Renewable Energy


NCAR is uniquely qualified to help support our nation's transition to renewable energy due to NCAR's breadth and depth of atmospheric science knowledge, experience with technology transfer, and access to university researchers. These capabilities led NCAR to include a new research frontier in the 2009 NCAR Strategic Plan.  RAL is collaborating with university researchers, DOE labs, and other NCAR entities to develop methods to more accurately analyze and predict wind energy to support the renewable energy industry.  RAL has projects both in analyzing wind potential for siting wind farms (Mesoscale Current Climate Downscaling Project) and in real time forecasting to improve operations and economics of incorporating wind energy into the power mix, which is described below.  It is anticipated that RAL’s renewable energy research will expand to include the prediction of direct and indirect solar radiation and the impact of aerosols and jet contrails on insolation and how that impacts solar energy harvesting.

Xcel Energy Wind Prediction Project

In 2011, RAL completed a 2.5 year collaborative project with Xcel Energy Services, Inc. to perform research and develop technologies to improve Xcel Energy’s ability to increase the amount of wind energy in their energy generation portfolio. The agreement and scope of work was designed to provide highly detailed, localized wind energy forecasts to enable Xcel Energy to more efficiently integrate electricity generated from wind into the power grid. The wind prediction technologies help operators make critical decisions about powering down traditional coal- and natural gas–powered plants when sufficient winds are predicted, enabling increased reliance on alternative energy while still meeting the needs of its customers. The U.S. Department of Energy’s National Renewable Energy Laboratory (NREL) is also collaborating by developing algorithms to calculate the amount of energy that the turbines generate by winds blowing at various speeds for a broad spectrum of wind facilities. The wind prediction technologies have been designed to cover Xcel Energy wind farms in Colorado, Minnesota, New Mexico, Texas, Wisconsin,  and Wyoming. It is anticipated that wind energy forecasting companies in the United States and overseas may adopt the developed technologies to help utilities that need more accurate wind predictions transition away from fossil fuels. NCAR has been communicating with scientists and energy companies throughout the world about such possibilities.

To generate wind energy forecasts, NCAR is incorporating observations of current atmospheric conditions from a variety of sources, including satellites, aircraft, weather radars, ground-based weather stations, and even sensors on the wind turbines. The information is utilized by various powerful NCAR-based tools:

  • The Weather Research and Forecasting (WRF) computer model, which generates high resolution deterministic finely detailed simulations of future atmospheric conditions
  • The Real-Time Four-Dimensional Data Assimilation System (RTFDDA), which continuously updates the WRF model simulations with the most recent observations
  • The Dynamic Integrated Forecast System (DICast®), which statistically optimizes the forecasts based on current observations, climatological data, and real-time validation of the model predictions
  • An Ensemble of numerical weather prediction forecasts based on both WRF and the MM5 (Penn State/NCAR) mesoscale models
  • NCAR calibration tools that improve the probabilistic forecast, specifically the Analogue Kalman Filter combined with Quantile Regression to remove bias, sharpen the forecast, and increase statistical reliability
  • NCAR’s Four Dimensional Variational Doppler Radar Analysis System (VDRAS), which combines Doppler radar data with a cloud scale model to predict short term weather events 
  • An NCAR-developed expert system that uses field observations to fine-tune when wind ramps are most likely to impact the wind plant
  • Customized Graphical User Interfaces to provide grid operators with a best forecast that includes error estimates as well as meteorological GUIs to visualize the weather graphics
Figure 1: Diagram of the WRF domains used in the wind energy prediction system. The grid spacings are as follows: D1=30km, D2=10km, and D3=3.3km.
Figure X: Diagram of the WRF domains used in the wind energy prediction system. The grid spacings are as follows: D1=30km, D2=10km, and D3=3.3km.

These models and analysis systems were originally developed for other applications, but have been combined to provide a comprehensive wind energy forecasting system.  Real-time information from Xcel Energy’s largest wind facilities is utilized by the wind energy system to refine the power curve calculations and tune the forecasts. Wind predictions are made for each wind turbine and a sophisticated post-processing algorithm converts the hub-height wind predictions into energy predictions. These turbine-based wind and energy predictions are then merged to create predictions of average wind, power generation, and expected error for each wind facility, connection node and public service region in Xcel Energy’s domain. The predictions are subsequently provided to Xcel Energy using a  real-time communication link.

In the first six months of the agreement, NCAR successfully developed the initial capabilities and began providing wind energy predictions. Since then RAL has been expanding the system capabilities and incorporating forecasts for additional wind farms as they are integrated in the Xcel system. During 2011, the system was transferred to a third party for continued operations.

The Real-Time Four Dimensional Data Assimilation (RTFDDA) and forecasting system, developed by RAL to satisfy the meteorological needs of Army test ranges, has been adapted for wind-energy prediction. RAL implemented an operational RTFDDA system over the western and central states for supporting wind-power forecasting. This system contains three modeling domains with grid sizes of 30, 10 and 3.3 km (Figure 1). The 3.3 km domain covers the Rocky Mountains from New Mexico to Montana, the High Plains states, and most areas of the Central Plains. The system runs with a 3-hour cycle. In each cycle, it produces 24-hour forecasts for the innermost domain and 72-hour forecasts for the two coarser domains.  The inner domain (3.3 km) generates output at 15-minute time steps.

Figure 2: Diagram of the VDRAS domains used in the wind energy prediction system.
Figure 2: Diagram of the VDRAS domains used in the wind energy prediction system.

An ensemble wind energy prediction system has been added to the system in 2010. This system combines 15 WRF members with 15 MM5 members run on the 10 km grid. The mean of each model system is used in the DICast® system and also transmitted to the Xcel meteorologists for person-in-the-loop forecasting. This ensemble system also provides estimates of the uncertainty in the wind that affect the resulting power forecasts. Calibration methods are being used to calibrate that uncertainty to reflect the actual expected error. 

The Variational Doppler Radar Analysis System (VDRAS) enhances the wind energy forecasts through its independent monitoring of the evolving weather in the immediate vicinity of the targeted wind farms.  This helps optimize the overall system by refining the short term forecasts by predicting the precise times that weather systems will impact the wind farms.  The VDRAS domain for the Xcel project is depicted in Figure 2. 

Figure 3: Conceptual diagram of the wind energy prediction technology components that will be incorporated into the final configuration.
Figure 3: Conceptual diagram of the wind energy prediction technology components that will be incorporated into the final configuration.

A final conceptual diagram of all the technology components is provided in Figure 3.  The forecast power output of the system is conveyed to a graphical user interface (GUI) display that presents wind power predictions out to 72 hours from present. The meteorological model output data is made available to Xcel meteorologists using images and HTML-based web pages. 

The system has been enhanced through studying specific cases with significant weather influence on power production.  Here studies were made assessing cases where icing of the turbine blades affected the power output.  NCAR scientists identified specific weather situations where such icing is expected to occur and assessed methods to predict them.   Various wind ramp events were also studied. In such wind ramp events power may substantially increase or decrease over short time periods, making integration of the energy into the grid particularly difficult.  Assessing such events enables development of more robust methods to fine-tune forecasts of these situations.

Plans for FY2012:

FY2012 will be an exciting time for renewable energy research at NCAR. Various projects are beginning that will leverage the expertise of NCAR scientists and engineers. New collaborations with national laboratories, university scientists, and private companies are beginning that will advance the state-of-the-science necessary to make a large penetration of wind capacity feasible.

NCAR scientists were recently awarded several DOE Offshore Wind projects that will begin in 2012. The first of these, headed by Dr. Edward Patton (MMM), seeks to characterize the coupling between wind, wave states and stratification toward reducing uncertainty in siting offshore wind plants, allowing improved maintenance scheduling, and more accurate estimates of turbine loads and energy production. Dr. Luca Delle Monache (RAL) heads the second project, which focuses on enhanced understanding of the interactions between the ocean and the atmosphere in order to better predict winds for both locating wind plants and for forecasting power for plant operations. In addition, RAL scientist Dr. Sue Ellen Haupt is a co-PI on a collaborative project with Penn State to develop a “Cyber Wind Facility”, a computational facility akin to a field wind turbine test facility. These projects will lead to better model capability that will remove market barriers to developing offshore wind energy.

In addition, RAL will work to:

  • Expand wind forecasting capability into new areas, including mountainous and coastal sites
  • Enhance the forecasting of wind power ramping events using statistical guidance techniques as well as the Variational Doppler Radar Analysis System (VDRAS) and regime dependent expert systems
  • Expand calibration of the uncertainty in wind forecasting systems
  • Forecast cloud cover for solar energy sites
  • Enhance very fine scale modeling capabilities with Large Eddy Simulations (LES)
  • Study the effects of wakes on downwind turbine ingest
  • Forecast conditions conducive to turbine icing
  • Downscale future climate scenarios for resource assessment
  • Collaborate with university researchers on topics including:
    •    assimilating mesoscale model data into  fine-scale computational fluid dynamics models
    •    studying the impact of shear across the turbine blade on power production
    •    assessing icing conditions on rotating blades
    •    coupling LES models with models of rotating blades using overset grid technology
    •    assessing the feasibility of integrating wind turbines into buildings
    •    applying new artificial intelligence techniques to wind prediction problems