Performance evaluation of predictive models for a Grid-Connected Building-Integrated photovoltaic system (BIPV)

Building-integrated photovoltaics (BIPV) emerge as essential components in sustainable urban infrastructure, yet predictive modeling of their thermal and electrical behaviour remains underexplored. This study evaluates two widely used numerical photovoltaic models, a Simplified and a Detailed model, for their ability to predict the thermal and electrical performance of a BIPV rainscreen system under varying irradiance conditions and temporal resolutions. High-resolution (5-minute), mid-resolution (15-minute), and low-resolution (1-hour) simulations are conducted to reflect applications in research and development (R&D), demand side management (DSM), and energy prediction, respectively. Model outputs are validated against experimental data using six statistical indicators. Results show that the Simplified model performs well across temporal resolutions with a consistent overprediction particularly under low-irradiance conditions with its overall accuracy improving at lower resolutions (1-hour). Conversely, the Detailed model excels in simulations with high temporal resolution (5-min), capturing transient irradiance and temperature effects, but shows reduced accuracy when input data are temporally aggregated. These findings highlight the importance of selecting models based on application-specific requirements. The Detailed model is recommended for research and development, while the Simplified model is more suitable for building performance simulations. The study also emphasizes the need for better thermal modelling and further validation of numerical models across a variety of BIPV configurations and climatic conditions.