Third Workshop on
Best Practice in the Use of
Short-term Forecasting of Wind Power

The idea of this workshop series is to bring together utility and TSO personnel using wind power forecasts daily, and to share some tips and tricks for making the best use of them. This is flanked with the latest news from research and notes from operational forecasters.

News: the talk slides are available when clicking on the titles. (For previous workshops, see here:  the first and second workshop).

SESSION: Intra-day issues in Germany

Matthias Lange, energy & meteo systems, DE: Current developments in wind power forecasting.
To meet the requirements of the end-users wind power predictions are continuously enhanced to e.g. cover ramp events, manage extreme events, allow probabilistic forecasting, or have a high regional resolution. The presentation will give an overview over current developments, in particular, the EU projects ANEMOS.plus and SAFEWIND.

Arne Wessel, Reinhard Mackensen, Bernhard Lange, ISET, DE: Development of a shortest-term wind power forecast for Germany including online wind data and implementation at three German TSO
As highly intermittent energy source, wind energy needs wind power prediction for a successful integration in the electrical network. Wind power prediction tools help the TSOs for managing the electricity balance on the grid. The ISET developed recently a shortest term forecast system for time horizons up to eight hours in advance and with an actualization period of one hour. The system is developed for the control zone of the electrical grid of complete Germany. It is based on the newest short term weather forecast from the DWD, the Cosmo-DE model, which is updated 8 times a day. This forecast system is currently installed at three of the four German TSOs.

Dietmar Graeber, EnBW Transportnetze AG: Balancing of wind energy with shortest-term forecasts at EnBW Transportnetze AG
For balancing of differences between day-ahead wind energy forecast and real wind power production EnBW Transportnetze AG introduced a new process in July 2009. In the EnBW approach the intraday-markets, specific reserves and traditional secondary and tertiary reserves are used. Critical to the process is the availibility of shortest term forecasts.

Coffee Break

SESSION: Ramp events

Jesper Thiesen, ConWX, DK: Now casting RAMP events
Since the installation of the first offshore parks and the major onshore parks it has been a challenge to the grid operators that the production from these wind parks is often changing very rapidly. These changes are often caused by small scale meteorological systems. On short term (9-120 hours) the probability of ramp events can be predicted by using a number of numerical weather predictions (NWP) or ensemble forecasts. However, on very short term (0-6 hours) the techniques of using meteorological models are outperformed by persistence forecasts. We have been working with statistical improvements for offshore parks, but the only way to improve forecasts on very short term is to use data assimilation on the wind power forecast from one or more surrounding measurements like met masts, other wind parks, and Doppler radar.

John Pease, Bonneville Power Administration, US: Critical Short-term Forecasting Needs for Large and Unscheduled Wind Energy on the BPA System
Bonneville Power Administration (BPA) just completed the 1st 2009 Trimester Review with its two contracted wind forecast providers, Energy & Meteo Systems and AWS Truewind, on the BPA Wind Ramp Event Tracking System project. This 2009 - 2010 R&D project's goal is to forecast large and often unscheduled wind energy in the BPA Balancing Area that is consuming increasingly scarce hydro flexibility needed to support wind. This project is now under consideration to become BPA's generation scheduling and transmission dispatch forecast system in 2010. Important new information was presented to the wind forecast providers at this Trimester review meeting on the challenges of accurately forecasting short-term wind energy to meet BPA needs. This presentation will cover the operational challenges of moving an R&D wind forecast project to a utility scale operational wind forecast system for the largest build out of wind of any utility in North America.

Jeremy Parkes, GarradHassan, UK: Temporal Forecast Uncertainty for Ramp Events.
Wind energy forecasting is an essential tool in facilitating wind energy integration. To enable the forecasts to be fully utilised, forecast uncertainty information is typically provided. This paper focuses on methodologies for generating temporal forecast uncertainty for rapid changes in the production from wind farms. The forecast uncertainty is generated using an ensemble of NWP inputs, statistical processing and adaptive algorithms, to provide a clear indication of the likely timing and amplitude of wind energy ramps.

Lunch

SESSION: Down Under and Over There

Graeme Ancell, Transpower, NZ: Operational uses of wind generation forecasts in New Zealand.
Wind generation forecast accuracy has been identified as a major area for development to enable the connection of large scale wind generation in New Zealand. Present and future uses of wind generation forecasts in the New Zealand electricity market and power system operation will be discussed. Operational experiences with managing the variability of New Zealand wind generation output will be presented.

Deep Juneja, AEMO, AU: Australian Wind Energy Forecasting System Overview.
The Australian Wind Energy Forecasting System (AWEFS) project was established by Australian Energy Market Operator (previously NEMMCO), in 2007, in response to the growth in intermittent generation in the Australian National Electricity Market (NEM) and the increasing impact this growth is having on NEM forecasting process. The project has been financed through the Commonwealth government's Department of Resources, Energy and Tourism (DRET) to develop a wind energy forecasting system and capability. This project provides a vehicle for improving the research, development and application for forecasting of renewable energy.  The project has two broad objectives:
1.        Facilitating the operation of the market through more accurate wind generation forecasts; and
2.        Facilitating research to improve the quality and dimension of the wind generation forecasts over time
This project provides a vehicle for improving the research, development and application for forecasting of renewable energy.  Researchers under licence will be provided access to data to allow them to integrate new forecasting models over time.  


The vendor for the project is the ANEMOS Consortium, a European consortium of 6 partners and is coordinated by ARMINES (France) and OVERSPEED (Germany), who have developed the Wind Forecasting Engine. The Anemos forecasting model combines statistical and physical approaches to get an improved forecast. The AWEFS system produces forecasts from the following inputs:

AEMO hosts the AWEFS system and maintains its interfaces to its existing market systems to give data access to the market participants and to individual wind farms. AWEFS provides a more accurate forecast of wind generation, which is an increasingly significant and highly variable component of total generation in some regions of the NEM. Better forecasts will drive improved efficiency of overall NEM dispatch, pricing, and permit better power system stability and security management.

This AWEFS system produces wind generation forecasts for all NEM forecasting timeframes noted below:

 

Wei Yu, Environment Canada, Yves Gagnon, Universite de Moncton, Christian Masson, École de technologie supérieure Montreal, CA: Wind Energy Forecasting Research and Innovation Project in Canada
The research and innovation program presented is part of a series of projects to develop and assess wind energy forecasting methodologies for the Canadian climate. The research program is a component of the Wind Energy Strategic Network (WESNet) of Canada, while the work is being done in partnership with the Wind Energy Institute of Canada and Hydro-Québec.
After a description of the context of the research program in Canada, the presentation will provide a brief overview of the wind energy forecasting system used in the research work. Results from the validation of the model will be presented for two specific locations with different topographic conditions. This will be followed by examples of application in high impact events such as mountain waves and turbulent flows. Finally, current ongoing work will be described, along with the future plan for the program.
The presentation will also provide a brief overview of the Wind Energy Strategic Network (WESNet, www.wesnet.ca) of Canada, along with the Wind Energy Institute of Canada (WEICan, www.weican.ca).

Coffee Break

SESSION: News from Research

Olivier Texier, Maïa Eolis, FR: Wind Power Forecasting: A practical evaluation study of different wind power prediction services and practical operation expectations.
In the state of mind of the ANEMOS project, Maia Eolis carried out an assessment study of several wind power prediction services. The predictions are setting-up on two wind farms, which present different length of historical data. The evaluation was focus one the performance and reliability of the data.
The performance of the models is evaluated by using a protocol inspired from the one of the ANEMOS project.
In addition the errors in terms of real production is also evaluation and operation expectations are mentioned.
Finally the a look will be given on turbines behaviour during the storm of the 10th February in France.

Corinna Möhrlen, Weprog, DK/DE: First experiences of the new EEG trading rules in Germany and introduction of an ensemble based short-term forecasting methodology for intra-day trading.
In the new German renewable energy law (EEG) from autumn 2008, direct marketing of all renewable energy has become an option. Although the purpose of the law was not necessarily to encourage direct marketing, but rather to allow for the option is has not been a success so far.
The current study will discuss some of the non-technical obstacles for direct marketing of wind power in Germany and some technical aspects required to make direct marketing feasible in the future. The focus will here be on short-term forecasting. The algorithm that will be introduced is based on an ensemble Kalman Filter technique and presents a milestone in short-term forecasting. It is capable of establishing a consistent physical treatment of measurements and hence is able to take the weather development and not only the distance to a measurement into account when extrapolating measurement information into the future. In this way,
persistence forecasts can be outperformed already after 1 hour.

Cristóbal Gallego and Alexandre Costa, CIEMAT, ES: Different approaches on switching-models for wind power forecasting.
Performing forecasts during ramp events has increased the interest on switching-models, which have the capability of detecting different regimes of the wind behaviour. This kind of models perform on-line classification depending on certain hypotheses, like considering the process observable or unobservable. Several approaches have been carried out by the Group of Forecasting of Ciemat, based on Autoregressive, Neural Nets and Markov-Switching models. Some results will be presented, as well as the future research guide-lines.

Bonus presentation: Henrik Madsen, DTU.IMM, DK: New Research Results and Projects for WPPT.

Date and Venue: Hotel Maritim, Bremen, 13 October 2009, 09:00 to 17:00.

Picture:

The crowd at the workshop.The crowd at the workshop.