Dear colleague,
on 9 June 2016, we held the workshop
Experiences in using Wind Power Predictions
and Gaps in Forecasting Research
in Barcelona, Auditorium, Barcelona Advanced Industry Park,
c/Marie Curie, 8-14, 08042 Barcelona.
We now have the slides online. The entire seminar can be watched at our YouTube channel. The main findings of the seminar are summarised in a paper we wrote for the conference "The Science of Making Torque from Wind" in Munich, 5-7 Okt. 2016.
The workshop is intended to serve as a basis for the work in IEA Wind Task 36 Forecasting of Wind Power. As one can learn from successes, but even more from one’s failures, in contrast to other platforms, this workshop will concentrate on situations where the forecasts were off the mark, the mechanisms behind it, and the meteorology, data or models needed to address it. Additionally, industry requirements not currently or not adequately covered by the forecasters will be another emphasis. In the end, a document detailing the forecast gaps and needs will be drafted and circulated to the interested parties, e.g. the EU Commission or other funding agencies.
To do this, the program is (see below for abstracts or click on the names to get the presentations):
09:00-11:00: State of the Art and planned improvements
Keynote speech by Helmut Frank, DWD, on planned improvements in operational NWP models, and their impact on wind power forecasts (including input from several other met services).
Keynote speech by Ana Rodriguez, Red Electrica de España, on best practice in the use of wind power predictions, and an industry wish list.
11:30-13:00: Industry experience
"The little shop of horror forecasts – incidents remembered vividly by end users"
In this session, the end users of the forecasts will show examples of forecasts gone wrong, the impacts those forecasts had, and suggestions for remedies.
Melih Kurt, Tennet Bayreuth: Challenges of integrating Wind from a TSO Perspective
Efthalia Gogolou, Statkraft Germany: Significance of the forecast in the Short term Power trading
Sophie Guignard, Maia Eolis: 10 years of forecast use.
14:00-15:30: Meteorology gone wrong, and other views from forecasters
This session address recent failures or successes of the forecasts, and the meteorological basics why that happened. Learnings for the future will be presented.
Stanislas Siblot and Aurélien Bendaoud, Companie Nationale du Rhône: Wind power forecasting in a French context : meteorological feedback
Stephan Vogt, Fraunhofer IWES: Intraday Forecasts of the Wind Energy Production for Transmission System Grid Nodes
Gil Lizcano, Vortex: Perceptions on doing forecasting for the wind industry
Corinna Möhrlen, WEPROG: Causes of wind power forecast uncertainty and how we can learn to deal with it
16:00-17:00: What research and development needed to cover identified gaps (in interaction with the audience).
Here, we present previous attempts at a research agenda, correlate them with the learnings of the day, and prepare the input for the new agenda to be circulated to funding agencies. At the same time, the list should be seen to guide the developments in the IEA Task.
Corinna Möhrlen: Preliminary results from the first interviews on the state-of-the-art use og uncertainties in the power industry
Corinna Möhrlen, Gregor Giebel and George Kariniotakis: Current use and gaps in forecasting - a discussion starter
Abstracts:
Efthalia Gogolou, Statkraft Germany: Significance of the forecast in the Short term Power trading
Traditionally, electricity was produced by power plants running either baseload (the full day) or during the peak hours (12-16 hours of higher demand). These were mainly coal and gas power plants and in a smaller extent nuclear plants. With wind and solar power increasing as a renewable energy source in many countries, the energy industry has to adjust its mechanisms to integrate this intermittent power source into the electricity grid. The wind and solar forecast is a part of the expected supply. It becomes apparent, that any fluctuations in the forecast will have an effect on this supply-demand equilibrium and as a result on the expectations that the market assumed and finally on market prices. This presentation will give an introduction in the Short Term Power Trading and highlight the significance of the Power Forecast in the intraday and Day ahead market. Specific examples will be given from the UK and German Market as well as special cases like extreme weather situations that may lead to a big financial loss.
Corinna Möhrlen, WEPROG: Causes of wind power forecast uncertainty and how we can learn to deal with it
In this presentation some cases of extreme weather uncertainty that lead to large forecast errors will be shown. It will be analyzed where the uncertainty stems from and why this uncertainty develops in the way it does seen with the eyes of a power trader or an operations manager in a control room.
The cases will show how uncertainty develops and why weather uncertainty has such large impact on the power production of wind farms. We will focus on different weather situations as well as different scales of weather systems and how these influence the wind power production in different ways.
We will also look into the difference of highly dispersed wind capacity distribution in comparison to concentrated areas with large amounts of power production, especially with respect to transmission.
Finally, the presentation will show tools and methods to interpret and deal with weather and power production uncertainty in the energy market.
Melih Kurt, Tennet Bayreuth: Challenges of integrating Wind from a TSO Perspective
The main focus of the talk will be on feed-in management, including short overview of forecast generation/supply and application in TenneT.
Stephan Vogt, Fraunhofer IWES: Intraday Forecasts of the Wind Energy Production for Transmission System Grid Nodes
Reliable integration of wind power in the transmission network must consider both temporal and spatial variability in the production; it is necessary to deliver spatially resolved forecasts of the summed wind power production of all wind farms connected to a certain grid node.
We will give a short overview of grid node forecasting techniques. In particular, three methods will be addressed which use numerical weather predictions (NWP) and/or online power measurements. The first method estimates power output by feeding NWP wind speed into a generic power curve that does not take into account individual farm characteristics. The second algorithm estimates power using statistical models (here Extreme Learning Machine) trained on historical power measurements and NWP at so-called reference farms to estimate the production of unmeasured neighbor farms. The third method combines the two ideas by weighting the power curve and trained statistical model outputs depending on the distance between unmeasured and reference farms.
For this study, historic power measurements and weather predictions, collected between 2013 and mid-2015, are used to train models for 82 German reference plants. The main result of this work is that the combination of both reference and NWP-only forecasts significantly improves single plant forecast quality, and therefore, the grid node forecast is also improved.
Gil Lizcano, Vortex: Perceptions on doing forecasting for the wind industry
Stanislas Siblot and Aurélien Bendaoud, Companie Nationale du Rhône: Wind power forecasting in a French context : meteorological feedback
Compagnie Nationale du Rhône (CNR), France’s leading producer of energy exclusively generated from renewable sources, runs 3020 MW of hydroelectric installed capacity, owns 33 wind parks for a total installed capacity of 382 MW and 14 photovoltaïc parks for a total installed capacity of 47 MWp. As balance responsible party on French electric network, CNR has been developing since 2008 its own intermittent energy production forecasting tool, named APOGEES. It was designed as a multi-model platform for an operational use, forecasting wind, PV and small hydro power production. Production forecasts are also provided to other electricity companies, in relation to the French and Romanian TSOs.
The main feature of APOGEES is the possibility to adjust the predicted wind speed according to the expertise of CNR meteorologists. Comparison of available NWP output at local and regional scales, knowledge of site characteristics, maintenance and technical constraints are taken into account. As state of the art has not reported much work on wind power forecasting at the wind park level in France, some feedback in this field will be given. The focus will be laid on meteorological aspects, and two interesting meteorological cases over France will be discussed.
Sophie Guignard, Maia Eolis: 10 years of forecast use.
- Short presentation of Maia Eolis
- Our use of the forecasts: history, what we set up, our current needs
- Presentation of a precise case of error