Oklahoma Wind Power Initiative OWPI Tim Hughes OU

Oklahoma Wind Power Initiative (OWPI) Tim Hughes (OU) Mark Shafer (OU) Troy Simonsen (OU) Jeremy Traurig (OU) Nick Mirskey (OU) Steve Stadler (OSU) Pete Earls (OSU)

Wind Energy: Cost of Wind-Generated Electricity 1980 to 2005 Levelized Cents/k. Wh 40 38 cents 35 Cents per k. Wh 30 25 20 15 10 5 0 '84 ‘ 85 '88 8 '89 '91 6 '92 '95 4 '97 ''00 ‘ 00 2. 5 - 3. 5* 2005 * Assumptions: Levelized cost at excellent wind sites, large project size, not including PTC (post 1994)

OWPI GOALS: • Resource Assessment • Policy Study • Outreach • Educational programs • Community meetings • Promote Economic Development

The Oklahoma Mesonet

The Oklahoma Mesonet 1. 115 Active stations, spaced ~32 km 2. 5 -minute resolution data 3. Standard Meteorological variables following WMO standards 1. 10 -meter wind speed and direction; scaled to 50 -meters using: U/Ur = (Z/Zr)1/6 • Rural sites; generally good fetch conditions • 1994 -2000 data used in study

OWPAI

Oklahoma Wind Resource Maps Purple = class 5 or better, blue = class 4, lt. blue = class 3

Steps for Developing Models • Review Mesonet site surroundings to qualify “fetch conditions” of site, using: – aerial photos (DOQQs) – vegetation (LU/LC) – site panorama photos • Assign subjective ratings of ‘poor’, ‘fair’, ‘good’ or ‘excellent’

NORMAN Air Photo (zoomed) 250 m NORM wind rose 500 m

Steps for Developing Models • Combine information from: • Mesonet Station data (wind, pressure, temperature - 735, 000 readings of each per station) h. DEM elevation data h. Vegetation data (roughness) • Input into two different models: • analytical model (Windmap) • empirical model (using neural networks for non-linear relationship)

Wind. Map Software (Analytical) INPUTS OUTPUTS Arc. View* DEM Data Elevation Grid Arc. View* LULC (GAP) Map 10 Meter Winds *A GIS Software Package Roughness Grid W I N D M A P Final Winds Map Power Density Map Turbine Output Map

MODELED LONG-TERM AVERAGE WIND POWER DENSITY 50 METERS (164 FT. ) Above Ground Level Analytical model output OWPI DRAFT 9/2001

Neural Network (Empirical) • Correlate wind power values calculated at Mesonet sites, with neural network scheme, to: – site elevations – north and south terrain exposures – north and south average roughness • Get equations for wind power density as function of the above • Fill in grid for whole state

Average Wind energy rose using wind data from 78 stations with ‘good-excellent’ rating on fetch conditions Wind energy in N + S wedges = 89% of total Realizable energy from turbines: > 95% from N & S North Wedge 146 deg 34 deg 326 deg 214 deg South Wedge

Sample calculated WPDs and elevation, terrain exposure, and roughness averages + 57 more …….

Wind Power Map for Oklahoma (Empirical Model using Neural Networks)

Analytical Model Findings • Initial run underestimated wind power density at most Mesonet sites • Linear regression of predicted vs. calculated wind power density yielded correction factor of 1. 33 • Better agreement with field data from validation site, but still underestimates in Southeast Oklahoma

Empirical Model Findings • Emphasizes ridge lines (areas of good exposure • Low Roughness • Good Terrain Exposure • Compared to Wind. Map and Tower data, likely underestimates, especially in Southeast Oklahoma

NREL Resource Maps Purple = class 5 or better, blue = class 4, lt. blue = class 3

Analytical Model OWPI DRAFT 9/2001

Empirical Model

OWPI’s Oklahoma Wind Climatology Products

Wind Climatology Cheyenne Mesonet Site Station ID: CHEY Class 3 Site (January ’ 94 – December ’ 00) Average 10 m Wind Speed = 5. 70 m/s (12. 8 mph) Average 10 m Power Density = 189 W/m 2

Cheyenne Wind Energy Rose

Cheyenne

Cheyenne

For information on OWPI: • Oklahoma Wind Power Initiative • www. seic. okstate. edu/owpai • Contact Tim Hughes: thughes@ou. edu For information on OREC: • Oklahoma Renewable Energy Council • www. seic. okstate. edu/OREC