6-3_McIntosh-PVPMC-2023-1

1 Keith McIntosh, Malcolm Abot and Ben Sudbury Differences betwen advanced and conventional models in bifacial yield simulations PVPMC Workshop 10-May-2023 2 Objective Quantify the diference betwen models for 8 physical mechanisms −Mostly tested with SunSolve-Yield −And PVSystfor its VF model Thre system configurations −SATs −Fixed-tilt −Waves Just one location −Southwest Utah 3 Results preview MBE 0 yield higher for conventional model than advanced model Models are convoluted −MBEs of individual models do not sum to BE of al models combined. MBE 0 does not imply two models are equivalent. Maybe −morning discrepancy compensated by noon discrepancy, −summer compensated by winter, −one poor sub-model compensated by another poor sub-model. Results for typical 1P SAT located at Cove Mountain, UT. 4 Results preview Average P mod 315 W Results for typical 1P SAT located at Cove Mountain, UT. 5 Three early coments This is not an investigation into the accuracy of the models. Results specific to chosen examples. −Model comparisons wil difer for other location, weather, system configurations. −Consider these results as a general guide, with more emphasis on CRMSE than MBE. By themselves, these results don’t promote any model over another. The value of a model depends on many things −Acuracy, precision, uncertainty −Ease of implementation determining inputs −Aceptance by industry −Modeling objective e.g., annual yield, morning power, structural shading, etc. 6 Simulation details 7 Simulation details site location Image htps/ww.kimley-horn.com/project/cove-mountain-solar/ Coordinates for Phase I of Cove Mountain from https/ww.gem.wiki/Cove_Mountain_Solar Site of Cove Mountain Solar Plant, Utah 37.62 N 113.62 W 1570 m Image Google Maps 8 Simulation details weather I 1.96 MWh/m 2 f D 26.5 T av 14.7 o C w av 3.5 m/s Good solar resource Low cloud Cold winter, warm sumer Mostly light winds 9 Simulation details atmosphere 0 0.5 1 1.5 2 2.5 3 3.5 P r ec i p w a t er v a p o u r c m Giovani 20-yr av Solcast TMY Default for AM1.5g 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 A e r o so l o p ti c a l d e p th a t 5 0 0 n m Giovani 18-yr av Default for AM1.5g 0 50 100 150 200 250 300 350 400 O zo n e D o b so n s Default for AM1.5g Giovani 18-yr av 800 850 900 950 1000 A tm o sp h e r i c p r e ssu r e m b Default for AM1.5g Giovani 12-yr av Solcast TMY Relatively dry Dusty Moderate ozone Low presure due to altitude 1570 m We use −Solcastfor PWV P −Giovani for O 3 and turbidity Blue data is long-term daily average from NASA’s satelites and analysis of 1 o bounding box. htps/giovanni.gsfc.nasa.gov/giovanni. Gren data is daily average of Solcast, recently acquired by DNV. https/solcast.com/ PWV Turbidty Presure Ozone 10 Simulation details SAT 1P one in portrait Six modules per bay Posts betwen bays Circular torque tube Smal clamps rails Max tilt 55 o , backtracking Example results of ray tracing for 9 am, 10-May, light cloud. Front Rear 11 Simulation details fixed-tilt 2P two in portrait. Six modules per bay Posts betwen bays Rafters Purlins Tilt 25 o Example results of ray tracing for 10 am, 10-May, light cloud. Front Rear 12 Simulation details waves 1P one in portrait. Ten modules per wave. Concrete slabs below modules. Rafters Purlins Tilt 10 o . Example results of ray tracing for 9 am, 10-May, light cloud Front Rear 13 Simulation details module 37.62 N 113.62 W P MP 550 W Bifi 70 V OC 49.80 V I SC 13.99 A V MP 41.95 V I MP 13.12 A FF0.790 Eff 21.31 LONGI LR5-72HBD 550M 144 half-cut cels 14 Simulation details general Infinitely large system no edge efects from system perimeter. DC module output average of al modules in a bay. 1 hourly time steps. 15 Models examined 16 1. Spectral albedo Conventional Advanced Constant 33.7 Wavelength-dependent Yellow-brown soil NASA 17 2. Electrical mismatch Conventional Advanced Constant −Here, we use the anual weighted average mismatch los f M as determined by SunSolve front rear combined Calculated at al time steps 1. Solve J L in each cell 2. Solve equivalent-circuit of module f M SAT 0.5 Fixed 0.5 Waves 0.7 Fixed f M omits row-to-row shading of direct light since that is acounted for in PVSyst. 18 3. Solar position Conventional Advanced PVSyst[1] −Simple equations. −Omits refraction. Reda–Andreas 204 [2] −Mases of tables and equations. −Accounts for refraction. −Zenith and azimuth to within 0.0003 o betwen 200 BCE and 600 CE. [1] https/ww.pvsyst.com/help/solar_geometry.htm [2] Reda and Andreas, “Solar position algorithm for solar radiation aplications,” Solar Energy765, 57–589, 2004. 19 3. Solar position Conventional Advanced PVSyst[1] −Simple equations. −Omits refraction. Reda–Andreas 204 [2] −Mases of tables and equations. −Accounts for refraction. −Zenith and azimuth to within 0.0003 o betwen 200 BCE and 600 CE. –0.3 o q s 0.3 o –0.6 o f s 0.5 o 20 4. Difuse sky distribution Conventional Alternative implementation PVSystPerez 190 [7] PVL Perez 190 [7] Perez et al., “Modelingdaylight availability and iradiance components from direct and global iradiance,” Solar energy44 5,.271-289, 1990. [8] Hay and Davies, “Calculation of solar radiation incident on an included surface,“ 1stCanadian Solar Radiation Data Workshop, Ontario, 16, 1980. NB Change in yield when using Hay–Davies [8] rather than Perez in our example Conventional PVSyst Alternative SunSolve-Yield SAT -1.350.15 Fixed -1.500.17 Waves -0.650.80 21 Models, like simple Perez 190, approximate the difuse light with thre sources isotropic circumsolar horizon 4. Difuse sky distribution isotropic horizon circumsolar 22 4. Difuse sky distribution Adaption for infinite field acounts for shading from odules. PVSyst’simplementation of Perez 190 isotropic −partial shading circumsolar −posible shading horizon −same shading as isotropic isotropic horizon mod u l e circumsolar 23 4. Difuse sky distribution Adaption for infinite field acounts for shading from odules. PVL’s implementation of Perez 190 isotropic −partial shading circumsolar −posible shading horizon −completely shaded circumsolar isotropic horizon mod u l e 24 5. Module optics Conventional Advanced Highly absorbing material 100 Transmision calculated from IAM 0 Si Glas Glas EVA SiN x Ag Al Al 2 O 3 SiN x ARC 100 96 95 22 3.5 3.8 3.3 3.0 Ex am p l e r a y Ex am p l e r a y 95 25 5. Module optics Conventional Advanced ‘Simple’ like PV Syst, SAM, etc. −No reflection −IAM from lok-up table −l-independent −No cel spacing −No frames −Spatialy uniform −J L responds linearly to absorption Ray tracing into the module −Reflection −Calcs with Fresnel thin-film optics −l-dependent −Cel layout −Frames −Fingers, ribons, backsheet, pyramids, etc. −J L responds linearly to absorption 26 5. Module optics Conventional –IAM PAN files sometimes contain an unrealistic IAM. Sometimes it’s even “certified”. PVSystallows a calculated IAM instead of PAN IAM For conventional, we use calculated IAM, “Fresnel, AR coating” “Certified” gives 1.45 more yield for SATs 27 6. Thermal model Conventional Advanced Faiman[3] PVL PVSC 202 [4] distinguishes −Radiative losses −Transient efects −Tilt dependence Inputs fit to experiment [3] D. Faiman, “Asesing the outdor operating temperature of photovoltaic modules,” Progres in Photovoltaics, 16 4, 307-315, 2008. [4] K.R. McIntosh et al., “The influence of wind and module tilt on the operating temperature of single-axis trackers,” 49th IEE PVSC, 103-1036, 2022. MBE 0 C RMSE 1.4 C MBE 3.2 C RMSE 4.5 C FTC SAT FTC SAT U C U V SAT 251.2 Fixed251.2 Waves 270 28 7. Solar spectra Conventional Advanced AM1.5g Calculated at al time steps −SPECTRL2 for clear skies [5] −Ernst modification for cloudy skies [6] Affected by −Air mass i.e., solar location −Precipitable water vapour −Turbidity −Ozone −Air presure −Far-field albedo [5] Bird and Riordon, “Simple solar spectral model for direct and diffuse iradiance,” Journal of Climate and Aplied Metrology, 25, 87–97, 1986. [6] M. Ernst, et al. “SUNCALCULATOR A program to calculate the angular and spectral” Solar Energy Materials and Solar Cels, 157913–922, 2016. 29 7. Solar spectra Air mas PWV Turbidity 30 8. System optics Conventional Advanced View-factors bifacial los factors −Structural shading, f S −Transmision, f T Ray tracing