FREE – Forecasting RenewFREE – Forecasting Renewable Energy in Egyptable Energy in Egypt

Project summary

There is a trend in Egypt and the world towards increased reliance on renewable energy sources. These energy sources are less taxing on the environment and have a minimal effect on global warming compared to traditional hydrocarbon-based energy sources. The two major renewable energy sources are the wind energy and the solar energy. Egypt has formulated a long term plan to increase its share of renewable energy. Egypt possesses vast tracks of land in the desert receiving considerable sunshine that can be exploited in solar energy generation. Also, the Gulf of Suez has some of the best regions of wind energy sources.
Having renewable energy together with conventional sources poses a challenge of how to integrate both. Periods when renewable energy generation is expected to be low have to be compensated by more fossil fuel generation in order to avert shortages or power outages. Forecasting renewable energy output is crucial to keep the delicate balance between both types of energy sources.
Machine learning models have been successfully applied in many forecasting tasks. Machine learning is based on learning relations among the variables from the data in an automated and intelligent way. Certain weather patterns are detected and associated with future changes. Wind measurements at neighbouring sites and rate of change of wind speed are used to forecast future wind speeds. Cloud movements and previous patterns of sunshine and other weather variables are correlated with future changes in sunshine and hence solar energy output.
While renewable forecasts were used for many years, most research was done in Northern countries. FREE aims at research in the meteorological special conditions for Egypt, like the strong concentration of renewable power plants leading to strong ramps in the power output, the specific thermal conditions for the wind regime, and dust impacting the power output from both solar and wind power plants. The increased knowledge generated from the meteo side is used as input to advanced data science tools like gradient boosting regression trees, singular value decomposition or long-short term memory models. The direct outputs from the project could eventually be implemented by project partner ENFOR in a dedicated forecasting model for the Egyptian power company EETC.