"It is probably correct that the main source of errors in Eltra's wind energy forecasts originate from the wind prognoses," says Laursen. "I do not agree to the term poor, however. Our forecasts are state of the art. We are leading the pack, so to speak, in numerical forecasts used to predict the wind, compared to our neighbouring countries."
He does acknowledge the severe impact of forecast errors for a power system. From a system operator's perspective, there are two kinds of error. In a quantity error, wind speed is predicted at, say, 9 m/s and it is only 6 m/s. This problem is relatively easy to adjust when planning for regulation power in coming minutes, because system operators can usually see from the amount of actual production when the wind has levelled off, says Eltra's Poul Mortensen.
In a phase error, wind speeds are more or less correctly predicted, but the timing is off. Phase errors bring the biggest deviations and have the most influence on a system. If the wind is predicted to start increasing at six o'clock in the morning, Eltra might prepare for a 5 MW per minute increase in generation from wind turbines by seven o'clock, or 300 MW extra from wind power. But if the wind does not rise until two hours later, Eltra is required to find the 300 MW of missing power from another source by seven o'clock.
"It's the worst kind of error," Mortensen says. "The deviation grows very quickly. It's difficult to plan for. Will the front come now? You wait an hour, two hours -- you can't plan far ahead," Mortensen says.
"Of course there is room for improvement," says Laursen. "It's not that easy to improve these things very substantially in the short term. Development in this area of research is rapid, however."