Incorporation of dynamic soiling loss into the physical model chain of photovoltaic (PV) systems

Understanding the long-term operational trends and dynamic characteristics of photovoltaic (PV) systems is crucial for maximizing system efficiency, enhancing grid stability, and reducing maintenance costs. Traditional methods typically calculate PV system losses as fixed values, which are suitable for stable environments or simple scenarios in the initial design phase. However, this static method fails to capture the impacts during long-term operations, particularly under changing environmental conditions. We propose a novel method for quantifying dynamic soiling loss to address this limitation. This method integrates the effects of dust accumulation into the existing physical model chain, significantly improving accuracy in long-term assessments. This study focuses on the dynamic impacts of dust accumulation over time, providing a comprehensive analysis of dust deposition, regular manual cleaning, and natural rainfall cleaning. We establish a dynamic soiling loss model to quantify the performance degradation of PV systems accurately. Public datasets from Australia, China, and Belgium are used to evaluate the model’s applicability and limitations in diverse environments and geographical areas and for different PV systems. A comparison of the proposed model’s results with operational data demonstrates its superior performance compared to the clear-sky and fixed-decay models. The proposed model has a lower mean bias error (MBE) for Australian data from 2017 to 2020, with an average of 0.0304 kW, and a lower normalized root mean square error (nRMSE) for data from the three countries, with a minimum value of 0.0916. We discuss the potential applications of the proposed method for PV power evaluation and data generation.