Estanqueiro, AnaCouto, AntónioSchimeczek, ChristophLopes, DuarteAlgarvio, HugoPreto, IsabelKochems, JohannesSantos, TiagoFaria, RicardoSperber, Evelyn2024-12-302024-12-302024Estanqueiro, A., Couto, A., Schimeczek, C., Lopes, D., Algarvio, H., Preto, I., Kochems, J., Santos, T., Faria, R. & Sperber, E. (2024) New forecast tools to enhance the value of VRE on the electricity markets - 2nd Edition: Deliverable D4.9. Project report of WP4 - Development of Open-access Market Simulation Models and Tools, Deliverable nº D4.9. LNEG, 2024, 69 pp.http://hdl.handle.net/10400.9/4422Project TradeRES - New Markets Design & Models for 100% Renewable Power SystemsABSTRACT: The present deliverable was developed as part of the research activities of the TradeRES project Task 4.4 - Enhancing the value of VRE on the electricity markets with advanced forecasting and ramping tools edition 2. This report presents the second edition of deliverable 4.9, which consists of the description and implementation of the forecasting models aiming to identify and explore the time synergies of meteorological effects and electricity market designs explored in the project to maximize the value of variable renewable energy systems and minimize market imbalances. An overview of key aspects that characterize a power forecast system is presented in this deliverable through a literature review. This overview addresses the: i) forecast time horizon; ii) type of approach (physical, statistical or hybrid); iii) data pre-processing procedures; iv) type of forecast output; and v) the most common metrics used to evaluate the performance of the forecast systems. While in the TradeRES project work package 3 the conception of new market designs and products are presented from a theoretical point of view, in this deliverable, the power forecast tools capable to address the new designs and products are presented and discussed. Complementarily to the first edition of this deliverable, the link between dayahead market time frames and the performance of the different power forecast approaches is analysed. This second edition of D4.9 also focuses on the short-term forecasts (below six hours) for new market designs. As a first step, a non-disruptive change in the day-ahead market is proposed by simply postponing the gate closure hour according to the meteorological data availability from the global numerical weather prediction (NWP) models while the 24 hours forecast periods are still used. In the second step, various short-term forecast approaches designed for time horizons below six hours are developed and implemented. These approaches are specifically tailored to attend the requirements of new electricity market designs currently under development in TradeRES. Another aspect regarding the meteorological time synergy and electricity markets analysed in this deliverable is the identification of extreme events. A wind power ramping forecast approach implemented in the TradeRES forecast tools is described. This approach is designed to complement the existing deterministic power forecasts and it can be used to increase the transmission system operators’ awareness level and helping them to better scale the level of reserve required. Market players can also take advantage of this information to strategically define the bids in the different market environments. Using different wind and solar power parks in Portugal, as well as the national aggregated Portuguese and German wind and solar generation, results regarding the potential certainty gain effect from changing the day-ahead market gate closure are presented and analysed in this deliverable. Results showed that the use of the TradeRES forecast methodology guaranteed better performance compared to an operational forecast from a forecast provider. Additionally, the results emphasized the benefits of including non-traditional variables such as air pressure and temperature at different heights, atmospheric boundary layer, and geopotential height for various pressure levels. The simulations also highlighted that incorporating both NWP features based on historical power series led to improvement when compared with models based solely on power series or NWP. Therefore, it is recommended that power forecast systems can have access to recent observed values to improve their accuracy. Despite the improvements achieved in the forecasts for the day-ahead market, high power forecast errors are still observed (a normalised root mean square error of nearly 30% for wind and solar in Portugal and nearly 20% for Spain). Market designs with shorter forecast time horizons can significantly reduce power forecast errors. Results also emphasize the importance of evaluating the most suitable forecast approach based on the forecast time horizon. To assess the value of renewable energy forecasting for the German day-ahead market, the Agent-based Market model for the Investigation of Renewable and Integrated energy Systems (AMIRIS) was enhanced to account for power forecast errors. For this purpose, a feature was developed that allows for the adjustment of forecasts for the feed-in of renewable energies using a Gaussian distributed error term. Furthermore, this deliverable presents a realistic forecast time series that was implemented in AMIRIS. The case study of the German day-ahead market in 2019 demonstrated that realistic power forecasts can reduce the profits of onshore wind turbine operators by approximately 8% compared to perfect foresight of wind infeed. Assuming Gauss-distributed errors, the losses are smaller (~ 5 % less profit compared to the perfect forecast). The power forecast tools developed in this task will be publicly shared and disseminated in the channels of the project. With this step, users can use the tools for obtaining power forecasts in future studies or use the approaches developed in TradeRES as a benchmark.engElectricity marketsOpen access marketsAgent-based modelingRenewable energy sourcesreport