A11H-01 INVITED
Research Needs for Wind Resource Characterization
Currently, wind energy provides about 1 percent of U.S. electricity generation. A recent analysis by DOE,
NREL, and AWEA showed the feasibility of expanding U.S. wind energy capacity to 20 percent, comprising
approximately 300 gigawatts. Though not a prediction of the future, this represents a plausible scenario for
U.S. wind energy.
To exploit these opportunities, a workshop on Research Needs for Wind Resource Characterization was held
during January 2008. This event was organized on behalf of two DOE organizations; the Office of Biological
and Environmental Research and the Office of Energy Efficiency and Renewable Energy. Over 120
atmospheric science and wind energy researchers attended the workshop from industry, academia, and
federal laboratories in North America and Europe. Attendees identified problems that could impede achieving
the 20 percent wind scenario and formulated research recommendations to attack these problems. Findings
were structured into four focus areas: 1) Turbine Dynamics, 2) Micrositing and Array Effects, 3) Mesoscale
Processes, and 4) Climate Effects.
In the Turbine Dynamics area, detailed characterizations of inflows and turbine flow fields were deemed
crucial to attaining accuracy levels in aerodynamics loads required for future designs. To address the
complexities inherent in this area, an incremental approach involving hierarchical computational modeling
and detailed measurements was recommended. Also recommended was work to model extreme and
anomalous atmospheric inflow events and aerostructural responses of turbines to these events.
The Micrositing and Array Effects area considered improved wake models important for large, multiple row
wind plants. Planetary boundary layer research was deemed necessary to accurately determine inflow
characteristics in the presence of atmospheric stability effects and complex surface characteristics. Finally, a
need was identified to acquire and exploit large wind inflow data sets, covering heights to 200 meters and
encompassing spatial and temporal resolution ranges unique to wind energy.
The Mesoscale Processes area deemed improved understanding of mesoscale and local flows crucial to
providing enhanced model outputs for wind energy production forecasts and wind plant siting. Modeling
approaches need to be developed to resolve spatial scales in the 100 to 1000 meter range, a notable gap in
current capabilities. Validation of these models will require new instruments and observational strategies,
including augmented analyses of existing measurements.
In the Climate Effects area, research was recommended to understand historical trends in wind resource
variability. This was considered a prerequisite for improved predictions of future wind climate and resources,
which would enable reliable wind resource estimation for future planning. Participants also considered it
important to characterize interactions between wind plants and climates through modeling and observations
that suitably emphasize atmospheric boundary layer dynamics.
High-penetration wind energy deployment represents a crucial and attainable U.S. strategic objective.
Achieving the 20 percent wind scenario will require an unprecedented ability for characterizing large wind
turbines arrayed in gigawatt wind plants and extracting elevated energy levels from the atmosphere. DOE
national laboratories, with industry and academia, represents a formidable capability for attaining these
objectives.
http://www.nrel.gov/docs/fy08osti/43521.pdf
A11H-02
Remote Sensing of the 3D Wind and Turbulence Field by Coherent Doppler Lidars for Wind Power Applications
For several decades Risø DTU has been involved in wind power meteorology and during the last half- decade the performance of commercially available coherent wind Doppler Lidars have been extensively studied at the test station for large wind turbines, Høvsøre, in Western Jutland, Denmark. One aspect of wind Lidars, in contrast to many in-situ wind-monitoring instruments, is that they are not truly point-monitoring devices but the wind speed measured is rather a weighted average of the line-of-sight velocity component over an extended spatial volume. The width of the weighting function along the beam is for pulsed systems mainly determined by the laser pulse length together with the sampling duration for a single Doppler spectrum, whereas for continuous-wave systems the focal depth of the laser beam determines the weighting width. Here, some recent results regarding the effect of this spatial volume averaging on turbulence measurements are presented. One common approach to obtain the whole wind vector is to perform a conical scan of the Lidar laser beam, which under the horizontal homogeneity assumption allows for the wind vector to be extracted. The wind vector measured is thus, in some sense, averaged over a substantial lateral area and time. However, temporal as well as spatial resolution of the wind field could be improved if instead three fully steerable Lidars were simultaneously measuring from three different locations around the air volume of interest. Based on this concept, a ground-based Doppler Lidar Windscanner facility capable of providing the wind vector in several hundred locations each second is currently under development within a Risø DTU project that aims at providing a useful research tool in the field of wind power meteorology for the decade ahead. A field campaign inter-comparison of the turbulence and the wind vector measured by a sonic anemometer and by three Lidars staring from three different directions towards the location of the sonic anemometer has already recently provided some initial prospective results of this approach to measure the 3D wind and turbulence field.
A11H-03 INVITED
Characterizing the Great Plains Low-Level Jet Wind Resource using Doppler Lidar
A major North American wind resource during the warm season is the nocturnal low-level jet (LLJ) of the Great Plains. A dynamic acceleration of flow above the surface in response to cooling of the earth surface in the evening, this LLJ occupies an expansive geographical region thousands of kilometers across, and thus represents a vast resource. LLJ properties there have been observed to be relatively homogeneous in the horizontal, but the jet speed varies from night to night. Thus, accurate LLJ forecasts are essential to anticipate the amount of power that will be generated and transferred to the power grid. Current numerical weather prediction (NWP) models do not predict LLJ speeds and heights to sufficient accuracy for this application. The height of the LLJ maximum in this region generally occurs at 100-200 m above ground, traditionally a difficult atmospheric layer to obtain measurements - but most important for wind energy as the layer occupied by the turbine blades. Here we use the high-resolution Doppler lidar, an ideal instrument for this difficult layer, to obtain properties of wind and turbulence at high spatial and temporal resolution, to document the evolution of the LLJ through entire nights. These observations show that the strongest evening accelerations occur 100 m or more above ground and that near-surface measurements do not reflect the intensity of this increase in wind speeds aloft. The shear in the layer below the jet maximum tends to be nearly constant with height and strong (~0.1 s-1), so that for this type of flow, power-law-exponent extrapolations significantly underestimate both shear and extrapolated wind speeds aloft. Routine turbulence in the layer below the LLJ varies directly with LLJ speed but occasional bursts or events of strong turbulence or periodic-wave activity can adversely affect or even disable turbine hardware. Critical areas where better understanding is urgently needed for wind energy applications include understanding and predictability of LLJ dynamics, improvement of the representation of stable boundary layer and stable mixing processes in NWP models, and understanding of turbulence characteristics under stable flow conditions.
A11H-04 INVITED
The Stable Atmospheric Boundary Layer: A Challenge for Wind Turbine Operations
The growth in the installation of very large wind farms has been increasing exponentially. It is not uncommon for such installations to have an aggregate nameplate capacity of 500 MW or more. Currently there are individual wind plants being planned with capacities exceeding 1 GW. While the latest wind turbine designs now provide individual capacities approaching and sometimes exceeding 3 MW, large numbers of such machines will need to be installed and operated in juxtaposition to one another. The challenge for the turbine manufacturers and wind plant designers is to provide an intersection of optimum designs that provides for reliable and efficient wind plant operation while at the same time minimizing the costs of maintenance and repair. Current experience in the operation of large wind plants has shown that a combination of under production and higher maintenance and operation costs are commonplace. The National Wind Technology Center has been involved with developing an understanding of the interaction of atmospheric boundary layer turbulence with operating wind turbines and its operational consequences for almost twenty years. Field measurement campaigns have been employed to acquire very detailed measurements of the turbulent inflow synchronized with the corresponding dynamic responses of operating wind turbines and many of their individual components. The results of this work have identified turbulent conditions associated with the nocturnal or stable atmospheric boundary layer as a being a major source of the structural loads responsible for fatigue accumulation in many wind turbine components. The repetitive nature of various nocturnal turbulence generating processes provides the environment to deliver relatively low levels of loading over a diurnal period. However over longer periods these loads contribute to increased wear and result in shortened component lifetimes. We will provide a brief overview of the atmospheric processes in the stable boundary layer that contribute to this accelerated wear and occasionally are of sufficient severity to cause turbines into fault conditions. We will also discuss what avenues are being investigated to minimize the impact of such conditions in order to improve the operational reliability and lifetime of modern wind turbines.
A11H-05
Implications of the GABLS LES Intercomparison Studies for Future Wind Energy Projects
Given the urgent need for higher accuracy and better reliability in wind resource assessment, turbine micro-
siting, wake modeling, and short-term forecasting, the wind industries around the world are more and more
opting for various computational fluid dynamics (CFD) codes instead of traditional mass-consistent or linear
models. All of these contemporary atmospheric CFD models (e.g., Meteodyn, WindSim) essentially solve the
ensemble averaged Navier-Stokes equations and utilize the Reynolds-Averaged Navier-Stokes (RANS)
turbulence modeling approach. An alternative and undoubtedly better approach would be to use the Large-
Eddy Simulation (LES) models. At present, due to high computational costs, these state-of-the-art
atmospheric boundary layer models have not gained popularity outside the academic and research
institutions. With the advent of petascale computing environments, it is plausible to envisage that the LES
models will eventually find their place in a variety of applied wind energy projects.
In the year 2003, the first LES intercomparison of atmospheric stable boundary layers was organized under
the auspices of the GEWEX atmospheric boundary layer study (GABLS). More than ten LES modeling groups
from around the world participated in this intercomparison study. They modeled a weakly stable barotropic
boundary layer utilizing several LES subgrid-scale (SGS) models. This intercomparison study highlighted that
LES of weakly stable boundary layers is quite feasible. However, some of the key turbulence statistics were
found to be sensitive to SGS models even at relatively fine resolutions, which is not desirable.
Quite a few novel SGS models have been proposed in the atmospheric boundary layer literature in the past
few years.
The existing SGS models have also evolved quite dramatically during this period. The GABLS community
realized that the time was opportune for another LES intercomparison in order to re-evaluate the capabilities
of the SGS models - in the context of stably stratified flows. In this presentation, we will describe the setup of
the ongoing GABLS3 LES intercomparison case along with preliminary results. This time, the focus is on a
moderately stratified, baroclinic, mid-latitude nighttime boundary layer. The boundary layer was observed
over Cabauw, Netherlands on July 1st, 2006. The initial condition for the LES intercomparison is created by
synthetically merging the observed 200-m Cabauw tower data and a high-resolution 00 UTC sounding from
DeBilt. Time-height-dependent geostrophic wind forcings are derived from a network of surface pressure
stations in Netherlands combined with analysis of a 3D weather-forecast model.
During this presentation, we will discuss the performances of several dynamic and static (stability corrected)
SGS models in capturing the characteristics of this moderately stable boundary layer. We will put a strong
emphasis on the evolution of low-level jets (LLJs) and morning transition. LLJs provide a vast resource of
wind energy, thus, it is of utmost importance to improve the LES modeling capabilities of these atmospheric
features. The ongoing GABLS LES intercomparison study will assess our current LLJ modeling capability and
will also highlight future modeling needs.
http://www.atmo.ttu.edu/basu/GABLS3
A11H-06
Nesting Large-Eddy Simulations Within Mesoscale Simulations for Wind Energy Applications
With increasing demand for more accurate atmospheric simulations for wind turbine micrositing, for operational wind power forecasting, and for more reliable turbine design, simulations of atmospheric flow with resolution of tens of meters or higher are required. These time-dependent large-eddy simulations (LES) account for complex terrain and resolve individual atmospheric eddies on length scales smaller than turbine blades. These small-domain high-resolution simulations are possible with a range of commercial and open- source software, including the Weather Research and Forecasting (WRF) model. In addition to "local" sources of turbulence within an LES domain, changing weather conditions outside the domain can also affect flow, suggesting that a mesoscale model provide boundary conditions to the large-eddy simulations. Nesting a large-eddy simulation within a mesoscale model requires nuanced representations of turbulence. Our group has improved the Weather and Research Forecating model's (WRF) LES capability by implementing the Nonlinear Backscatter and Anisotropy (NBA) subfilter stress model following Kosoviæ (1997) and an explicit filtering and reconstruction technique to compute the Resolvable Subfilter-Scale (RSFS) stresses (following Chow et al, 2005). We have also implemented an immersed boundary method (IBM) in WRF to accommodate complex terrain. These new models improve WRF's LES capabilities over complex terrain and in stable atmospheric conditions. We demonstrate approaches to nesting LES within a mesoscale simulation for farms of wind turbines in hilly regions. Results are sensitive to the nesting method, indicating that care must be taken to provide appropriate boundary conditions, and to allow adequate spin-up of turbulence in the LES domain. This work is performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
A11H-07
The Use of Full-Physics Atmospheric Modeling for Wind Power Plants
This presentation will describe a number of issues relevant to the use of mesoscale meteorological models for the development and operation of wind power plants. An accurate description of the local winds over a range of spatial and temporal scales is important for wind plants. In addition to various measurement methods, a number of modeling tools can be used to explore winds at these scales, including a full-physics mesoscale model such as the Weather Research and Forecasting (WRF) model. Simulations in regions of complex terrain can, however, have significant amounts of uncertainty, and results can be sensitive to the model parameters such as turbulence representation, the horizontal and vertical grid spacing, and initial and boundary conditions. Numerous studies conducted by PNNL scientists have quantified the performance of WRF. These evaluations included simulation of low-level winds in a number of geographic areas with both simple and complex terrain. However, previous research focused on comparisons with data from intensive, short-duration field campaigns that may not be completely relevant for wind plants. The identification of long- term, high quality data sets is therefore an important aspect of evaluating and improving model performance in wind energy applications. One such source of quality-assured meteorological data is from the US Department of Energy's Hanford Site. It is an ideal location for evaluating the performance of the WRF model for both prognosis of the local winds, as would be appropriate for a wind resource characterization, and for an analysis of severe wind events, which is important for wind turbine safety. The Hanford Site is located in southeastern Washington State and covers approximately 1500 sq km. The topography in this region is dominated by a number of significant ridges around a central basin, and severe wind events are frequent, especially during the springtime. Wind energy has been explored in this area, and a wind plant with a 96 mW capacity has recently been installed nearby. The Hanford Meteorological Monitoring Network, which consists of 30 stations, including a station near the top of Rattlesnake Mountain (approximately 1 km above the basin floor), a 120 m tower near the center of the basin, and three 60 m towers, was developed to provide real-time support to activities around the Hanford Site. The network has been operational, with its calibrated data archived, for more than 60 years, providing high-quality long-term observations well-suited for these wind energy modeling studies. This presentation will describe the evaluation of WRF using data from the Hanford Meteorological Monitoring Network. PNNL-SA-62301
A11H-08
Wind Energy Forecasting: How Useful are the Forecasts?
To meet the rapidly growing demands for renewable energy, the number of wind energy projects in the U.S. is growing at an unprecedented rate. For many sectors of this industry, wind energy forecasts play an increasingly important role in the viability and profitability of these projects. Highly accurate forecasts are demanded, but at times the metrics commonly used as indicators of forecast quality do not coincide with the utility of these forecasts in decision-making. One such example is the importance in being able to predict events in which wind power production fluctuates rapidly (ramp events). These events are of foremost importance to wind energy forecast users, however, many commonly used metrics are not necessarily indicative of skill in capturing these events. In this presentation, wind energy forecasts are assessed through examining user needs and expectations as compared to the state of the science. Although it is common for much emphasis to be placed on a limited number of verification metrics as indicators of forecast success or failure, other approaches provide more insight into forecast quality. These include the examination of high- impact events and distributions-based verification approaches, among others. This presentation demonstrates the application of a range of verification methods both as indicators of forecast quality and in the diagnosis of forecast errors.