Eric White, Director of Engineering

Presentation on theme: "Eric White, Director of Engineering"— Presentation transcript:

Eric White, Director of Engineering
Wind Resource Assessment and Energy Production: From Pre-Construction to Post Construction May 2008 Eric White, Director of Engineering AWS Truewind, LLC 463 New Karner Road Albany, NY © 2007 AWS Truewind, LLC

AWS Truewind - Overview
Industry Leader & Consultant for 20,000+ MW Full spectrum of wind farm development and evaluation services Wind Assessment, Mapping, Engineering, Performance Assessment, Forecasting In business 25 years Project roles in over 50 countries Albany, New York based; 55+ employees © 2007 AWS Truewind, LLC

Topics Wind resource assessment process Energy production modeling
Wind Farms Measurements Siting Wind resource assessment process Energy production modeling Effects on wind farm siting and design Operational plants and actual yields © 2007 AWS Truewind, LLC

Establishing Project Viability
Wind Resources Determine: Project Location & Size Tower Height Turbine Selection & Layout Energy Production annual, seasonal on- & off-peak Cost of Energy/Cash Flow Warranty Terms Size of Emissions Credits The wind energy industry is more demanding of wind speed accuracy than any other industry. © 2007 AWS Truewind, LLC

Power in the Wind (Watts)
= 1/2 x air density x swept rotor area x (wind speed)3 A V3 Density = P/(RxT) P - pressure (Pa) R - specific gas constant (287 J/kgK) T - air temperature (K) Area =  r2 Instantaneous Speed (not mean speed) kg/m3 m2 m/s Knowledge of local wind speed is critical to evaluating the available power © 2007 AWS Truewind, LLC

Wind Shear The change in horizontal wind speed with height
A function of wind speed, surface roughness (may vary with wind direction), and atmospheric stability (changes from day to night) Wind shear exponents are higher at low wind speeds, above rough surfaces, and during stable conditions Typical exponent () values: : water/beach : gently rolling farmland : forests/mountains Wind Shear Profile V2= 7.7 m/s V1 = 7.0 m/s Z2= 80 m Z1= 50 m Wind speed, and available power, generally increase significantly with height V2 = V1(Z2/Z1)  = Log10 [V2/V1] Log10 [Z2/Z1] © 2007 AWS Truewind, LLC

Wind Resource Assessment Handbook Fundamentals for Conducting a Successful Monitoring Program
Prepared for: National Renewable Energy Laboratory 1617 Cole Boulevard Golden, CO NREL Subcontract No. TAT Prepared By: AWS Scientific, Inc. 255 Fuller Road Albany, NY April 1997 Published by NREL Peer reviewed Technical & comprehensive Topics include: Siting tools Measurement instrumentation Installation Operation & maintenance Data collection & handling Data validation & reporting Costs & labor requirements © 2007 AWS Truewind, LLC

Summary of Wind Resource Assessment Process
Identify Attractive Candidate Sites Collect >1 yr On Site Wind Data Using Tall Towers Adjust Data for Height and for Long-Term Climatic Conditions Use Model to Extrapolate Measurements to All Proposed Wind Turbine Locations Predict Energy Output From Turbines Quantify Uncertainties © 2007 AWS Truewind, LLC

Siting Main Objective: Identify viable wind project sites
Main Attributes: Adequate winds Generally > 7 hub height Access to transmission Permit approval reasonably attainable Sufficient land area for target project size 30 – 50 acres per MW for arrays 8 – 12 MW per mile for single row on ridgeline © 2007 AWS Truewind, LLC

Sources of Wind Resource Info
Existing Data (surface & upper air) usually not where needed use limited to general impressions potentially misleading Modeling/Mapping integrates wind data with terrain, surface roughness & other features New Measurements site specific using towers & other measurement systems © 2007 AWS Truewind, LLC

Alternative Sources of Wind Speed Data
Visual Indications and local knowledge Beaufort Scale Wind Speed Estimated by Visual Effects on Land Features Accuracy : 15% Without Height Adjustment National Weather Service (NWS) Measurements of Wind Speeds for Weather Conditions Accuracy : +/- 1 m/s up to 10 m/s % above 10 m/s Environmental Protection Agency Accuracy: 0.25 m/s < 5 m/s 5% > 2 m/s not to exceed 2.5 m/s World Meteorological Organization Accuracy: 0.5 m/s < 5 m/s 10% > 5 m/s Wind Industry expects 1-2% speed measurement accuracy! Alternative sources and techniques have large uncertainty; the wind industry is more demanding of accuracy than traditional users © 2008 AWS Truewind, LLC © 2007 AWS Truewind, LLC 11

Old and new wind maps of the Dakotas
Modern Wind Maps Utilize mesoscale numerical weather models High spatial resolution ( m grid = 3-10 acre squares) Simulate land/sea breezes, low level jets, channeling Give wind speed estimates at multiple heights Extensively validated Std error typically 4-7% GIS compatible Reduce development risks Old and new wind maps of the Dakotas Source: NREL © 2007 AWS Truewind, LLC

Wind Mapping © 2007 AWS Truewind, LLC

Typical Monitoring Tower
Heights up to 60 m Tubular pole supported by guy wires Installed in ~ 2 days without foundation using 4-5 people Solar powered; cellular data communications © 2007 AWS Truewind, LLC

How and What To Measure Anemometers, Vanes, Data Loggers, Masts
Measured Parameters wind speed, direction, temperature 1-3 second sampling; 10-min or hourly recording Derived Parameters wind shear, turbulence intensity, air density Multiple measurement heights best to measure at hub height can use shorter masts by using wind shear derived from two other heights to extrapolate speeds to hub height Multiple tower locations, especially in complex terrain Specialty measurements of growing importance Sodar, vertical velocity & turbulence in complex terrain © 2007 AWS Truewind, LLC

Predicting Long-Term Wind Conditions From Short-Term Measurements
Measure - Correlate - Predict Technique Measure one year of data on-site using a tall tower Correlate with one or more regional climate reference stations Need high r2 Reference station must have long-term stability Upper-air rawinsonde data may be better than other sources for correlation purposes Predict long-term (7+ yrs) wind characteristics at project site Airport A Regression y = x R 2 = Airport B Regression y = x = 0.875 Airport C Regression y = x = 5 10 15 20 25 Reference Station Mean Wind Speed (m/s) Project Site 60 m Wind Speed (m/s) Airport A Airport B Airport C This plot compares a site’s hourly data with three regional airport stations. A multiple regression resulted in an r2 of 0.92. © 2007 AWS Truewind, LLC

Conceptual Project Estimated Net Capacity Factor ~ 29.0 – 31.5%
Software tools (WindFarmer, WindFarm, WindPro) are available to optimize the location and performance of wind turbines, once the wind resource grid within a project area is defined. Estimated Net Capacity Factor ~ 29.0 – 31.5% © 2007 AWS Truewind, LLC

Elements of Energy Production Analysis & Reporting
Site/Instrument Description Wind Data Summary Long-term Speed Projection Turbine Power Curve Turbine Number & Layout Gross Energy Production Loss Estimates Uncertainty Analysis Net Annual Energy Production (P50, P75, P90, etc.) © 2007 AWS Truewind, LLC

Influences on Uncertainty
(Typical Range of Impact on Lifetime Energy Production) Measured Speed Shear Climate Resource Model Plant Losses (2-4%) Sensor Types, Calibration & Redundancy, Ice-Free, Exposure on Mast, # of Masts (1-3%) Height of Masts, Multiple Data Heights, Sodar, Terrain & Land Cover Variability (4-9%) Measurement Duration, Period of Reference Station, Quality of Correlation (5-10%) Microscale Model Type, Project Size, Terrain Complexity, # of Masts, Grid Res. (1-3%) Turbine Spacing (wakes), Blade Icing & Soiling, Cold Temp Shutdown, High Wind Hysteresis, etc. © 2007 AWS Truewind, LLC

So What’s a P90? For a function with an assumed normal probability distribution, P50 = mean of the distribution P90 = the point where 90% of the results are expected to be above Both are important © 2007 AWS Truewind, LLC

Operational Plant Performance:
Understanding operational wind farm performance can be non-trivial Typical Plant Operational Data Project Availability Plant Production Need to separate wind variability from other effects to understand real long term expectations © 2007 AWS Truewind, LLC

Characteristics of an Operational Assessment
Information from actual operations improves estimate Many sources of uncertainty can be removed (measured speed, shear, resource model, plant losses, actual turbine performance) Monthly farm level numbers help smooth and linearize results (Power vs. wind speed) Climatological adjustment with reference station data to provide long term trends Can focus directly on the bottom line data (Revenue Meter) Typical Wind Farm Power Curve (After correcting for Availability) 100% Avail Plant Output Nacelle Average Wind Speed © 2007 AWS Truewind, LLC

Case Comparison - Reference Wind
12 month Rolling Average Wind Speed Nacelle Average Trends are similar Annualized Wind Speed shows low period in 04 and 05; should expect low production Wind Speed (m/s) Ref Station 2 Ref Station 1 © 2007 AWS Truewind, LLC Month

Case Comparison – Operational Performance

A Few Words on Plant Underperformance - Understanding the Issue
Plant underperformance vs preconstruction estimates is a real and significant issue for the industry Many plants averaging ~ 8 to 10% below projections Contributions from numerous sources Can be hard to pinpoint and evaluate Wind variability complicates the analysis Many aspects can be addressed Need to peel back the onion to understand the real issues involved - Studies in process © 2008 AWS Truewind Confidential © 2007 AWS Truewind, LLC

Early Findings – Some Key Contributors
Regional Climate and Variability Resource Assessment Campaign Biases Actual Plant Availability As-Built Plant Characteristics Changes from Plan Sub-Optimal Operation Many contributors – across the project development cycle © 2008 AWS Truewind Confidential © 2007 AWS Truewind, LLC

Summary Wind conditions are site-specific and variable, but predictable over the long term. Accuracy is important. Wind resource assessment programs must be designed to maximize accuracy. Combination of measurement and modeling techniques can give projections close to those experienced in actual operations. Operational plant evaluations can be used to improve projections © 2007 AWS Truewind, LLC

Similar presentations