IBM Launches Advanced Renewable Forecasting Tool

Better models allow for higher penetration of renewable energy


Weather has always been an important factor in planning for the needs of the electrical grid, but lately it has become even more crucial with the proliferation of grid-scale wind and solar resources.

To meet the needs of utilities that are installing large amounts of renewable energy, IBM just launched software that brings together data analytics and cutting-edge weather prediction. The Hybrid Renewable Energy Forecasting, or HyRef, incorporates cloud imaging technology, sky-facing cameras, and operational and environmental sensors to build customized models of renewable outputs.

“If you can’t do the weather forecast better, you’re done,” says Lloyd A. Treinish, chief scientist of IBM's Deep Thunder program, which aims to improve local weather forecasting through the use of high-performance computing.

Wind farms, for example, have had sensors on the turbines, but they’re used to monitor the turbines and not the weather because the data has been too contaminated to be of use, Treinish says.

IBM took its expertise in renewables and weather forecasting, including its micro-forecasts used in Deep Thunder, to develop HyRef. One of the biggest challenges was taking inputs from the front of the wind blades, rather than behind it. Taking measurements from the front of the blade increased the acoustic noise, which had to be filtered out to get an accurate reading.

Once IBM had data that was free of contamination, it built a statistical model that could drill down to the individual turbine scale and provide forecasts for 15-minute intervals or up to a month in advance. “With accuracy and precision, you have that much better information so then you can have inputs that offer far greater fidelity for power output,” says Treinish.

The first client to use the software, Jibei Electricity Power Company Limited—a subsidiary company of the State Grid Corporation of China—hopes to increase the integration of renewable power generation by 10 percent. The utility is part of State’s Grid’s the Zhangbei 670 MW demonstration project, the world’s largest utility-scale renewable power plant, which combines wind and solar with energy storage and transmission.

“Clients keep telling us forecasts aren’t good enough,” Treinish says of current weather forecasts for renewables. IBM’s approach is to customize the software for each utility’s needs, whether that’s a large-scale solar array or offshore wind farm. In the case of Jibei, the utility is interested in day-ahead forecasting for its wind resources.

“We have some of the most high-resolution weather modeling software, and we bring that to bear,” says Treinish. IBM also relishes the multi-disciplinary challenge of building better forecasting tools, which combine expertise from math disciplines, computing, atmospheric physics and other sciences.

Much of the research for HyRef was borrowed from solving other weather-related challenges, such as flood forecasting for cities or outage detection for grid operators. “We’re data scavengers,” says Treinish. “We’ll use whatever we can use to make the forecast better.”

The interest in HyRef, so far, has been in areas around the globe that have a high penetration of renewable energy and are already having intermittency problems, according to Stephen Callahan, a partner in IBM’s global business services for the energy and utilities industry. For some clients, the goal is not just better forecasting for the individual utility, but bringing that accuracy to market at the wholesale level so that the economic value of renewable energy can be priced more effectively. “That is the threshold we’re all working to get to,” says Callahan.

Photo: Yagi Studio/Getty Images

The Energywise Newsletter

Monthly newsletter on latest news & opinion for the power & energy industry, green technology, and conservation.

About the Energywise blog

IEEE Spectrum’s energy, power, and green tech blog, featuring news and analysis about the future of energy, climate, and the smart grid.