approach to public smart charging, activating flexibility automatically across small and large EV fleets, while maintaining high user acceptance (4.5/5 app rating across more than 30,000 reviews). Further uptake depends on continued large scale demonstrations with energy suppliers and BRPs, broader access to vehicle data through OEM APIs, supportive regulatory frameworks enabling EV flexibility participation in electricity markets, and continued optimisation of AI models using growing volumes of real world charging data.