Advancing PV temperature modeling: comparative evaluation and a novel empirical model

Accurate estimation of PV module temperature is essential for reliable prediction of solar electrical generation and system-level performance assessment. However, commonly used empirical steady-state models often oversimplify thermal processes by assuming that heat transfer depends solely on wind speed, while radiative losses are largely neglected. This simplification can introduce temperature-prediction errors exceeding 10 °C compared with measured values. This study aims to develop a simple yet accurate steady-state empirical PV temperature model that accounts for radiative effects without increasing model complexity or instrumentation requirements. This study proposes a novel empirical PV temperature model that goes beyond existing formulations by incorporating an irradiance-dependent correction to the effective heat transfer coefficient, enabling an implicit representation of radiative heat dissipation. Rather than explicitly modeling radiative processes, the proposed approach preserves the simplicity of conventional empirical models while improving physical realism and prediction accuracy. The proposed formulation is benchmarked against four established models: Ross, NOCT, Sandia, and the Faiman radiation model, using long-term field data from multiple climates and PV technologies. The results show that the proposed model achieves the lowest annual temperature prediction error, in terms of RMSE, across all sites and PV technologies considered. The improvement is particularly evident under high-irradiance conditions, where irradiance-driven thermal dynamics dominate module heating. For example, under strong solar irradiance, the proposed model exhibits an nMBE as low as 4.3%, whereas the benchmark models show substantially larger biases (up to approximately 17.5% in magnitude). These results also indicate that the advantages of the proposed model become more pronounced in sunnier climates. The enhanced temperature accuracy also results in reduced annual PV power prediction errors, highlighting the importance of refined thermal modeling for system-level performance assessment. Overall, the proposed empirical model offers a robust, accurate, and instrumentation-free approach for estimating PV temperature, providing practical advantages for large-scale performance modeling and forecasting applications.