After the good success of the first workshop, we decided to have another one, also together with the Thomas Ackermann Workshop on Wind Integration. The idea was 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 was flanked with the latest news from research.
Note: A third workshop will be held in Bremen, on October 13, 2009. See here.
Program:
09:00 Introduction and Notes from Operational Forecasters
Ian Pearman, MetOffice, UK:
Wind forecasting advances at the Met
Office
A brief look at recent developments in Met Office wind
forecasting, from high resolution model nowcasts through to climate change
impacts, to ensemble forecasts and post-processing techniques over the UK,
Europe and worldwide, over land and sea. We are exploring ways to improve
accuracy in forecasting wind and other meteorological parameters at turbine hub
height and thereby to produce reliable forecasts of power output, as well as
flagging potential weather hazards.
Ignacio Martí Perez, CENER, ES: Wind Power Forecasting - An Introduction (and Experiences from the Spanish Market)
Matthias Lange, Energy & Meteo Systems, DE:
Operational Wind
Power Forecasting in Europe, North America and Australia.
In many different regions in world wind power forecasts have
become indispensable for the integration of wind energy into grids and markets.
But, of course, the requirements of the customers differ according to the market
framework and, in addition, the locations of the wind farms are very distinct
regarding their predictability. We show examples of operational wind power
forecasting in three countries: Germany, Canada (Alberta) and Australia. The
same forecasting system Previento was used in all three cases with input from
different weather models. The experience shows that our general approaches lead
to successful
forecasts in different areas which includes the optimization of the forecasts
for the individual applications.
Bernhard Lange, ISET, DE: Experiences from the Development of Wind Power Forecasts for Six European TSOs.
10:30 Coffee break
11:00 Notes from TSOs and Traders
Alan Kennedy, System Operator of Northern Ireland (SONI), UK: Finding the right model for Northern Ireland
Gerardo Gonzalez Morales, Red Electrica de Espana (Spanish TSO): Wind power prediction in the Spanish system operation.
John H Pease, Jr, Bonneville Power Administration, US:
Wind
Power Forecasts in the US context.
Bonneville Power Administration (BPA) is a regional
utility in the Pacific NW (USA), serving a annual load of 9500 MW but managing a
control area of approximately 16,200 MW. BPA is a wholesale, hydro based utility
that must support all Regional intermittency (wind), regardless if it serves a
native load or not. Hydro resources are statutorily obligated to serve preferred
utility customers, yet federal transmission mandates (FERC) require the control
area operator (BPA) to integrate any and all renewable resources that request
interconnection. Consequently, capacity and energy that could be sold at peak to
maximize revenue is diverted to support a real time need (wind intermittency)
for our transmission obligations, potentially impacting the revenue stream of
the organization. The Pacific NW does not have an established wholesale energy
or capacity market.
Today BPA manages 1425 MW of wind that is expected to grow to 3300 MW by 2009
with increasing wind integration in the subsequent years. Due to transmission
constraints in the BPA control area, large and concentrated wind projects are
built in areas where transmission is available. This has created large and
unpredictable wind ramps that must be managed by hydro resources that were not
originally built to serve this need. If wind is integrated at the current rate,
potential reliability issues may occur.
This has led to an ambitious wind forecasting effort to reduce wind forecast
error. BPA completed a two year wind forecasting effort that forecasted hourly
wind from real time to seven days ahead. This effort was successful. Due to
emerging wind ramps in the control area, BPA is now interested in an R&D effort
to forecast wind ramps from 48 hours ahead to real time at high accuracy. These
forecasts will include magnitude, timing and the probability of wind ramps with
a special effort to eliminate false alarms. Along with wind ramp forecasting,
BPA is forecasting wind every five minutes during the operating hour (12
forecasts every 5 minutes) at high accuracy - point wind energy forecasts at
minute 5, 10 and 15 with 1% error up to 3% error at minute 30, which will be
automatically input into BPA's automatic generation control (AGC) system to
determine reserve requirements in real time.
13:00 Lunch
14:00 More news from users
Frank Hochmuth, NUON, NL:
Wind Imbalance Management Cost.
In markets where wind power is subject
to programme and balancing responsibility, the Programme Responsible Party (PRPs),
typically the trading division of an energy utility company, incurs
substantial imbalance costs in the process of selling wind power into the
market. These costs can amount to several Euro per MWh and have to be
charged to the wind power producer, therefore decreasing the contract price
and revenue of the wind farm owner.
These costs can be significantly
decreased by advanced hedging strategies, which in turn are making use of
short-term, medium-term and long-term wind forecasting. This work will
outline the benefits of week-ahead wind forecasting (i.e. 10 day wind
forecasts) that was in practice proven to be highly beneficial for the value
of wind power.
Shanti Majithia, UK National Grid:
Reducing the volatility of predicting wind energy as part of renewable energy.
Wind is widely expected to be the dominant renewable technology
to meet Government targets on renewable energy for 2010 and 2020. What are the
statistical issues and challenges facing the Energy Industry in predicting power
from wind farms? How can these issues be overcome?
15:00 Coffee break
15:30 Notes from Research and Operational Forecasters
Rui Pestana, REN (Portuguese TSO): Dealing with limited connections and large installation rates in Portugal
Jeremy Parkes, Garrad Hassan (UK):
Forecasting For Utility-Scale Wind
Farms, The Power Model Challenge.
To date, a number of studies have focused on
the mathematical modelling techniques for forecasting the production from wind
farms, often focusing on the task of predicting the meteorological conditions at
the site. This talk focuses on the conversion from meteorological data to the
power produced from wind farms. This challenge is particular significant for
Utility scale wind farms, where the simple application of a manufacturers power
curve is insufficient to capture the true behaviour and interaction of the wind
turbines.
As the penetration of wind energy continues to increase around the world, with a trend towards large Utility-Scale wind farms, greater than 100 MW, effective wind energy forecasting will become increasingly important, and sophisticated wind farm power models will have a greater affect on the overall forecast accuracy. This talk summarises typical techniques for generating wind energy forecasts, highlighting some of the alternative methods of converting meteorological predictions to final wind farm production. A detailed assessment of a sophisticated power model is provided, including a description of the method and evaluation of the contribution to overall forecast accuracy when compared with a standard power model. The generation of sophisticated power models by combining theoretical flow modelling with detailed analysis of SCADA data, and statistical adaptive algorithms, enables the forecast error due to the power model to be significantly reduced.
The results presented in this talk are all for live forecasting services, taken from a range of sites in Europe and the USA. The results clearly demonstrate that there are significant impacts on the accuracy of the forecasts due to the conversion from meteorological conditions to wind farm output. Simple power models can introduce an error of the order of 10% Mean Absolute Error (MAE, normalised by wind farm capacity), where as sophisticated power models can reduce this down to the order of 2% MAE.
Nicholas Cutler, University of New South Wales, AU:
Seeing a
bigger picture from NWPs to assist power system management during uncertain
periods.
I will demonstrate a new approach to wind energy forecasting,
founded on the strengths and weaknesses of NWPs. A visually-aided decision
making tool is under development where transformed predicted wind fields are
displayed during periods of high uncertainty in the wind forecasts at specific
locations. Existing wind farm locations are shown on the wind fields to allow
the forecast user to a) decide on the likely nature wind changes at the wind
farms, b) assess the possibility for multiple wind farms to be affected
simultaneously, and c) consider possible timings for the changes taking NWP
uncertainty into account. This information can add value to deterministic or
probabilistic wind power forecasts for managing power system security.
17:00 Close
Location: Madrid, Spain. For hotels please check the relevant website of the Integration Workshop.
Date: 28 May 2008, 09:00 to ca 17:30.
In collaboration with: Seventh International Workshop on Large-Scale Integration of Wind Power into Power Systems as well as on Transmission Networks for Offshore Wind Farms, 26-27 May, 2008.
For all participants, but especially participants from INCO countries (essentially most non-OECD countries) and EU Less Favoured Regions, some sponsoring of the travel can be given on request.