Feature review of photovoltaic modeling software utilizing blind performance assessment

While confidence in photovoltaic (PV) modeling software has always been essential, the rapid pace of new PV plant developments makes accuracy and credibility more critical than ever. Independent assessments, particularly through blind modeling comparisons, are therefore necessary to ensure unbiased benchmarking across PV modeling software. Previous studies have been limited by a narrow range of models compared, anonymized results, or system size. This study presents results from the first-ever onymous blind modeling comparison, evaluated using both lab- and utility-scale fixed-tilt, monofacial, south-facing systems at sub-hourly time intervals. Seven commercially used PV software tools were compared: 3E SynaptiQ, PlantPredict, PVsyst, RatedPower, SAM, SolarFarmer, and Solargis Evaluate. Predictions were submitted directly by software representatives, providing unique insights into each software’s implementation and resulting prediction behavior. Notable features, including plane-of-array (POA) transposition model, module temperature model, shading model, and performance model were analyzed and compared. Four summary tables compile these features of the software, serving as a resource to help users understand the methodological differences and select the most suitable software for their applications. The software tools show deviations from mean error in annual yield up to 2.5 % in the lab-scale system, increasing to 6.0 % for the utility-scale system. These differences arise from a combination of user decisions and the inherent behavior of the software, indicating the need for continuous and rigorous validation of modeling methods using these software tools against complex, real-world systems. • Standardization and transparency of the Perez model implementation are needed across photovoltaic modeling software. • The software model shading differently varying in assumed terrain, stringing configuration, and irradiance decomposition. • Results from the blind comparison demonstrate a direct correlation between annual energy NBE and the value of assumed derate. • The software show agreement for lab-scale systems, but diverge in performance as system complexity grows.