1 Integrating Wind into the Transmission Grid Michael C Brower, PhD AWS Truewind LLC Albany, New York
2 About AWS Truewind Providing integrated consulting services to the wind industry Responsible for the Irish Wind Atlas (with ESBI, initiated by SEI) Forecasting for MW of wind plant in US and Europe Conducted wind integration studies in US
3 Time Scales – Electric Power Regulation: seconds to minutes Load following: minutes to hours Unit commitment: hours to days Reliability: months to years
4 Time Scales – Wind
5 53 MW Capacity Wind and Wind Plant Variability Not the Same +33% +10%
6 Propagation of Gusts Through a Wind Farm
7 Mean Change in Power vs Number of Turbines at Flat Rock
8 Spatial Diversity of Turbine Output Correlation coefficient of power change for different average times over the distance From Ernst et al, 1999
9 Typical 4-Hr PIRP Forecast Performance San Gorgonio Pass, California - May 2003 Wind Forecasting
10 Forecast Accuracy Vs Time
11 Forecast Accuracy Vs Output 3-Hour Ahead Forecasts
12 New York Integration Study Evaluating 3300 MW of wind on a 33,000 MW system Time scales from seconds to days AWS Truewind provided wind data GE PSEC performing grid analysis (from AGC to day-ahead scheduling)
13 NY Study: The Challenge How to simulate the behavior of 3300 MW of wind with little site data? –Must capture spatial and temporal correlations –Met stations often not in windy areas and exhibit wrong diurnal pattern Solution: Mesoscale modeling
14 NY Study: Tasks Selected 33 potential project sites with MW capacity Used a mesoscale weather model to simulate hourly wind speed, direction, temperature for 5 continuous years Sampled 1-min and 1-sec data to synthesize sub-hourly fluctuations Created statistical model to synthesize plant forecasts – based on actual forecasts
15
16
17 Forecasting Error Distribution
18 Validation of Dynamic Behavior
19 “Extreme” Wind Events
20 “Extreme” Event System Response
21 Conclusions Wind, turbine, and wind plant variability are not the same The more spatial diversity, the less temporal variability Mesoscale modeling provides a powerful tool for analyzing scenarios of large wind penetration