advantage of TIGHT
hour ahead forecasts
One day before electricity is produced from the 300 MW FPL Stateline Energy Center wind farm on the border of Oregon and Washington in the US Northwest, Marcie Cole of PPM Energy, which buys power from the plant, reserves space for it on the crowded transmission system of the Bonneville Power Administration (BPA). With an intermittently generating resource like wind, scheduling the wind farm's output just a year ago was on a best guess basis. Now, using a prediction tool developed by 3TIER of Seattle, one of Cole's three screen displays shows day-ahead and hourly wind production forecasts updated every ten minutes until an hour before generation time, when the final call is made by one of PPM's real time schedulers.
PPM's Gerry Froese says efforts to get accurate wind forecasts have cut the costs of hourly imbalance penalties levied on its generation by a factor of five, though he declines to go into detail about figures. "We hope to continue to drive costs lower," Froese says. With the September energising of FPL Energy's 162 MW High Winds project in northern California -- from which PPM is buying the output from -- the nation's largest power marketer of wind energy hopes that good forecasting will gain it a competitive edge in these two locations. California is the only state that requires wind forecasting.
The hourly imbalance penalties imposed on generators which err when predicting output -- within the BPA transmission control area in the Northwest -- have not surprisingly fallen heaviest on wind farms. Until August 2002, BPA penalised hourly generation imbalance deviations a whopping $1/MWh, but an agreement with Northwest wind advocates led BPA to devise a system of penalties that is much kinder to wind projects, especially if marketers can more accurately schedule power (Windpower Monthly, September 2002).
BPA allows a 1.5% or 2 MW error, whichever is greater, in a schedule without penalty, but the second band -- up to 7.5% or 10 MW deviation -- charges 110% of market price for the unmet portion of the scheduled power and only gives 90% of the market price for over-producing energy. The third level is above that and harsh. These penalties, which in the BPA control area can still amount to $1-1.89/MWh, represent the potential savings that good wind forecasts promise, says 3TIER's Pascal Storck. They also point out the need for consistently accurate forecasts.
Accuracy in hour-ahead forecasts is crucial, says David Ryan, a meteorologist working for PPM, and not just to avoid imbalance charges. "In terms of how the turbine turns wind into energy, as wind increases it has the ability to create a four-fold increase in power," with just a small increase in wind speed, he says. "We need the [forecast] accuracy in the steepest part of the power curve the most."
If beating persistence -- the tendency of weather in a very short time frame to remain constant -- is the object of forecasting, Storck says, then 3TIER provides the accuracy PPM needs. Its forecasts are 10-25% better than persistence in next-hour schedules and 50-60% better in day-ahead schedules. No matter how good the forecast, however, true savings come down to how the forecast is used, he says. "The value for a power marketing agency is how they use the information. We could give them a one hundred percent perfect forecast that could save tens of millions of dollars, but if they don't act on it correctly, there is no value."
Collection of long term wind data at Stateline -- which now hosts 450 Vestas V47 660 kW turbines -- began in the 1970s when Bob Baker, an Oregon State University meteorologist and now a PPM employee, first erected meteorological measurement towers on the hilly farmland topping Vansycle Ridge. The data collected over the years since, plus FPL's ten minute wind and turbine performance data along with numerical weather prediction models are all blended into a wind forecast.
"We have some idea of how the atmosphere behaves and some ability to predict how the atmosphere changes, so we look for signals from what's happening right now," Storck says. What 3TIER wants as instant data and what it gets is not always the same, however. Storck's ideal would be to get ten minute predictions of wind speed and direction for each turbine at hub height. "Then we could predict the variability of each turbine and explain the variability [of performance] by wind direction," Storck says. Instead, the data comes as aggregate statistics every ten minutes. That allows 3TIER to determine total plant output.
Still looking for more improvements in forecasting and to gain sweeter economics, PPM installed a number of meteorological stations in three compass directions 20 miles from the Stateline wind farm to help anticipate wind direction and speed and thus give more certainty to its hour ahead forecasts, says Froese. PPM also markets the power output from its 24 MW Klondike wind plant in Oregon; because this is upwind of Stateline, it adds one more check station for fine tuning hourly wind predictions.
In addition to Stateline, Klondike and High Winds, PPM markets portions of Wyoming's Foote Creek plant. It will also market output from two more stations due online by the end of the year, including the 44 MW Flying Cloud plant in Iowa and the 51 MW Moraine plant in Minnesota. 3TIER is expected to provide wind forecasting to these projects. The company also services Energy Northwest for its 48 MW Nine Canyon project in Washington and Seawest for its 49.8 MW Condon project in Oregon.