IBM takes weather forecasting to the next level
Utilities and grid managers need to know precisely how much energy wind farms and solar-energy plants will produce at any given moment in order to plug that power into the grid. Inaccuracies in day-ahead wind and solar forecasting average around 20 percent, frequently forcing grid operators to under-utilize renewable energy resources, and allowing hundreds of megawatts of clean energy go to waste.
IBM is aiming to reduce that waste by using Big Data and machine-learning tools to splice together multiple weather models, along with terabytes of data from more than 1,600 weather-monitoring stations, solar installations and wind farms, into a single framework. By combining realtime data with historical records, geographical records and other sources, IBM’s system dramatically improves forecasting accuracy. “The more data you put in, the smarter it gets,” explains Hendrik Hamann, a research manager at IBM.
Improving forecasting accuracy is a vital step as utilities increase their deployment of renewable energy sources, Hamann notes. Most experts believe that the unpredictable, intermittent nature of sunshine means that solar energy can’t account for more than 20 to 30 percent of the U.S. energy supply — but better forecasting could reduce that uncertainty, and increase the theoretical solar-energy ceiling to around 50 percent of U.S. energy needs, Hamann writes.
By reducing waste, more accurate forecasts would also dramatically reduce the renewable energy sector’s operating costs, according to a National Renewable Energy Laboratory study. In the wind sector alone, a 20 percent improvement in forecasting accuracy could reduce the sector’s costs by up to $975 million a year, depending on market penetration, researchers found. "By improving the accuracy of forecasting, utilities can operate more efficiently and profitably," says Bri-Mathias Hodge, who oversees the NREL’s Transmission and Grid Integration Group. "That can increase the use of renewable energy sources as a more accepted energy generation option."
It’s unclear how utilities will incorporate IBM’s free-to-use forecasting system into their current forecasting and supply-management protocols. Commercial forecasting fees have fallen sharply in recent years, with most energy firms now paying just $300 to $400 per month per plant, according to an NREL survey — a relatively trivial sum considering the urgent need for accurate forecasts. Several utilities are also researching their own in-house forecasting systems, with BPA developing algorithms capable of choosing between multiple forecasts on an hour-by-hour basis, and Xcel Energy (NYSE: XEL) developing a probabilistic forecasting model designed to draw multiple forecasts into a single data stream.
Equities analysts remain cautiously optimistic about IBM’s prospects, with the firm winning buy or hold recommendations from all but one of 14 surveyed analyst firms.
Companies to watch
* Colorado’s Xcel Energy (NYSE: XEL) is working with Boulder’s National Center for Atmospheric Research to create more accurate forecasts using data from its wind turbines, giving grid operators the confidence to source up to 60 percent of the state’s energy from wind. “That kind of wind penetration would have given dispatchers a heart attack a few years ago,” says Xcel’s Drake Bartlett.
* General Electric (NYSE: GE) is testing “smart” turbines capable of monitoring their own performance, communicating with neighboring turbines, and using in-turbine forecasting and energy storage technologies to provide a smoother, more reliable supply of electricity.
* Clean Power Research is working with California ISO to track the state’s 200,000 or so small-scale solar installations, and to predict how much energy each will feed into the grid based on cloud-tracking data from a geostationary satellite.
* Finnish firm Vaisala (HEL: VAI) acquired U.S. renewable-energy forecasting specialist 3TIER in late 2013 to improve its forecasting offerings for the wind, solar and hydro sectors. The company was this year selected by the Department of Energy to run a $2.5 million project aimed at improving wind forecasting in complex terrains
Ben Whitford is the U.S. correspondent for The Ecologist. He has written for the Guardian, Newsweek, Mother Jones, Slate, and many other publications.