HONEYMOON - a high resolution numerical wind energy model for on- and offshore forecasting using ensemble predictions

Principal Investigator

Dr Jess U. Joergensen and Dr. C. Moehrlen
University College Cork
Sustainable Energy Research Group
Dept of Civil & Environm Engineering
University College Cork
Ireland

jess.joergensen@wanadoo.dk

Project description

Objectives and problems to be solved
Using an operational numerical weather prediction model as a development basis for a real-time wind energy predictions is a sophisticated and yet secure way to meet the challenges of this topic. Such a model makes use of newest methods in meteorological forecasting, but also provides experience in the operational use and maintenance of a prediction system. The NWP model that will be used in this project has been upgraded with a wind power module inside the operational weather prediction model. The coupling of the NWP model with a wave model has also started in low resolution. The promising results of these implementations have prompted to this project. Therefore the lower level objectives have been identified as:

=> Improvement of winds in the boundary layer
=> Improved efficiency based parameterisation of wind power
=> Improved offshore predictions through coupling of atmosphere, wave and ocean model
=> Reduction of large scale phase errors by using ensemble predictions

Description of work
The project consists of 4 parts, three of which are research related. The forth part is the demonstration of the developed model system at various end-users. The end-users are utilities and renewable energy suppliers spread over Europe, namely in Denmark, Germany, France and the UK. A wind database will be used to simulate a real-time setup in Ireland. The scientific part is divided into 4 Work packages:

=> Numerical Weather Prediction Development

• Coupling of a atmosphere-wave and ocean model
• Direction dependent roughness parameterisation
• Diffusion parameterisation

=> Development of an operational ensemble prediction System

• Ensemble System
• Statistical Ensemble Prediction Analysis
  

=> Development of an efficiency based wind power parameterisation

• Power Curve Analysis
• Efficiency based Power Parameterisation
• Spectral Analysis on high frequency data

=> Demonstration and verification of the model system at end-user

• Installation and testing the model at test sites
• Demonstration verification with Statistics
  

The project is designed in a Plug-and Play mode for highest flexibility of the work packages. That means the project is designed to enable simple and yet robust connectivity among the work packages. These are designed similar to stand-alone devices for a common operating system. If the model system is regarded as an operating system, then this means that the model system is independent of the single modules. A default setup makes it possible to operate the model system without the newly developed modules, but also allows plugging them in at any stage of their development. The modules won't be necessarily a complete application but rather a subject based, self-contained piece of software that cooperates with the other modules in a plug-and-play fashion via a defined interface. In this way scientific development and technical implementation, operation and upgrading can be organised hand in hand. This architecture therefore offers pervasive applicability for developer and user, and is an intelligent appliance to allow the extension of the system with the end-users own modules, e.g. GUIs.

This project has the following scientific and technological objectives:

Higher accuracy of winds in the boundary layer in the met services' operational weather prediction model
With more accurate forecasts of production, network operators will be in a position to plan more effectively for the integration of wind generated electricity into the network.

Supplying an uncertainty estimate of power predictions using ensemble predictions
A numerical wind power prediction tool will however not only facilitate network operators in accommodating an increasing penetration of wind-generated electricity. In addition, an implemented uncertainty estimate of the power predictions in the model system will provide an economic measure of the wind power prediction in a "bidding" environment. The ensemble predictions will therefore not only serve to improve the forecast accuracy. It could be delivered not as weather forecasts, but probability forecasts in units meaningful to the customer (e.g. megawatts). For example an ensemble of weather and power forecasts can be transformed into an ensemble of bidding scenarios on the electricity market. For a given level of production, each member of this ensemble can be converted into a profit or loss. The expected profit/loss is the average value of the profit/loss across the ensemble. The production level that maximizes this expected profit or minimizes the expected loss should then be chosen.

Real-time forecasting and maintenance made available for a large clientele
Operational forecasting requires a specific environment. The transformation to the requirements of the customers of a wind energy forecasting system and its maintenance will be handled with a Plug and Play setup of the model system. A major challenge is the portability of the model and flexibility to computer resources of potential clients. The model system by default should be available for large clients such as utilities as well as for smaller clients such as green electricity suppliers, or even wind farm owners. Although the source code of the NWP model that will be used is capable to run on a large variety of platforms, the integrated new model system has to be tested and adopted in a similar way to ensure the envisaged portability.

Increased predictability of offshore winds and waves
The increase of predictability of offshore winds and waves can be obtained through coupling of an atmospheric model with wave and ocean models. Apart from the direct impact on the wind power prediction from the ocean coupling, offshore wind farms will also require prediction of waves and currents for maintenance purposes. Hence, the coupling opens the possibility of delivering offshore predictions for maintenance of offshore turbines and masts to customers.

Expected Results and Exploitation Plans
Using an operational numerical weather prediction model as a development basis for a real-time wind energy predictions is a sophisticated and yet secure way to meet the challenges of the wind energy forecasting. Such a model makes use of newest modeling methods, provides experience in operational use and maintenance. The demonstration sites where the new model will be tested range from markets with high growth rate (UK, France, Ireland) to well-established markets (Denmark, Germany) with high wind energy deployment. In these areas, accurate and timely predictions are essential. In the markets with high growth rate it is also essential to verify constraints and prepare for the future. In this way new markets can grow sustainable and previous failures can be avoided. The choice of a combination of well-established and growing markets for the demonstration of the project seems to be most beneficial in the prospect of sustainable economic development and the ambitious goal of developing a fully portable model system.

Final report

Additional information

Project started in 2003 - finished 2005.