Understanding the Variation in Long-Term Solar Resource Estimates

Transcription

1 Understanding the Variation in Long-Term Solar Resource Estimates Which Data Set Represents Your Project Site? Marie Schnitzer Senior Director, Solar and Investor Services

2 Outline Purpose What is a P50? Data Sources Methodology Results/Impact

3 Purpose Industry tends to default to TMY2 or TMY3 data sets Need for understanding variation between various data sets What to consider? Availability Period of record Accuracy Others Which data source is selected to represent a project?

4 What is a P50? Expected or most likely estimate Site-specific Derived for long-term solar resource AND long-term energy Baseline for higher confidence resource or energy estimates

5 Data Sources Selection criteria High quality Long period of record Multiple solar components

6 Data Sources Compared to SUNY Satellite Model Meteonorm Closest NSRDB TMY3 Data Set Closest Class I NSRDB TMY3 Data Set NASA SSE Other Satellite Model Closest TMY2 Data Set Nearby USCRN Station Modeled data disclaimer

7 Methodology True long-term solar resource P50 estimate from each data source Calculation of: Average relative bias Standard deviation of biases Maximum absolute bias Most and least accurate Comparison of P50 estimates

8 Results 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% -0.5% -1.0% -1.5% -2.0% GHI Average annual bias across 9 sites Data Source DNI DHI Other Satellite 0.5% -4.0% SUNY -0.5% -0.5% Meteonorm 2.0% 2.8% Nearest TMY3-3.7% 1.0% Nearest C1 TMY3-2.3% 3.6% TMY2-2.5% 7.5% NASA SSE 5.9% -3.9% Nearby USCRN NA NA

9 Results Average monthly bias across 9 sites

10 Results Standard deviation of bias across 9 sites Data Source GHI DNI DHI Other Satellite 4.0% 6.7% 4.8% SUNY 3.5% 5.6% 4.0% Meteonorm 5.7% 9.8% 4.8% Nearest TMY3 4.1% 7.8% 6.5% Nearest C1 TMY3 1.4% 4.7% 6.2% TMY2 2.5% 4.7% 8.3% NASA SSE 2.7% 8.0% 6.5% Nearby USCRN 5.1% N/A N/A Modeled data disclaimer Maximum absolute bias across 9 sites Data Source GHI DNI DHI Other Satellite 9.1% 11.1% 14.1% SUNY 9.2% 12.5% 7.1% Meteonorm 10.8% 18.4% 10.7% Nearest TMY3 6.9% 17.0% 14.6% Nearest C1 TMY3 2.7% 12.6% 11.9% TMY2 6.0% 11.3% 21.3% NASA SSE 5.0% 21.4% 10.6% Nearby USCRN 8.6% N/A N/A

11 Impact on P50 Estimates Maximum and minimum difference compared to true long-term Site Min GHI Max GHI Min DNI Max DNI Min DHI Max DHI Boulder -4.2% 2.1% -4.2% 5.2% -6.6% 5.9% Bondville -6.9% 4.1% -17.0% 7.0% -9.8% 5.2% Desert Rock -4.4% 1.4% -6.7% 6.1% -14.1% 21.3% Penn State -1.8% 10.8% -12.6% 18.4% -5.9% 11.9% Salt Lake 1.6% 8.6% -3.7% 21.4% -8.7% 10.4% Goodwin -9.2% 6.0% -13.1% 3.2% -1.5% 16.6% Fort Peck -5.1% 3.3% -7.3% 5.7% -6.4% 10.6% Sterling -8.4% 5.2% -7.4% 9.9% -5.4% 14.9% Sioux Falls -6.0% 3.3% -10.4% 7.7% -9.1% 2.9% Average -4.9% 5.0% -9.2% 9.4% -7.5% 11.1% Direct Impact on Energy Data source most likely to over and under predict Solar Resource Component Extreme P50 (Low) Extreme P50 (High) GHI Nearest TMY3 TMY2 DNI Nearest TMY3 NASA SSE DHI NASA SSE N/A

12 Conclusions Many data sources to choose from P50 estimates vary on an annual and monthly basis GHI differences ranged to greater than 10% in some cases Greater for DNI and DHI Impact on Energy Level of risk in using modeled data to represent a project site Higher confidence desired/needed On-site monitoring

13 Thank You awstruepower.com Marie Schnitzer Christopher Thuman Peter Johnson