Final Report Summary - METEORES SERVICES (Turning Wind Energy Meteorology into System Integration Services for Energy Market Participants)
The main objective of MeteoRES Services is to enhance the efficiency of the European internal electricity market with high share of wind power, through the development of services integrating forecasting methods. Services developed through the MeteoRES project focus on advanced forecasting methods for wind power production and electric loads, with the aim of improving the market value of wind power production. To some extent, also forecasting of power generation from photovoltaics has been studied.
One of the key challenges for the further development of wind energy is to ensure that power generation and consumption are continuously balanced. As wind-generated electricity is mostly traded on the Day Ahead Market (DAM), the deregulation of the energy industries in recent years has greatly increased the importance of load forecasting. It is now imperative to improve not only the physical efficiencies of wind power generation but the predictive power and accuracy of both load and production forecasts.
Currently, in unbundled electricity markets, balance responsibility always requires separate forecasts per injection point to the transmission system. As a consequence, wind power forecasts are now needed at single wind farm level. The bases of wind power forecasting are Numerical Weather Predictions (NWP) which provide the wind speed predictions. The wind speed predictions are then post-processed to yield to wind power forecasting. In addition to the deterministic wind power forecast, an uncertainty estimation associated with prediction error is computed based on Ensemble Prediction Systems (EPS).
Wind power forecasting objectives:
• Implementation of advanced algorithms to improve the statistical correction (MOS) of wind power prediction models, using an auto-recursive model with forgetting factor instead of a linear regression algorithm
• Development of an online application to dynamically collect turbine availability planning from maintenance teams to be integrated in forecasting system
• Reduction and quantification of uncertainty on forecasts through the use of ensemble forecasting techniques at wind power prediction level
Load forecasts can be divided into three categories: short-term forecasts for up to several days, medium forecasts covering seasonal influences up to one year, and long-term forecasts of more than one year. Oldenburg University has developed a load forecasting system based on the method of Principal Component Analysis (PCA). The main area of application of this type of model is load forecasting for sub-grids. This requires a light and robust model allowing for the reduction of the number of parameters of the system model to the few most significant parameters which optimally represent daily load patterns.
Load forecasting objectives:
• Adaptation of existing (utility-scale) load forecasting schemes to smaller scale and specific consumption sectors, with special emphasis on intra-day aspects
• Set-up and integration of a commercially available load forecasting service with wind power forecasting tools
• Leverage a load forecasting system for the load forecasting of sub-grids
• Application of methods to a real life study case and evaluation of the market advantages
• Application tool that integrates methodologies to an IT platform
The project has successfully achieved its scientific and technical targets and milestones. These include:
• Demand control: inventory of the state of the art and a prototype set-up for validation
• Wind power forecasting: set-up and validation of basic and advanced wind power forecasting systems
• Commercial exploitation on power markets: a review of valorisation mechanisms
• ICT: approach and architecture for forecasting and communication with loads
• Dissemination & valorisation: website, workshops and publications
http://www.meteores.eu/
One of the key challenges for the further development of wind energy is to ensure that power generation and consumption are continuously balanced. As wind-generated electricity is mostly traded on the Day Ahead Market (DAM), the deregulation of the energy industries in recent years has greatly increased the importance of load forecasting. It is now imperative to improve not only the physical efficiencies of wind power generation but the predictive power and accuracy of both load and production forecasts.
Currently, in unbundled electricity markets, balance responsibility always requires separate forecasts per injection point to the transmission system. As a consequence, wind power forecasts are now needed at single wind farm level. The bases of wind power forecasting are Numerical Weather Predictions (NWP) which provide the wind speed predictions. The wind speed predictions are then post-processed to yield to wind power forecasting. In addition to the deterministic wind power forecast, an uncertainty estimation associated with prediction error is computed based on Ensemble Prediction Systems (EPS).
Wind power forecasting objectives:
• Implementation of advanced algorithms to improve the statistical correction (MOS) of wind power prediction models, using an auto-recursive model with forgetting factor instead of a linear regression algorithm
• Development of an online application to dynamically collect turbine availability planning from maintenance teams to be integrated in forecasting system
• Reduction and quantification of uncertainty on forecasts through the use of ensemble forecasting techniques at wind power prediction level
Load forecasts can be divided into three categories: short-term forecasts for up to several days, medium forecasts covering seasonal influences up to one year, and long-term forecasts of more than one year. Oldenburg University has developed a load forecasting system based on the method of Principal Component Analysis (PCA). The main area of application of this type of model is load forecasting for sub-grids. This requires a light and robust model allowing for the reduction of the number of parameters of the system model to the few most significant parameters which optimally represent daily load patterns.
Load forecasting objectives:
• Adaptation of existing (utility-scale) load forecasting schemes to smaller scale and specific consumption sectors, with special emphasis on intra-day aspects
• Set-up and integration of a commercially available load forecasting service with wind power forecasting tools
• Leverage a load forecasting system for the load forecasting of sub-grids
• Application of methods to a real life study case and evaluation of the market advantages
• Application tool that integrates methodologies to an IT platform
The project has successfully achieved its scientific and technical targets and milestones. These include:
• Demand control: inventory of the state of the art and a prototype set-up for validation
• Wind power forecasting: set-up and validation of basic and advanced wind power forecasting systems
• Commercial exploitation on power markets: a review of valorisation mechanisms
• ICT: approach and architecture for forecasting and communication with loads
• Dissemination & valorisation: website, workshops and publications
http://www.meteores.eu/