DEALING WITH UNCERTAINTY: Wind Resource
Assessment
D. C. McKay ORTECH Power
Presented atPresented atEnvironmental Finance Environmental Finance
Workshop Series Workshop Series University of TorontoUniversity of Toronto
October 12, 2005October 12, 2005
10 Steps in Building a Wind Farm
• Understand Your Wind Resource
• Determine Proximity to Existing Transmission
Lines
• Secure Access to Land
• Establish Access to Capital
• Identify Reliable Power Purchaser or Market
• Address Siting and Project Feasibility
Considerations
• Understand Wind Energy’s Economics
• Obtain Zoning and Permitting Expertise
• Establish Dialogue with Turbine Manufacturers
and
Project Developers
• Secure Agreements to Meet O&M Needs
Considering a Wind Farm?
Need to Consider• Revenue• Capital Costs• Operational CostsAll carry uncertainty
Why Estimate Uncertainty?• Uncertainty is inevitable• Understanding its origin is
important to:– Know it– Control it– Be prepared for it
Who wants to know?
• You– To set contingencies– To conduct realistic sensitivity
analyses with financial model– To assess project feasibility– To qualify for competitive
financing
• Your lender/ financier
Uncertainty on Revenue side: Wind Resource Assessment• Wind shear• Long term Variation• Monitoring• Wake Estimate• Noise• Power Curve
Sources of Uncertainty:Wind Shear
Sources of Uncertainty:Wind Shear• Profiling & Extrapolation
– Log law or power lawU(z1)/U(z2)=(z1/z2)^p
– p ~ height, roughness, terrain, direction & stability
– wake & turbulence
Sources of Uncertainty:Wind Shear• Can only be eliminated if wind
is monitored at hub height• Often no hub height
measurement available when feasibility of project is assessed
Sources of Uncertainty:Wind Shear
•Uncertainty value for Wind Shear
+/- 20-25%Sources:• Wind Resource Analysis Program 2002,
Minnesota Department of Commerce, http://www.state.mn.us/mn/externalDocs/WRAP_Report_110702040352_WRAP2002.pdf
• Project specific estimates
Sources of Uncertainty:Long Term Variation
Annual Variations in Wind Speed and Energy Production
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
Year
No
rma
liz
ed
Win
d S
pe
ed
/ E
ne
rgy
Normalized Wind Speed
Normalized Wind Energy
Sources of Uncertainty:Long Term Variation• E.g.
– 25 years long term data available (d.o.f. = 24), standard deviation of sample (s = 15%)=> good measure of year to year variation
– 99% confidence interval = 7%
t…student-t , t(d.o.f., confidence level)=> good measure of long term average
Sources of Uncertainty:Long Term Variation
• Climate Change– Mean levels of wind energy– Fluctuations of wind energy
Sources of Uncertainty:Wind Resource Monitoring• Systematic error• Calibration of instrument
– Quality of instrument– Installation (effects of tower,
mounting arrangements)– Surrounding terrain, obstructions,
etc.– Instrument icing/ malfunctioning– Type B ≈ +/- 5%
• Random error– Data recovery rate, electronic
noise– Reduced by increasing number
of samples– Type A ≈ +/- 1%
Sources of Uncertainty:Wind Modelling
• Wind Models– Flow Model Vs Wind Climate Model– Diagnostic Vs Prognostic– Meso-scale Vs Micro-scale (Coupling)– Physics (hydrostatic / non-hydrostatic,
compressible / non-compressible, friction, turbulence closure)
Sources of Uncertainty:Wind Modeling
• Input to Models– Land Use, Seasonal Variations– Terrain (resolution & accuracy)– External Forcing (pressure
gradients, solar radiation, stratification, temperature difference between land and water)
Sources of Uncertainty:Wind Modelling
• Wake Modelling(project specific estimate of 2%)
• Model ValidationDifference between WAsP and MS-Micro Models <2% on project example
Difference between WAsP and more advanced models 25%+
• Noise Modelling
Sources of Uncertainty:Turbine Power Curve
Sources of Uncertainty:Turbine Power Curve
Other Factors in Production Estimating
• Power curve guarantee• Availability and maintenance time• Electrical losses• Time dependent performance
deterioration (blade soiling)• Blade icing and extreme weather
Combination of Uncertainties• Project example
Contribution +/- TotalWind Shear 20.0%Long Term Variation 7.0%Monitoring 5.1%Modeling 5.0%Wake Estimate 2.0%Power Curve 7.0%
23.5%
Summary
• Rational quantification of revenue estimate uncertainty is essential
• Wind shear is often biggest contributor to uncertainty
• Redundant modeling helps to keep model uncertainty down
• Monitoring at as many locations as possible and as close as possible to hub height will reduce uncertainty
• Other loss factors need be considered