Energy lab software models wind farm performance
Connecting state and local government leaders
Open source software developed by researchers at the National Renewable Energy Laboratory can calculate how variables alter the air flow and energy production of not just one wind turbine, but a whole wind farm.
According to experts, wind energy is notoriously difficult to predict because it is affected not just by the turbines but also by topography, temperature and surrounding vegetation.
Now software developed by researchers at the National Renewable Energy Laboratory can calculate how rolling hills, whipping blades, surface temperatures and other variables alter the air flow and energy production of not just one wind turbine, but a whole wind farm.
The Simulator for Wind Farm Applications (SOWFA) allows users to investigate wind turbine and wind plant performance under the full range of atmospheric conditions and in terrain.
Supercomputers made SOWFA possible, said NREL senior engineer Pat Moriarty, noting that he and his colleagues started work on the project in 2005, but that its success accelerated as computers reached the ability to do trillions of calculations per second. To accurately predict the motions of wind at a wind farm, computers need to simulate the dynamics in the space occupied by up to a hundred turbines – and model the flows, counterflows, cross flow and temperatures in every square meter. Other models have looked at pieces of the whole, but now that the wind industry is maturing and the cost margins are decreasing, it's increasingly important to model the whole system and maximize energy production.
Moriarty’s team uses NREL's Peregrine high performance computer to simulate the movement of air through rows of turbines.
SOWFA is available on GitHub, an open-source "community" software platform users can add their own modules and have the support of an online forum where they can discuss problems and solutions with other SOWFA users.
With SOWFA, the central super-controller can receive information from individual turbines and send command messages to them. SOWFA's central controller can send a message asking a turbine to adjust its yaw—but the turbine can override that command if conditions at the turbine are such that adjusting the yaw would be unsafe.
The architecture is designed to mimic as closely as possible how wind plant controls might be implemented at a real wind farm.
The Energy Department, which provided funding for the SOWFA project, is in the process of developing experiments to test that the modeling predicted by SOWFA matches real-life conditions and energy output at the wind farms.
SOWFA has drawn acute interest from manufacturers, developers and utilities. "One of the most important outputs of SOWFA is the reduced uncertainty," said Moriarty. "The finance community cares not just about energy production, but about the uncertainty of energy production over the next 20 years that wind farm would be operating. They particularly want to know what the worst-case energy production can be."
SOWFA's improved look at physics will help the community better understand the uncertainty gap between the average energy production and the worst-case energy production. That, according to NREL, should lower the interest rates for financing a wind farm, which can be a huge part of the total cost.
NEXT STORY: Workflow key to case management in the cloud