Long-term forecasting of frequency ancillary services in the spanish power market
The increasing integration of variable renewable energy sources in the Spanish power system is driving the need for greater system flexibility and reshaping the dynamics of power markets. In this context, frequency ancillary services play a crucial role in maintaining grid stability and can also rep...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/444044 |
| Acceso en línea: | https://hdl.handle.net/2117/444044 |
| Access Level: | acceso embargado |
| Palabra clave: | Electric power -- Spain Renewable energy sources Energy storage Energia elèctrica -- Espanya Energies renovables Energia -- Emmagatzematge Àrees temàtiques de la UPC::Energies |
| Sumario: | The increasing integration of variable renewable energy sources in the Spanish power system is driving the need for greater system flexibility and reshaping the dynamics of power markets. In this context, frequency ancillary services play a crucial role in maintaining grid stability and can also represent a significant source of revenue, as generators and demand-side participants are able to participate in dedicated markets. Developing accurate and robust forecasting methodologies is essential for assessing business cases and maximizing revenue from services such as aFRR (automatic Frequency Restoration Reserve) and mFRR (manual Frequency Restoration Reserve), especially for Battery Energy Storage Systems (BESS), which are expected to stack revenues from ancillary services alongside those from the energy market. This thesis investigates the use of machine learning-based models for forecasting frequency ancillary services in the Spanish electricity market, utilizing historical data from Red Eléctrica and long-term power price projections from DNV. A comprehensive exploratory analysis of the aFRR and mFRR markets is conducted, examining the evolution of prices and volumes and identifying correlations with other market variables such as DayAhead Market prices, generation, and demand. Forecasting models are developed using DNV’s proprietary tool (PLT) and validated against historical data to predict hourly prices for both services and aFRR power volumes. Model selection is based on performance metrics (MAPE, RMSE, MdAPE) and qualitative alignment with expected market behavior. This work improves DNV’s long-term power market forecasting capabilities and provides inputs and results for consulting studies focused on revenue potential from ancillary services. Finally, several potential improvements and extensions to the methodology are discussed in the conclusions, laying the groundwork for future developments in this field. |
|---|