Increased RES Penetration to Isolated Power System

Home » Concept & Methodology » Use cases » Increased RES Penetration to Isolated Power System

Increased RES penetration to isolated power system reducing RES curtailment

Partner: ICCS / HEDNO

Actors involved: RES producers, Aggregators, Forecast providers

Context

System operators of isolated power systems apply curtailment strategies to RES units, for security reasons, reducing the RES penetration. To minimize the energy curtailed and the operating costs while the system is kept flexible and stable, system operators need accurate forecasts of the RES production and of the system load for both day-ahead and intraday periods.

Summary

This use case aims at making estimations of RES curtailment using weather observations in the case of isolated power systems which will result to increased RES penetration and to optimal energy management.

Machine-learning based prediction of the available active power of wind farms subject to curtailment in isolated power systems needs to be applied. A more accurate prediction of the available power can help optimize the forecasting model performance and therefore potentially increase RES penetration in isolated power systems.

Challenge

How to improve set-points and to estimate available RES production?

The improved wind power prediction can help to increase the RES penetration from a wind farm when some wind farm production is lower than the initial setpoint.

The recorded RES production by the meters corresponds to the penetrated power which may be curtailed. Machine learning techniques need to be applied to adjust the wind power timeseries with the power rejection.

Forecasting models should be trained with the adjusted timeseries in order to provide forecasts of how a wind farm would operate without the curtailment.

Approach
  • Combination of different sources and measurements to upscale wind power recordings: The forecast provider receives data from RES plants, from Power system operators and from weather data providers.
  • Deep learning techniques are performed to optimize forecasts under curtailment:
    • Estimation of RES curtailment
    • Calculate the RES forecast based on weather forecast and the adjusted production data from RES plants.

 

Innovative content of forecasting solution
Curtailed power estimation using machine learning.
KPI1 Increased RES hosting capacity
KPI2 Reduced RES curtailment
KPI3  Amount of curtailed energy
KPI4 Revenue losses per production unit due to curtailment