Taming Wind Power

Xcel Energy purchases energy from NextERA’s Limon Wind Farm in CO (Image credit: Xcel Energy)

At midnight on Christmas Day in 2007, Xcel Energy had just upgraded a nascent wind generating capability on its Colorado electrical system. When a huge winter storm pushed through the area and delivered three-fourths of its entire capacity in little more than an hour.

It was a literal shock to the system.

“We call it the Santa Claus ramp,” said Drake Bartlett, Renewable Energy Analyst for Xcel Energy, the Minneapolis-based utility (a ‘load ramp’ is a measure of how quickly generating capacity can be dialed up or down).

The company had been basing its load (or demand) requirements on weather forecasting “forever,” according to Bartlett, determining power needs dictated by temperature and time on a day-ahead basis with an hourly average error rate of 1-2%.

The Santa Claus ramp demonstrated that it had to know a lot more about the variability of wind as a source of energy (or supply).

Xcel had been using publicly available free information, which meant that it was guaranteed to be inaccurate, if only because it provided surface-level measurements when the average wind turbine hub could be 80-100 meters above the ground. Also, some measuring stations were 30+ miles away from the turbines, and NOAA forecasts were only updated every 6 hours.

So it funded research at two leading labs in Colorado, and got a bespoke, state-of-the-art wind forecasting system that was specific to each farm, provided hub-height speeds, updated every 15 minutes, and extended out to 168 hours. Its backbone is comprised of large numeric weather prediction models, like the one that drives the North American Mesoscale Forecast System, and develops consensus forecasts that are highly reliable.

Then it had to be able to do something about it.

Drake’s special projects group didn’t exist 8 years ago; instead, staff from resource planning, machinery maintenance, dispatching and other departments from across its operations took on the task of innovating a technical solution to improve system reliability and reduce costs.

“It’s not linear, but instead the balance of two concepts, uncertainty and variability,” he explained.

“You have uncertainty in the accuracy of your forecast, and the variability of your wind supply.”

“Our goal was to enable automatic management of load and generation [called automatic dispatch], so our system could be smart about turning units on, or keeping them offline when using them would be inefficient or unnecessary.” It’s enabled by an Its IT infrastructure that sends thousands of data points every five minutes in order to connect the system to real-time circumstances.

The resulting cost savings were (and are) passed on to its customers, to date yielding approximately $60 million in savings for a company investment of $3.8 million. Its wind production displaces approximately 11.7 million tons of CO2 emissions annually, too.

Last year, the reliability of its forecasting allowed it to provide more than half of all its Colorado customers’ energy for an entire day.

It also recently added to its wind capacity, purchasing two wind farms and buying from two new partners last year, and announced plans to add another supplier and build a huge wind farm in North Dakota later this year.

It has qualified as the nation’s “top utility wind energy provider” for the past 12 years.

LEAVE A REPLY