Bankable solar resource assessment and risk management in planning - PowerPoint PPT Presentation

Bankable solar resource assessment and risk management in planning and operation of Solar Energy Projects Marcel Suri, PhD GeoModel Solar s.r.o., Bratislava, Slovakia marcel.suri@geomodel.eu http://solargis.info http://geomodelsolar.eu Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [1] 13-15 September 2011, Jodhpur, Rajasthan, India

About GeoModel Solar Expert consultancy: • Solar resource assessment and PV yield prediction • Performance characterization • Country optimization potential European Commission • Grid integration studies PVGIS 2001-2008 SolarGIS : Real-time solar and meteo data services for: • Site selection and prefeasibility • Planning and project design • Monitoring and forecasting of solar power • Solar data infrastructure SolarGIS from 2008 http://geomodelsolar.eu http://solargis.info Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [2] 13-15 September 2011, Jodhpur, Rajasthan, India

International collaboration International Energy Agency , Solar Heating and Cooling Program: • Task 36 Solar Resource Knowledge Management • Task 46 Solar Resource Assessment and Forecasting • EU COST Action Weather Intelligence for Renewable Energies • EU project Management and Exploitation of Solar Resource Knowledge (finished) • National Renewable Energy Laboratory (NREL, US) • SUNY (US) • DLR (DE) • Fraunhofer ISE (DE) • Stellenbosch University (ZA) • University of Geneva (CH) • European Commission JRC (IT) • CENER (ES) • SUPSI ISAAC (CH) Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [3] 13-15 September 2011, Jodhpur, Rajasthan, India

Uncertainty in solar resource assessment Solar resource estimate • High quality ground measurements of solar radiation missing • Diverse results from the existing databases • Poor understanding of the potential of the modern satellite-derived data Weather interannual variability • Long and continuous record of data is needed (10+ years) • Changing weather (natural and human induced) and extreme events (e.g. volcanoes) to be considered • In the recent history • In the future Bankable data = low uncertainty, high reliability Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [4] 13-15 September 2011, Jodhpur, Rajasthan, India

Solar resource – requirements for solar projects Data available at any location Long-climate record (10 years minimum) Cleaned, validated, harmonized and without gaps High accuracy, low uncertainty (no systematic errors, good representation) High level of detail (temporal, spatial) Modern data products (time series, TMY, long-term averages) Standardized data formats Real-time data supply: • historical • monitoring All this is possible with satellite-based data, • nowcasting supported by high-quality ground measurements! • forecasting + Meteo and other geodata for energy modeling (temperature, wind, humidity) Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [5] 13-15 September 2011, Jodhpur, Rajasthan, India

Solar resource – how to obtain site-specific information Ground instruments Satellite-based solar data (interpolation/extrapolation) (solar radiation models & atmospheric data) WRDC network (~1200 archive stations) sources: NASA, EUMETSAT, Stoffel et al. 2010 sources: NREL, WRDC Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [6] 13-15 September 2011, Jodhpur, Rajasthan, India

Available solar databases - Gujarat GHI >10% more for DNI! The databases differ in many aspects: • Input data (satellite/ground) • Time coverage (period) • Applied methods/models • Time and spatial resolutions Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [7] 13-15 September 2011, Jodhpur, Rajasthan, India

Ground instruments ADVANTAGES LIMITATIONS High accuracy at the point of measurement Historical data: High frequency measurements (sec. to min.) Limited time of measurement High-quality data Limited number of sites Unknown accuracy (in historical data) THIS APPLIES ONLY IN THE CONTROLLED Different periods of measurement AND RIGORIUSLY MANAGED CONDITIONS … Operation of a ground station: Regular maintenance and calibration Data management Issues of aggregation statistics High costs for acquisition and operation Extrapolation/interpolation ignores site-specific info source: Gueymard 2010AWI Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [8] 13-15 September 2011, Jodhpur, Rajasthan, India

Uncertainty in ground observations Issues • Sensors accuracy • Installation and maintenance routines • Cleaning of the sensor • Calibration • Time shifts, shading Needed procedures • Data post-processing • Quality checking (only high-frequency data!) • Filling the gaps in the measurements • Missing data results in skewed aggregation statistics (e.g. daily and monthly sums) • High probability of systematic deviation (BIAS) and occurrence of extreme values • Uncleaned data result in unreliable values Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [9] 13-15 September 2011, Jodhpur, Rajasthan, India

Solar radiation models: satellite-derived data ADVANTAGES LIMITATIONS Available everywhere (continuous coverage) Lower instantaneous accuracy for the point estimate (when compared to high quality Spatial resolution from 3 km ground measurements) Frequency of measurements from 15 minutes Spatial and temporal consistency High calibration stability Availability ~99.5% History of up to 20 years Continuous geographical coverage (global) Data sources: EUMETSAT, ECMWF Source: SolarGIS Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [10] 13-15 September 2011, Jodhpur, Rajasthan, India

Uncertainty in satellite-derived DNI and GHI Highest uncertainty Clouds Aerosols Water vapour Terrain ±10% (up to ± 50%) DNI 0 to 100% ±3 to 4% 100% ±2 to 3% GHI 0 to 80% (up to ± 12%) ±0.5 to 1% 60 to 80% Terrain Atmosperic Optical Depth Water vapour Clouds Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [11] 13-15 September 2011, Jodhpur, Rajasthan, India

Uncertainty of Aerosol Optical Depth (AOD) Kanpur Critical for DNI AERONET MACC GEMS MACC model compared to ground measured AERONET data Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [12] 13-15 September 2011, Jodhpur, Rajasthan, India

Typical uncertainty of ground-measured vs. satellite-derived solar data GHI Thermopile pyranometer Satellite ISO Classification Secondary Standard First Class Second Class WMO Classification High Quality Good Quality Mod. Quality RMSD Bias Hourly uncertainty 3% 8% 20% 9-20% ±2-7% Daily uncertainty 2% 5% 10% 4-12% bias depends on the calibration and maintenance DNI Thermopile pyrheliometer RSR Satellite WMO Classification High quality Good quality RMSD Bias Hourly uncertainty 0.7% 1.5% 2-4% 24-60% ±4-12% Daily uncertainty 0.5% 1.0% 1.5% 15-25% Bias: • It is natural for satellite-derived data and can be reduced/removed • For ground-measured data it is very challenging and costly to keep bias close to 0 Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [13] 13-15 September 2011, Jodhpur, Rajasthan, India

Typical uncertainty of ground-measured vs. satellite-derived solar data GHI Thermopile pyranometer Satellite ISO Classification Secondary Standard First Class Second Class WMO Classification High Quality Good Quality Mod. Quality RMSD Bias Hourly uncertainty 3% 8% 20% 9-20% ±2-7% Daily uncertainty 2% 5% 10% 4-12% bias depends on the calibration and maintenance DNI Thermopile pyrheliometer RSR Satellite WMO Classification High quality Good quality RMSD Bias Hourly uncertainty 0.7% 1.5% 2-4% 24-60% ±4-12% Daily uncertainty 0.5% 1.0% 1.5% 15-25% GHI: • satellite already competitive in RMSD with good-quality sensors Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [14] 13-15 September 2011, Jodhpur, Rajasthan, India

Typical uncertainty of ground-measured vs. satellite-derived solar data GHI Thermopile pyranometer Satellite ISO Classification Secondary Standard First Class Second Class WMO Classification High Quality Good Quality Mod. Quality RMSD Bias Hourly uncertainty 3% 8% 20% 9-20% ±2-7% Daily uncertainty 2% 5% 10% 4-12% bias depends on the calibration and maintenance DNI Thermopile pyrheliometer RSR Satellite WMO Classification High quality Good quality RMSD Bias Hourly uncertainty 0.7% 1.5% 2-4% 24-35% ±4-12% Daily uncertainty 0.5% 1.0% 1.5% 15-25% DNI: • It is very challenging to keep high standard of DNI ground measurements • Satellite data can be correlated with ground measurements to obtain improved site solar statistics Workshop for Solar Energy and Smart Grid Development, Asian Development Bank, Regional Task Force [15] 13-15 September 2011, Jodhpur, Rajasthan, India