Image

The wind is a variable and uncertain power source that relies on a host of complex atmospheric forces. Reducing the uncertainty of wind power forecasts, upon which wind farm operators and power grid operators depend, is the goal of a team of researchers at the U.S. Department of Energy’s Lawrence Livermore National Laboratory (LLNL), who combine fieldwork, advanced simulations, and statistical analysis in their efforts.

Observations Guide and Verify Simulations

LLNL field researchers are characterizing winds in numerous locations with distinct conditions. They have made significant discoveries studying the winds in the lower atmosphere and their effects on wind farm power output. A particularly important variable is turbulence, as it affects the power extracted from wind turbines as well as the reliability and life spans of turbine components.

Using field instrumentation and data from wind farms, LLNL researchers determine the effects of wind and other atmospheric variables on power production. Accurate descriptions of how wind velocity and turbulence vary across the turbine rotor disk are important in improving wind power forecasts. LLNL activities include the first effort to explore the relationship between three-dimensional turbulence and turbine power production.

Advanced Simulations Address Key Challenges

LLNL simulation and modeling efforts use high-performance computation to study atmospheric flows and turbine aerodynamics in fine detail. The task is enormous because the length scales involved span eight orders of magnitude—from regional weather patterns that include fronts, sea breezes, and flows over mountain ranges (~1,000 km) down to aerodynamic effects around the wind turbine’s rotor (~1 mm).

To analyze the impacts of atmospheric flows and turbine aerodynamics on wind farm performance, LLNL adapted the National Center for Atmospheric Research’s popular Weather Researching and Forecasting (WRF) atmospheric simulation code to include turbine models and improved turbulence simulations to extend capabilities to wind farm analysis. However, the standard WRF model is restricted to simple terrain with shallow slopes. To eliminate this restriction, LLNL developed the immersed boundary method (IBM) that simulates flow in highly complex terrain with near vertical slopes without compromising accuracy.

Image

Minimizing Uncertainties

LLNL researchers are studying how to reduce uncertainties and errors in wind forecasts and power predictions using statistical modeling and uncertainty quantification. They have developed a statistical power curve that can adapt to changing conditions and greatly improve predicted power output. Traditional turbine power curves model power as a function of only the wind speed at the hub height of the turbine and can err by 50%. In reality, power output is a function of many additional variables.

To reduce uncertainties in wind forecasts, ensemble modeling is employed, which entails running a family of forecasts using slightly different assumptions. One of the chief advantages of ensemble modeling is the ability to spot outliers such as a wind ramp. Because power is proportional to the cube of the wind speed, it is important to be aware of outliers.

Accurate Predictions, Lower Costs for Everyone

Taken together, the field observations, simulations, and statistical modeling are significantly improving wind power predictions. LLNL is sharing this work with the wind industry and governing bodies to help them refine their power curves and incorporate findings about what atmospheric processes are important in wind power forecasting. With improved models, wind farm operators will know how to better maximize their sizable investments, more skillfully bid into the energy market, and plan new development.

To learn more, read Predicting Wind Power with Greater Accuracy.