Forecasting the Future

Because wind provides power only intermittently, grid operators working to maximize efficiency need to know how much energy will be available at any given time. Under Spanish regulations, wind-farm operators sell their power to the grid and must predict how much wind they will be contributing; the operators pay penalties for inaccurate prediction. (In other national markets, operators are not penalized for these errors.)

The apparent burden this requirement places upon companies has turned into an opportunity. Spanish companies have taken the lead in microsite prediction-forecasting what will happen at a specific turbine, given the meteorological conditions. In fact, 90 percent of Spanish wind farms use prediction services from one Madrid-based company, Meteológica. The small firm has the largest market share of wind forecasting in the world.

“There was a highly competitive environment, because companies needed to be able to forecast as accurately as possible,” says Manuel Blanco of Meteológica. “In Spain this has made us very successful; we developed a simple system that is able to very accurately forecast the generation of wind farms.”

Meteológica began operations in 1997, developing an automated forecasting system that it began marketing in 2000. Soon the company was providing services for Spanish national and local government offices that needed to predict, for example, agricultural conditions or the chance of forest fires.

“After a few years we started to develop systems to forecast variables of interest to our clients,” says Blanco. The company developed models not just to predict the likelihood of forest fires but to model how those fires would evolve; not only to predict rainfall but to forecast water flows in specific rivers and, thus, hydrological power generation in the country.

Wind power was a natural business evolution. “In 2002 we developed a system that was able to forecast the output of the plant in kilowatt-hours, not just predict the wind speed,” says Blanco. The company began working with four wind farms that year and now manages forecasting for more than 600, predicting outcomes for about 15,000 megawatts of power.

The model works by taking advantage of different global atmospheric models and the meteorological information available from various satellite systems. Engineers input this information along with the conditions on the ground. They have developed statistical models that learn from the wind farm's actual performance and use that information, coupled with the input, to forecast conditions an hour to two days ahead.

“If a wind farm has been working for the last year, we know every hour how many megawatt-hours it produced,” Blanco says. “So we take this on-the-ground experience and use it with the information from the global atmospheric model. And we've developed the statistical relationship, the equation, that translates these atmospheric conditions into generation in megawatts.”

He admits that the model is weakest when a turbine first begins operation and works best after a wind farm has been running for a while. For the first week or two, the company makes its forecasts partly on the basis of simulations from other wind farms, but that's not necessary for long.

“It has an exponential learning curve,” says Blanco. “In a few months the system has seen almost every atmospheric condition possible on that farm. And a month is nothing in the life of the wind farm”-a typical facility is usually expected to produce power for about 20 years. He says the company has probably improved the accuracy of its wind-power forecasting by about 40 percent since 2003.

Meteológica works with wind farms across Europe and in North America and Asia as well, though international expansion has been cautious and deliberate. Instead of opening marketing offices in a variety of countries, the company attracts new clients through personal meetings and the strength of its track record. All the computations are done from the 17-person office in Madrid.

Another company important in forecasting is Kintech, which provides technology and equipment for collecting meteorological information. Its sensors and data loggers have captured most of the Spanish market, and its devices measure conditions in more than 50 countries. According to general manager Tirso Vasquez, the company has succeeded by customizing remote satellite communications from the data loggers to the customers' needs.

Systems for understanding wind potential in a given region still have the capacity to get better. To that end, AWS Truewind, an American powerhouse in wind-farm siting and forecasting, joined forces with the Spanish company Meteosim, a spinoff of a meteorological research team at the University of Barcelona. Having honed its technique in small, narrowly focused areas, Meteosim began working with new models for mapping wind resources around the world, providing that information to governments, nonprofit institutions, and potential wind-farm developers.

“Typically the main approach has been to install a tower, wait for two years, and see how much wind the wind developers or someone has been measuring on that tower,” says Meteosim director Joan Aymamí. Instead, Meteosim provides mapping information on specific, narrowly defined sites. “With this approach, a client has a very accurate idea of where to go, where are the best places in a big region.”

Extract from an article published in the magazine Technology Review
http://www.technologyreview.com/microsites/spain/wind

Forecasting the Future

Because wind provides power only intermittently, grid operators working to maximize efficiency need to know how much energy will be available at any given time. Under Spanish regulations, wind-farm operators sell their power to the grid and must predict how much wind they will be contributing; the operators pay penalties for inaccurate prediction. (In other national markets, operators are not penalized for these errors.)

The apparent burden this requirement places upon companies has turned into an opportunity. Spanish companies have taken the lead in microsite prediction-forecasting what will happen at a specific turbine, given the meteorological conditions. In fact, 90 percent of Spanish wind farms use prediction services from one Madrid-based company, Meteológica. The small firm has the largest market share of wind forecasting in the world.

“There was a highly competitive environment, because companies needed to be able to forecast as accurately as possible,” says Manuel Blanco of Meteológica. “In Spain this has made us very successful; we developed a simple system that is able to very accurately forecast the generation of wind farms.”

Meteológica began operations in 1997, developing an automated forecasting system that it began marketing in 2000. Soon the company was providing services for Spanish national and local government offices that needed to predict, for example, agricultural conditions or the chance of forest fires.

“After a few years we started to develop systems to forecast variables of interest to our clients,” says Blanco. The company developed models not just to predict the likelihood of forest fires but to model how those fires would evolve; not only to predict rainfall but to forecast water flows in specific rivers and, thus, hydrological power generation in the country.

Wind power was a natural business evolution. “In 2002 we developed a system that was able to forecast the output of the plant in kilowatt-hours, not just predict the wind speed,” says Blanco. The company began working with four wind farms that year and now manages forecasting for more than 600, predicting outcomes for about 15,000 megawatts of power.

The model works by taking advantage of different global atmospheric models and the meteorological information available from various satellite systems. Engineers input this information along with the conditions on the ground. They have developed statistical models that learn from the wind farm's actual performance and use that information, coupled with the input, to forecast conditions an hour to two days ahead.

“If a wind farm has been working for the last year, we know every hour how many megawatt-hours it produced,” Blanco says. “So we take this on-the-ground experience and use it with the information from the global atmospheric model. And we've developed the statistical relationship, the equation, that translates these atmospheric conditions into generation in megawatts.”

He admits that the model is weakest when a turbine first begins operation and works best after a wind farm has been running for a while. For the first week or two, the company makes its forecasts partly on the basis of simulations from other wind farms, but that's not necessary for long.

“It has an exponential learning curve,” says Blanco. “In a few months the system has seen almost every atmospheric condition possible on that farm. And a month is nothing in the life of the wind farm”-a typical facility is usually expected to produce power for about 20 years. He says the company has probably improved the accuracy of its wind-power forecasting by about 40 percent since 2003.

Meteológica works with wind farms across Europe and in North America and Asia as well, though international expansion has been cautious and deliberate. Instead of opening marketing offices in a variety of countries, the company attracts new clients through personal meetings and the strength of its track record. All the computations are done from the 17-person office in Madrid.

Another company important in forecasting is Kintech, which provides technology and equipment for collecting meteorological information. Its sensors and data loggers have captured most of the Spanish market, and its devices measure conditions in more than 50 countries. According to general manager Tirso Vasquez, the company has succeeded by customizing remote satellite communications from the data loggers to the customers' needs.

Systems for understanding wind potential in a given region still have the capacity to get better. To that end, AWS Truewind, an American powerhouse in wind-farm siting and forecasting, joined forces with the Spanish company Meteosim, a spinoff of a meteorological research team at the University of Barcelona. Having honed its technique in small, narrowly focused areas, Meteosim began working with new models for mapping wind resources around the world, providing that information to governments, nonprofit institutions, and potential wind-farm developers.

“Typically the main approach has been to install a tower, wait for two years, and see how much wind the wind developers or someone has been measuring on that tower,” says Meteosim director Joan Aymamí. Instead, Meteosim provides mapping information on specific, narrowly defined sites. “With this approach, a client has a very accurate idea of where to go, where are the best places in a big region.”

Extract from an article published in the magazine Technology Review
http://www.technologyreview.com/microsites/spain/wind