A21E-0221
Model for wind resource analysis and for wind farm planning
Due to the ever increasing anthropogenic environmental pollution and the worldwide energy demand, the research and exploitation of environment-friendly renewable energy sources like wind, solar, geothermal, biomass become more and more important. During the last decade wind energy utilization has developed dynamically with big steps. Over just the past seven years, annual worldwide growth in installed wind capacity is near 30 %. Over 94 000 MW installed currently all over the world. Besides important economic incentives, the most extensive and most accurate scientific results are required in order to provide beneficial help for regional planning of wind farms to find appropriate sites for optimal exploitation of this renewable energy source. This research is on the spatial allocation of possible wind energy usage for wind farms. In order to carry this out a new model (CMPAM = Complex Multifactoral Polygenetic Adaptive Model) is being developed, which basically is a wind climate-oriented system, but other kind of factors are also considered. With this model those areas and terrains can be located where construction of large wind farms would be reasonable under the given conditions. This model consist of different sub- modules such as wind field modeling sub module (CMPAM/W) that is in high focus in this model development procedure. The wind field modeling core of CMPAM is mainly based on sGs (sequential Gaussian simulation) hence geostatistics, but atmospheric physics and GIS are used as well. For the application developed for the test area (Hungary) WAsP visualization results were used from 10 m height as input data. This data was geocorrected (GIS geometric correction) before it was used for further calculations. Using optimized variography and sequential Gaussian simulation, results were applied for the test area (Hungary) at different heights. Simulation results were produced and summarized for different heights. Furthermore an exponential regressive function describing the vertical wind profile was also established. The following altitudes were examined: 10 m, 30 m, 60 m, 80 m, 100 m, 120 m and 140 m. By the help of the complex analyses of CMPAM, where not just mere wind climatic and meteorological factors are considered, detailed results have been produced to 100 m height. Results at this altitude were analyzed and explained in a more detailed way because this altitude proved to be the first height that can ensure adequate wind speed for larger wind farms for wind energy exploitation in the test area. Keywords: wind site assessment, wind field modeling, complex modeling for planning of wind farm, sequential Gaussian simulation, GIS, wind profile
A21E-0222
United States Offshore Wind Resource Assessment
The utilization of the offshore wind resource will be necessary if the United States is to meet the goal of having 20% of its electricity generated by wind power because many of the electrical load centers in the country are located along the coastlines. The United States Department of Energy, through its National Renewable Energy Laboratory (NREL), has supported an ongoing project to assess the wind resource for the offshore regions of the contiguous United States including the Great Lakes. Final offshore maps with a horizontal resolution of 200 meters (m) have been completed for Texas, Louisiana, Georgia, northern New England, and the Great Lakes. The ocean wind resource maps extend from the coastline to 50 nautical miles (nm) offshore. The Great Lake maps show the resource for all of the individual lakes. These maps depict the wind resource at 50 m above the water as classes of wind power density. Class 1 represents the lowest available wind resource, while Class 7 is the highest resource. Areas with Class 5 and higher wind resource can be economical for offshore project development. As offshore wind turbine technology improves, areas with Class 4 and higher resource should become economically viable. The wind resource maps are generated using output from a modified numerical weather prediction model combined with a wind flow model. The preliminary modeling is performed by AWS Truewind under subcontract to NREL. The preliminary model estimates are sent to NREL to be validated. NREL validates the preliminary estimates by comparing 50 m model data to available measurements that are extrapolated to 50 m. The validation results are used to modify the preliminary map and produce the final resource map. The sources of offshore wind measurement data include buoys, automated stations, lighthouses, and satellite- derived ocean wind speed data. The wind electric potential is represented as Megawatts (MW) of potential installed capacity and is based on the square kilometers (sq. km) of Class 5 and higher wind resource found in a specific region. NREL uses a factor of 5 MW of installed capacity per sq. km of "windy water" for its raw electric potential calculations. NREL uses Geographic Information System data to break down the offshore wind potential by state, water depth, and distance from shore. The wind potential estimates are based on the updated maps, and on previous offshore resource information for regions where new maps are not available. The estimates are updated as new maps are completed. For example, the updated Texas offshore map shows almost 3000 sq. km of Class 5 resource within 10 nm of shore and nearly 2000 sq. km of Class 5 resource or 10,000 MW of potential installed capacity in water depths of less than 30 m. NREL plans to develop exclusion criteria to further refine the offshore wind potential
A21E-0223
Analysis of Wind Characteristics at United States Tall Tower Measurement Sites
A major initiative of the U.S. Department of Energy (DOE) is to ensure that 20% of the country's electricity is produced by wind energy by the year 2030. An understanding of the boundary layer characteristics, especially at elevated heights greater than 80 meters (m) above the surface is a key factor for wind turbine design, wind plant layout, and identifying potential markets for advanced wind technology. The wind resource group at the DOE National Renewable Energy Laboratory is analyzing wind data collected at tall (80+ m) towers across the United States. The towers established by both public and private initiative, measure wind characteristics at multiple levels above the surface, with the highest measurement levels generally between 80 and 110 m. A few locations have measurements above 200 m. Measurements of wind characteristics over a wide range of heights are useful to: (1) characterize the local and regional wind climate; (2) validate wind resource estimates derived from numerical models; and (3) directly assess and analyze specific wind resource characteristics such as wind speed shear over the turbine blade swept area. The majority of the available public tall tower measurement sites are located between the Appalachian and Rocky Mountains. The towers are not evenly distributed among the states. The states with the largest number of towers include Indiana, Iowa, Missouri, and Kansas. These states have five or six towers collecting data. Other states with multiple tower locations include Texas, Oklahoma, Minnesota, and Ohio. The primary consideration when analyzing the data from the tall towers is identifying tower flow effects that not only can produce slightly misleading average wind speeds, but also significantly misleading wind speed shear values. In addition, the periods-of-record of most tall tower data are only one to two years in length. The short data collection time frame does not significantly affect the diurnal wind speed pattern though it does complicate analysis of seasonal wind patterns. The tall tower data analysis revealed some distinct regional features of wind shear climatology. For example, the wind shear exponent (alpha) at the towers in the Central Plains is generally between 0.15 and 0.25, greater than the commonly used 1/7 power law exponent value of 0.143. Another characteristic of Central Plains wind climatology was that winds from the south had alpha values of 0.2 to 0.3, while northerly winds had lower alpha values from 0.1 to 0.2. The wind resource at a particular tower is affected not only by the regional climatology but also by local conditions such as terrain, surface roughness, and structure of the lower boundary layer.
A21E-0224
Temporal oscillations in convective boundary layers forced by mesoscale surface heat flux variation
A theoretical approach with the equations of horizontal velocity and potential temperature that are low-pass filtered with a mesoscale cutoff wavelength suggests that the surface heterogeneity on a scale of tens of kilometers can generate mesoscale motions that are not in a quasi-stationary state. The transition of the generated mesoscale motions to nonquasi-stationary state occurs when horizontal advection is strong enough to decrease the potential temperature gradient on the surface heterogeneity scale. Large eddy simulations (LES) suggest that the convective boundary layer (CBL) forced by surface heat flux variation with an amplitude of 100 Wm-2 or higher and a wavelength of the order of 10 km becomes in a nonquasi-stationary state. Spectral analysis of the LES reveals that when the mesoscale motions are in a quasi-stationary state, the energy given by the surface heat flux variation remains in organized mesoscale motions on the variation scale. However, in a nonquasi-stationary state of the organized mesoscale motions, the energy cascades to smaller scales. The cascade reaches down to the turbulence in the CBL forced by surface heat flux variation on a scale smaller than 100 times the CBL height. The energy transfer from the generated mesoscale motions to the turbulence makes the spectral gap between the two scales absent. The absence of an obvious spectral gap between the generated mesoscale motions and the turbulence raises a question to the applicability of mesoscale model to the studies on the effect of high-amplitude surface heterogeneity on a scale of tens of kilometers.
A21E-0225
Scanning Doppler Lidar Measurements for Wind Energy Applications
The development of wind energy has increased rapidly along with the size and capacity of wind turbines. These larger machines require detailed wind resource measurements at higher and higher altitudes. Accurate wind speed, wind direction, and turbulence statistics are required for wind resource assessment and efficient wind farm operation. Tower measurements are limited in coverage and do not provide the three dimensional sampling of the atmospheric processes required for accurate model initialization or resource assessment. Remote sensing measurements are the most attractive option for wind energy meteorology. However, the measurement volume must be sufficiently small to resolve the important atmospheric scales and the spatial and temporal measurement domain must satisfy the requirements of the wind energy industry. High resolution profiles of mean and turbulent statistics of the wind field upstream of a wind farm can be produced using a scanning Doppler lidar. Careful corrections for the spatial filtering of the wind field by the lidar pulse produce turbulence estimates equivalent to point sensors but with the added advantage of a larger sampling volume to increase the statistical accuracy of the estimates. For a well designed lidar system, this permits accurate estimates of the mean windspeed and the turbulent statistics over various subdomains and with sufficiently short observation times to monitor rapid changes in conditions. These features may be ideally suited for optimal operation of wind farms and for improved data assimilation for local high resolution forecast models. Results from the analysis of scanning Doppler lidar data collected at the National Renewable Energy Laboratory (NREL) will be presented to highlight some of the fundamental atmospheric processes for wind power meteorology. The unresolved issues for future applications of this technology will be outlined.
A21E-0226
Investigating Interactions Between Wind Turbines and the Atmosphere
With wind energy gaining in popularity and wind farms being developed faster than ever, it is important to quantify any effect these farms have on the atmosphere. Knowing these effects will assist in site selection as well as in optimizing array efficiency. Previous studies have looked at atmosphere-wind farm interaction using models at coarse resolutions. This study goes down to resolutions as fine as 15 m in order to resolve the turbine blades themselves. A simple wind turbine model based on the Blade Element Momentum method is used to determine the forces exerted by the turbine blades onto the incoming air flow. The turbine model results are validated against three different wind turbines – the NREL/NASA Ames UAE turbine, an LM Glasfiber turbine, and a Tjaereborg turbine. Once validated, parameters from the model are used to estimate the wind velocity in the wake of a turbine. The energy lost in a single wake is extrapolated to get a rough estimate of how much energy is lost from the atmosphere from large wind farms. The next step will be to apply the forces from the turbine model into a 3D atmospheric model which will couple the blade forces into the atmospheric flow field. The coupled model will be used to examine turbine wake development and how this feeds back into atmospheric processes.
A21E-0227
Power Flow Simulations of a More Renewable California Grid Utilizing Wind and Solar Insolation Forecasting
Time series power flow analyses of the California electricity grid are performed with extensive addition of intermittent renewable power. The study focuses on the effects of replacing non-renewable and imported (out-of-state) electricity with wind and solar power on the reliability of the transmission grid. Simulations are performed for specific days chosen throughout the year to capture seasonal fluctuations in load, wind, and insolation. Wind farm expansions and new wind farms are proposed based on regional wind resources and time-dependent wind power output is calculated using a meteorological model and the power curves of specific wind turbines. Solar power is incorporated both as centralized and distributed generation. Concentrating solar thermal plants are modeled using local insolation data and the efficiencies of pre-existing plants. Distributed generation from rooftop PV systems is included using regional insolation data, efficiencies of common PV systems, and census data. The additional power output of these technologies offsets power from large natural gas plants and is balanced for the purposes of load matching largely with hydroelectric power and by curtailment when necessary. A quantitative analysis of the effects of this significant shift in the electricity portfolio of the state of California on power availability and transmission line congestion, using a transmission load-flow model, is presented. A sensitivity analysis is also performed to determine the effects of forecasting errors in wind and insolation on load-matching and transmission line congestion.
A21E-0228
Applying canopy flow model for estimation of wind turbine wake
For the planning of large offshore wind farm the optimal spatial placing of wind turbines as well as wind farms relatively to each other is highly important to reduce the wake losses of energy. Conventional instrumental investigations of airflow characteristics around and inside an offshore wind farm aimed at understanding of far-wake behavior are very difficult and expensive. Computational fluid dynamic (CFD) models can provide the information on spatial patterns of wind and turbulence and thus, help to develop the optimal wind farm design. With limited level of model resolution, however, there is still a problem of how to describe the effect of a wind turbine itself on air flow. Having this problem solved the joint effects of a given number of wind turbines could be easily estimated. In present work, to describe the influence of a wind turbine on the flow a coupled canopy-atmospheric boundary-layer model SCADIS is implemented. It has been shown that this model, based on two-equation closure and modified to account for plant drag, is able to simulate airflow through a wide range of vegetation reasonably. In the numerical experiment with SCADIS the turbine's rotor was replaced by a disk of limited thickness, with diameter (D) and location of real rotor but with properties of vegetation. Aerodynamic drag values for this rotor with some 'plant' surface density can be derived from the trust coefficient Cp of the wind turbine of interest. Model results were compared with measurements from the Danish offshore wind farm Vindeby consisted of 11 Bonus 450 kW turbines (hub height and rotor diameter are 38 m and 35 m, respectively). The comparison show that the approach can describe well the single- and double-wake cases (at distance 9.6D behind the last turbine), and quintuple-wake case (at distance 8.6D). Taking in account relatively low the computing time demands of the approach, it is a promising tool for further studies of wakes of offshore wind turbines of any size and composition.
A21E-0229
A Comparison of Synoptic Classification Methods for Application to Wind Power Prediction
Wind energy is a highly variable resource. To make it competitive with other sources of energy for integration on the power grid, at the very least, a day-ahead forecast of power output must be available. In many grid operations worldwide, next-day power output is scheduled in 30 minute intervals and grid management routinely occurs at real time. Maintenance and repairs require costly time to complete and must be scheduled along with normal operations. Revenue is dependent on the reliability of the entire system. In other words, there is financial and managerial benefit to short-term prediction of wind power. One approach to short-term forecasting is to combine a data centric method such as an artificial neural network with a physically based approach like numerical weather prediction (NWP). The key is in associating high-dimensional NWP model output with the most appropriately trained neural network. Because neural networks perform the best in the situations they are designed for, one can hypothesize that if one can identify similar recurring states in historical weather data, this data can be used to train multiple custom designed neural networks to be used when called upon by numerical prediction. Identifying similar recurring states may offer insight to how a neural network forecast can be improved, but amassing the knowledge and utilizing it efficiently in the time required for power prediction would be difficult for a human to master, thus showing the advantage of classification. Classification methods are important tools for short-term forecasting because they can be unsupervised, objective, and computationally quick. They primarily involve categorizing data sets in to dominant weather classes, but there are numerous ways to define a class and a great variety in interpretation of the results. In the present study a collection of classification methods are used on a sampling of atmospheric variables from the North American Regional Reanalysis data set. The results will be discussed in relation to their use for short-term wind power forecasting by neural networks.
A21E-0230
Combining Wind and Wave Energy in Offshore Power Plants to Reduce Variability in Electrical Generation
While wave energy is primarily a wind driven phenomenon, at a particular location and time the energy levels in the wind and waves may be different. The correlation between wind and wave energy is sufficiently weak that combining the two energy sources in a collocated offshore power plant reduces the variability in electrical generation. A preliminary examination of offshore locations along the west coast of the U.S. using buoy data shows two advantages of combining the two energy sources: 1) the number of hours of no power generation in a given year is significantly decreased, which reduces the intermittency of the power plant; 2) a decrease in the variability of the generation curve, which reduces the drops and surges of voltage at the grid interconnection point. The power generation curves for the hypothetical combined wind and wave offshore power plants use atmospheric conditions, wind speed, and wave statistics collected by NOAA buoys, and a common commercial offshore wind turbine model paired with a wave energy convertor in early commercial development in a reasonable array configuration. The hypothetical offshore power plants are located in areas with both a quality wind and wave resource near existing or feasible transmission corridors. Multiple locations along the west coast of the U.S. are used to demonstrate this reduction in power variability and intermittency.
A21E-0231
Wind Power Production and Climate Change--a Modeling Study
Recent studies using Global Climate Models (GCMs) for several climate change scenarios are inconclusive as to the sign of the change in surface wind speeds. Some regions may experience a net increase in boundary layer winds, while other areas observe a decrease. Areas within the U.S. that are most susceptible to climate change also contain substantial wind resources (for example, California and the Great Plains). The next few decades under a changing climate may also see greater variation in seasonal and annual wind speeds, making long-term planning for air quality and wind energy purposes problematic. Thus, the purpose of this presentation is to show preliminary results from a regional-scale study focusing on the effects of climate change on wind in the boundary layer. Under the sponsorship of the California Energy Commission (CEC) and the Lawrence Livermore National Laboratory (LLNL), simulations of future-climate (2040-2060) wind speeds in the surface layer were performed in in order to estimate affects upon wind power production under the IPCC A2 greenhouse gas emissions scenario. Output from high-resolution (50 km) global climate simulations conducted at LLNL for the Department of Energy (DOE) National Science Foundation (NSF)-funded North American Regional Climate Change Prediction Project (NARCCAP) was used to initialize AWS Truewind's Mesoscale Atmospheric Simulation System (MASS) model. Simulations covering the entire state of California, with a grid size of 15 km, and "inner nests" with finer resolution (4.0 km) in the Tehachapi Pass and other wind resource regions were performed. We present the results of these simulations and will discuss the implications for future wind energy resource assessment, air dispersion applications, and energy balance consequences.
A21E-0232
Improved Large-Eddy Simulation For Wind Energy Applications Using the Weather Research and Forecasting Model
Future expansion of wind power production requires resolution of several outstanding research issues, many of which involve the complicated and highly variable near-surface atmospheric flow field. To these ends we have made several improvements to the Weather Research and Forecasting model (WRF) to improve its Large Eddy Simulation (LES) capability. These improvements enable exploration of how terrain heterogeneity, turbulence and low-level shear interact with the larger-scale flow to impact many facets of wind power production, including resource characterization, micrositing and turbine performance and lifespan. We demonstrate the improvements afforded by our improved WRF-LES in high-resolution simulations of flow over complex terrain. Results are shown for flow over both an isolated ridge and flow over terrain consisting of several hills and valleys of various shapes and sizes. The simulations cover a variety of wind speeds and stability conditions. The importance of boundary conditions as well as horizontal and vertical resolution, and the grid aspect ratio, are discussed. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. LLNL-ABS-406846
A21E-0233
Dynamic WRF-WAsP Downscaling
To determine if accurate numerical weather prediction model-based wind power resource assessments can be accomplished, a framework was developed to downscale the coarse Weather Research and Forecasting (WRF) model output with a wind resource analysis tool commonly used within the industry, the Wind Atlas Analysis and Application Program (WAsP). The dynamic downscaling accounts for fine scale topography and surface roughness features that can have a large impact on low-level wind fields. It was found that the WRF model can be used as input into WAsP, and in the future could possibly be a replacement to tower observations when completing preliminary resource assessment projects. This framework would allow the wind power industry to complete site assessment projects in a timely and economically efficient manner.
A21E-0234
Evaluation Of Meteorological Data For Wind Energy Analysis
This study was undertaken to compare wind turbine energy estimates from different meteorological models and to evaluate the strengths and weaknesses in using these models to predict wind patterns and model wind power production. The specific data sources included the Penn State/UCAR Mesoscale Model version 5 (MM5), National Center for Environmental Prediction (NCEP) North American Regional Reanalysis (NARR) and ground-based airport weather data. MM5 is a widely used weather and climate prediction model which employed a 4 km x 4 km resolution over Minnesota and North and South Dakota for the years 2004, 2005, 2006 in the Minnesota Wind Integration Study (MWIS). The NARR dataset is only available at a 32 km x 32 km resolution, but can be retrieved over a long temporal scale, from 1979 to present, and covers the entire North American region. NARR data have previously not been used to investigate wind power potential. The ground-based airport weather data has been used to predict for wind power but wind speeds need to be extrapolated to a wind turbine hub height and data is only available on a limited basis throughout North America. We compared predictions of wind farm capacity of the MWIS MM5 data with the corresponding locations using NARR data. At 1000 millibars of pressure, representing the lowest boundary layer of the earth's troposphere, the NARR wind speed data provided capacity factors that most strongly correlated with the MWIS data and had the lowest average error. Wind energy estimates produced from the NARR database were also used to analyze spatial, diurnal, seasonal and interannual variability. As distance becomes greater between wind turbine locations, the NARR data has shown decreased correlation of wind speeds; this suggests that by having an interconnected wind farm network, challenges of intermittency will be reduced. Seasonal and interannual variation can be observed using the NARR data, suggesting that long term planning is necessary in building wind farms. The NARR data shows promise in siting wind turbine farms, and has potential to analyze wind regimes for planning, operation and storage of wind power production.
A21E-0235
Modeling impact of large wind farms on local and regional hydrometeorology
Previous studies suggest that large-scale wind farms may affect local and regional hydrometeorology. Rigorous assesment of this impact is hampered by lack of adequate parameterization of wind farms in climate models. This paper compares two different approaches: (i) wind farm as a surface roughness element and (ii) turbine rotors as elevated sinks of momentum and source of TKE. Experiments with the RAMS model shows significant differences in simulated local hydrometeorology between these 2 approaches. Preliminary results suggest that the roughness approach is not appropriate for simulating vertical profiles of temperature and moisture near the surface.
A21E-0236
Wind Farms and Weather Modification.
Electrical generation by wind turbines is increasing rapidly, and has been projected to satisfy 15 percent of world electric demand by 2030. The extensive installation of wind farms would alter surface roughness and significantly impact the atmospheric circulation. This forcing could be changed deliberately by adjusting the attitude of the turbine blades with respect to the wind. Using the NCAR Community Atmosphere Model, we model the impact of time-dependent surface roughness changes due to manipulation of a continent-scale wind farm. We show that initial disturbances caused by a step change in roughness grow within four days such that the flow is altered at synoptic scales. The growth rate of the induced perturbations is largest in regions of high atmospheric instability. For a roughness change imposed over North America, the induced perturbations involve substantial changes in the track and development of cyclones over the North Atlantic. For example, in some cases, weather over the British Isles changes from cloudy to clear, depending on whether wind turbines in the American Midwest are "on" or "off" three days beforehand. We explore the dependence of the downstream effects on the size and roughness of the wind farm installation, showing that as the size of individual wind farms and turbines grows, the scale of atmospheric impacts increases in extent and magnitude. We also look at the dependence of the wind farm impacts on the initial state of the atmosphere, confirming that the impacts are largest when the wind farm perturbation projects onto growing error modes in the atmosphere. In particular, rapid growth occurs when the initial disturbance is carried into regions of high baroclinic instability such as the North Atlantic. By running ensemble experiments, we estimate the robustness of the wind farm impacts with respect to realistic uncertainty in the initial conditions. Our results suggest the possibility of a method for weather modification that in some cases could provide added value to very large wind farms.
A21E-0237
Evaluation of Sub-Kilometer Simulations Using Towers for Wind Energy Assessment
It is generally known that to accurately resolve atmospheric processes and winds in topographically developed terrain, regional/mesoscale model resolution should be on the order of a kilometer or so. Since these models are still common tools for producing wind density potential maps, it is important to investigate their capabilities and accuracy compared to measurements. Operation of tall towers in western Nevada offered excellent reference information for verification of the Mesoscale Model 5 (MM5) and Weather and Research Forecasting (WRF) simulations in complex terrain. A principal question is whether a nominal increase in the horizontal resolution uniquely produces more accurate wind and turbulence predictions. In this study, we examined the effect of horizontal resolution on the accuracy of regional/mesoscale model predictions for the near surface (standard) height as well as hub heights using tower data in the first 80 m AGL. The models were set up with nine domains, with the first four having horizontal resolutions of 27, 9, 3, and 1 km, respectively. Within the fourth domain there were five sub-domains, all with 333 m resolution. Three of the sub-domains were centered on a 50m tower at each of the four locations, one of sub-domains is centered on an 80m tower, and one on a surface station equipped with a sonic anemometer. The 50m towers were equipped with wind speed measurements at every 10 m and wind direction measurements at 10 and 50 m. The 80m tower had wind speed and direction measurements at 40, 60, and 80m using standard cup and vane as well as 3D sonic anemometers at each level. These measurements provided an excellent set for evaluation of the predicted wind and turbulence aloft with respect to extremes, diurnal variation, and gustiness as well as the success of the various models' physical parameterizations.
A21E-0238
A wind tunnel investigation of wind turbine wakes: Boundary-layer turbulence and surface roughness effects
Wind turbine wakes are known to have an important effect on power generation and fatigue loads in wind energy parks. Wake characteristics are expected to depend on the incoming atmospheric boundary layer flow statistics (mean velocity and turbulence levels). Here, results are presented from a wind tunnel experiment carried out at the St. Anthony Falls Laboratory atmospheric boundary layer wind tunnel to study turbulence levels in the wake of a model wind turbine placed over both rough and smooth surfaces. How-wire anemometry was used to characterize the cross-sectional distribution of turbulent intensity, kinematic shear stress and mean velocity at different locations downwind of the turbine for both surface roughness cases. Non-axisymmetric behavior of the wake is observed over both roughness types in response to the non- uniformity of the incoming boundary layer flow and the presence of the surface. Nevertheless, the velocity deficit with respect to the average incoming flow is nearly axisymmetric everywhere except near the surface in the far wake, where the wake interacts with the surface. It was found that the wind turbine induces a large enhancement of turbulence levels in the upper part of the wake. This is due to the effect of relatively large velocity fluctuations associated with helicoidal tip vortices near the wake edge. In the lower part of the wake, where the incoming flow has lower average velocity and higher turbulence levels, the turbulence intensity shows a small reduction. The non-axisymmetry of the turbulent intensity distribution of the wake is found to be stronger over the rough surface, where the incoming flow is less uniform at the turbine level. It was found that the average turbulent intensity produced by the wake, its positive and negative components and its local maximum decay as a power law of downwind distance (with a power of -0.3 to -0.5 for the rough surface and with a wider range for the smooth surface). Preliminary results will also be presented on the effect of thermal stratification and the presence of multiple turbines.
A21E-0239
Wind energy resource assessment using coupled groundwater-land-surface atmospheric models
By modifying the amounts of sensible and latent heat available to drive atmospheric boundary layer
dynamics, soil moisture variability can substantially modify winds and wind shear in the lower atmosphere.
Soil moisture in turn depends on groundwater flow as well as atmospheric forcing. This model
intercomparison study assesses the impact of heterogeneous and time-dependent soil moisture forcing on
low-level winds relevant for wind energy generation in the southern Great Plains. Simulations using coupled
models, variably-saturated groundwater flow model (ParFlow) and mesoscale atmospheric models (WRF and
ARPS), allow examination of the effects of soil moisture heterogeneity on atmospheric boundary layer
processes. These parallel, integrated models can simulate spatial variations in landsurface forcing driven by
three-dimensional (3D) atmospheric and subsurface components.
Test cases are presented with both fully-coupled models (which include 3D groundwater flow and surface
water routing) and the uncoupled atmospheric models. The effects of the different soil moisture initializations
and lateral subsurface and surface water flow are seen in the differences in atmospheric evolution, boundary
layer and wind and wind shear over a seven-day time period. Sensitivity to vertical resolution of the
atmospheric model, particularly for resolving nocturnal flow features, is explored.
This work is performed under the auspices of the U.S. Department of Energy by Lawrence Livermore
National Laboratory under Contract DE-AC52-07NA27344.
A21E-0240
Remote sensing of the wind and turbulence characteristics at the heights of modern wind turbines
Remote sensing estimates of wind resource potential and turbulence structure of the boundary layer at the heights of turbine rotors is very important as the height reached by commercial turbines increases up to 200- 250 m to take advantage of stronger wind speeds at higher altitudes. The fine temporal and spatial resolution of Doppler lidar observations can provide near-continuous information about wind and turbulence conditions at turbine height and above the range of tower measurements. Flexibility of lidar to perform conical, vertical-slice or fixed-beam measurements, allows researchers to obtain a variety of boundary-layer characteristics and to investigate wind-field structure from different points of view, for example as mean wind and turbulent profiles or as images of individual scans to investigate flow features. This paper, as a part of an ongoing effort to extract quantities of interest for wind-power meteorology from existing lidar data sets, presents information on quantities, such as nighttime boundary layer height, distribution of wind, turbulence intensity, wind and directional shear across the turbine rotor layer, and power- law exponent, obtained from lidar observations during two experiments conducted in the Great Plains determined to be an important wind-resource region in the U.S. These measurements focus on the nighttime low-level jet, a flow feature that provides the potential for wind energy during the warm season. Accurate estimates of dynamics and nighttime evolution of these quantities through remote sensing measurements provide a better understanding of the atmosphere-turbine interactions under nighttime stable conditions and offer wind-farm developers a needed decision-making information important to the success of wind energy.
A21E-0241
Electricity from Wind: Global Perspectives
The potential of wind power as a source of electricity both globally and more specifically for the continental US is assessed using wind fields assimilated by GEOS-5. Analysis of data for 2006 suggests that a network of land-based 2.5 MW turbines restricted to non-forested, ice-free, non-urban areas with wind resources sufficient to accommodate turbines operating with capacity factors greater than 20%, could supply electricity greater than current global consumption by a factor of 40, greater than current total global use of energy in all forms by a factor of 5. Resources in the US, specifically in the central plain states, subject to the same constraints, would be sufficient to supply more than 16 times current total US demand. Estimates are offered also for the potential yield of electricity from 3.6 MW turbines deployed both globally and for the US in ocean waters with depths less than 200m within 50 nm of closest coastlines.
A21E-0242
Dynamical and Statistical Wind Downscaling in the Northeast of the Iberian Peninsula
Estimations of possible changes of wind variability at the regional scale as a response to the evolution of large scale climate entail relevant economic and ecological implications for society, as for instance, the assessment of the variations and sustainability in wind energy resources. Not only in this context but also from a meteorological point of view, the evaluation of surface wind variability involves many interesting aspects that are worth to be analyzed. The limited reliability of the general circulation models at the regional/local scale requires the use of downscaling techniques to derive regional climate variability from the large scale circulation. Dynamical, statistical or a combination or both approaches can be applied to the downscaling problem to explore the wind field behavior in the region of interest. In this work, the potential predictability of the wind speed is evaluated by means of its relationship with the atmospheric circulation over the North Atlantic area using different methodologies. For this aim, wind speed observations from the region of Navarra, Northeast of the Iberian Peninsula, are employed; the data span a 14 years period, from 1992 to 2005. A dynamical downscaling using the Weather Research and Forecast (WRF) model is used to analyze the wind variability at daily time scales. The spatial wind variability is analyzed by dividing the region into various subregions by means of cluster analysis. The temporal variability is addressed by classifying the wind fields into weather types (wind circulation types) with similar spatial structure. The model is skillful in identifying the observed subregions and in reproducing the temporal wind variability at most of them. In addition, the spatial structure of the wind circulation types is generally reproduced by the simulation, with a tendency to underestimate the spatial wind speed variability. The statistical methodology explores the variability of wind speed and also wind power production at monthly timescales and consists in a linear technique which isolates optimal correlated modes of variability between the synoptic fields over the North Atlantic and the observed wind velocity (Canonical Correlation Analysis). Results evidence the existence of wind predictability in the region of study at monthly timescales. An assessment of the sensitivity of the methodology is performed as a first step in the evaluation of the potential sources of uncertainty affecting the regional estimations of the wind field. The statistical relationship found during the period of available observations is used to perform a climatological reconstruction of the surface wind field within the last five centuries using reanalysis, observational and reconstruction data sources. This evaluation of past wind variability could have relevant applications for the study of regional wind predictability over the 21th century.
A21E-0243
Offshore Wind Resource Assessment Based on Mesoscale Modeling
A methodology for assessing regional offshore wind energy development potential using mesoscale modeling for wind fields has been developed. Recommendations are made on selecting the best mesoscale modeling domain resolution, as well as choosing the best data for model initial and boundary conditions, based on a sensitivity study using the Penn State/NCAR MM5 mesoscale model near California coast validated with offshore buoy wind data and coastal meteorological stations. Annual wind speed averages are developed by modeling four seasonal months to reduce total computational time, as well as to allow study of the innterannual variability. Four seasonal months of 2005, 2006, and 2007 were compared to using a complete modeled year for 2007 to calculate how the overall energy answer changed. Results from summer 2006 MM5 simulations show the average 10 m wind speed to be calculated within one percent when using three months of data (Jun, Jul, Aug) versus using July alone. Siting restrictions were developed based on bathymetry depth limits for offshore turbine tower support structures with economic and structural limitations for monopiles, multi-leg, and future floating tower support types corresponding to 30, 70, and 200 m depth respectively. Other exclusionary entities such as shipping lanes and avarian flyways were also considered as exclusion zones inside of areas amenable for offshore wind energy farms. A method to validate the modeled wind fields though error calculations against offshore buoy wind data, as well as onshore coastal meteorological towers is presented.
A21E-0244
LES of wind turbine wakes: Evaluation of turbine parameterizations and dynamic subgrid- scale models
Large-eddy simulation (LES), coupled with a wind turbine model, is used to investigate the characteristics of wind turbine wake turbulence in both idealized free stream flow and turbulent boundary layer flow. Three different subgrid-scale (SGS) models for SGS stresses are tested: (a) the Smagorinsky model, (b) the Lagrangian dynamic model, and (c) the recently developed scale-dependent Lagrangian dynamic model (Stoll and Porté-Agel, 2006). The turbine-induced drag force is parameterized using two methods, the actuator-disk method and the actuator-line technique, which distribute the force loading of blades on a disk and along lines, respectively. Simulation results obtained with all SGS models and wind turbine models are compared with both field and wind tunnel measurements. The wind tunnel data were collected using hot wire anemometry in the wake of a miniature wind turbine at the St. Anthony Falls Laboratory atmospheric boundary layer wind tunnel. We find that the characteristics of the turbine wake are sensitive to the parameters used in the actuator line/disk methods. For a given turbine parameterization, the scale- dependent dynamic SGS model is able to account, without any tuning, for the local changes in the eddy- viscosity model coefficient at different positions in the wake. It can also capture the scale dependence of this coefficient associated with flow anisotropy at the smallest resolved and sub-grid scales in regions of the flow with strong mean shear. As a result, the scale-dependent dynamic model yields results that are more realistic than the ones obtained with the standard Smagorinsky model and the scale-invariant Lagrangian dynamic model.
A21E-0245
The Influence of Stable Boundary Layer Flows on Wind Turbine Fatigue Loads
This study aims to address two main issues: (i) the generation of high-resolution four-dimensional inflow turbulence fields for a range of atmospheric stability conditions; and (ii) the comparison of fatigue loads on utility-scale turbines for different atmospheric flows (e.g., night-time stable boundary layers and associated low-level jet events). Presented here is an attempt to exploit recent computational, observational, and statistical developments with a view to gaining a better understanding of atmospheric boundary layer flows and its influence on wind turbine loads and design. Over the last two decades, there has been considerable wind energy development in the Great Plains region of the U.S. Nocturnal low-level jets (LLJs) occur quite frequently in this region. The peaks (local wind maxima) of LLJs are typically centered 100-1,000 m above the ground level and make the Great Plains' wind resources very favorable for wind energy production. At the same time, though, the presence of LLJs can significantly modify vertical shear and night-time turbulence environments in the vicinities of wind turbine hub heights and have detrimental effects on rotors. Thus, accurate numerical modeling and forecasting of LLJs are needed for robust wind turbine design and more reliable power-generation prediction. Since stable stratification is a prerequisite for the occurrence of nocturnal LLJs, this then requires an improved modeling capability of stable boundary layers (SBLs). By statistically merging different types of LES-generated flow data sets, it is hoped that a consistent atmospheric turbulence database can be created for various wind turbine loads studies planned. The need for such a database is important in simulating wind turbine loads since it can provide an opportunity to examine the influence of realistic inflow turbulence over the rotor in different atmospheric conditions. Wind turbine aeroelastic simulations are carried out; these are employed to quantify turbine loads in the form of time histories that are then used to statistically estimate fatigue loads.
A21E-0246
Testbeds for Wind Resource Characterization: Needs and Potential Facilities
With the emergence of wind as a significant source of alternative energy, it is becoming increasingly clear that some problems associated with the installation and operation of wind plants arise because of continuing gaps in our knowledge of fundamental physical processes in the lower atmospheric boundary layer. Over the years, a number of well-designed intensive field campaigns have yielded significant insight into boundary layer structure and turbulence under targeted conditions. However, to be able to usefully simulate the atmosphere for applications of wind power, it is important to evaluate the resulting parameterizations under a realistic spectrum of atmospheric conditions. To do this, facilities — testbeds — are required that operate continually over long periods. Such facilities could also be used, among other things, to establish long-term statistics of mean wind and low-level shear, to explore the representativeness of shorter-period (e.g. one year) statistics, to explore techniques for extrapolating wind statistics in space, and to serve as host infrastructure for boundary layer campaigns targeted to wind energy applications. During the last half of the 20th century, a number of tall instrumented towers were installed at locations around the United States for studies of atmospheric dispersion and other purposes. Many of these are no longer in service, but some have operated continuously for decades and continue to collect calibrated wind and temperature information from multiple heights extending to hub height or higher for many current operational wind turbines. This talk will review the status of tall towers in the U.S. that could anchor testbeds for research related wind power production and will use data from the 120-m meteorological tower on the Hanford Site in southeastern Washington State to illustrate the kind of information is available.
A21E-0247
Short Term Wind Forecasting for Sites with Abrupt Roughness and Thermal Changes
With the increasing penetration of wind power to the electrical grid, the importance of short-term wind energy forecasting from hours to 2-3 days ahead has been recognized. Typically, the outputs of Numerical Weather Prediction (NWP) models are used to drive the forecasting system with forecast horizons of a few hours. We are studying the coupling of the local-scale models with North America regional NWP models such as GEM and NAM and nested meso-scale models for site specific wind and wind energy forecasting for wind farms near abrupt roughness and thermal changes. The goal is for real-time wind forecasts from 1 hour to 48 hours and comparison with field measurements at one or more sites in southern Ontario. Local, site specific, winds are affected on a local scale by a variety of factors. These include topography, on a range of scales, surface roughness and its spatial variation, surface temperatures or thermal properties and wakes behind surface mounted obstacles. On the meso-scale, effects such as sea or lake breezes and channelling effects are important factors. These local effects are generally not properly represented in meso-scale models, with a resolution of order 2-10 km. We will use various methods to simulate these local effects in our forecasting system. These include numerical vertical and horizontal interpolations and the use of models of flow in complex terrain. The results will be examined using the Mean Absolute Error and Root Mean Squared Error (RMSE). A decomposition of RMSE into amplitude and phase error will assist in identifying the forecasting errors and selecting MOS procedures for improving the forecasting. A comparison between the forecasted and measured wind speed at 80-m, the typical turbine hub height, shows encouraging results.
A21E-0248
An 80m Coastal Wind Power Assessment Using QuikSCAT
Steadier and faster offshore winds provide a potentially higher and more continuous source of energy. Companies are actively pursuing technology which allows for wind turbines to be placed in deeper waters (>100 m) farther away from the coast. Typical hub heights of modern wind turbines are near 80 m. We use wind profile correction methods and bathymetric contours to highlight coastal regions where extraction of wind power at 80 m is feasible. Observed (2000--2006) 10 m surface winds from NASA's SeaWinds scatterometer measurements onboard QuikSCAT are extrapolated to 80 m using Monin- Obukhov similarity theory. A Weibull probability distribution function (PDF) is fitted to these twice-daily wind speed observations. 80 m wind power density is calculated using the full and truncated (between cut-in and cut-out speeds of typical wind turbines) PDF. Mean 2000--2006 80-10 m wind speed differences range from <2 m s-1 for unstable boundary layers to >3 m s-1 for stably stratified boundary layers over coastal waters near Nova Scotia and east of Argentina. Near Japan, climatological 80 m wind power densities are double 10 m wind power densities. Boreal wintertime wind power densities calculated for usable wind speeds are 15% and 17% lower than full PDF wind power densities for gap wind regions near Vladivostok and Japan, respectively.