Accelerating the energy transition using data science Big Data Expo 12-09-2023 Judith te Selle | Eneco Energy Trade Titus Kervezee | Eneco Energy Trade

To reach climate goals we need (a lot) more renewable power • Eneco is investing heavily to increase renewable power production • Renewable sources of energy are intermittent and uncertain • Electricity needs to be consumed at the same time as it is produced • We need flexibility to make sure supply and demand match each other

“Power system flexibility is the cornerstone of electricity security in modern power systems.” IEA, World Energy Outlook 2020

4

How does this work in practice? 5

Virtual Power Plant Micro-service platform that operates between the market and the assets Data science plays a crucial role in enabling propositions on the platform: forecasting, optimization, trading algorithms Today’s use case: curtailment of B2B rooftop solar installations

No content

Increased solar production leads to imbalance & negative wholesale prices Total production of electricity exceeds demand: • grid stability is at risk • prices on the market become negative Excess solar should be curtailed in case of overproduction Number of negative wholesale prices (cumulative) Win-win-win for the customer, the grid stability and Eneco

We cannot just curtail the solar panels 9 When curtailing we need to know what would have happened if we did not curtail • VPP needs to monitor what would happen if we stop curtailing • Realtime prediction, frequency as high as possible • Compensation for customer of missed production • Monthly batch prediction, 15-minute frequency Machine learning to estimate the what-if production ?

Solar power forecasting 10 Features: • Solar irradiation • Temperature • Row shading • Cloud cover • Relative humidity • Hour-of-the-day • And so on

We need a local solar power nowcast 11 X Y Solar irradiation Temperature Production Measured at customer site Sensors KNMI Weather stations

Clear correlation for customers in the proximity of a weather station

XGBoost outperformed other models (MAE 10.60 kW)

Accelerating the energy transition using data science 14 • Using publicly available data sources resulted in being able to offer this proposition to B2B customers • Currently two customers are connected • This proposition helps us providing flexibility to the grid

No content