Data-Driven Soiling Estimation and Optimized Cleaning Strategies for Industrial Rooftop PV Systems

Abstract

The accumulation of dust and dirt on solar photovoltaic (PV) panels, known as soiling, reduces energy generation and conversion efficiency of a PV plant. Therefore, regular cleaning is essential to maintain optimal plant performance and economic viability. Fixed-interval cleaning schedules become uneconomical during periods such as low-insolation, rainy, or cloudy events. This study proposes a data-driven method to estimate the soiling ratio (SR) for a 504-kWp rooftop PV plant in India using power, temperature, and irradiance data. A PV panel temperature estimation model is employed, based on ambient temperature and solar irradiance, which simplifies the process by eliminating the need for direct temperature measurements. The analysis reveals that regular cleaning is essential despite rainfall, with energy losses due to soiling ranging from 32% to 47% across inverters, with soiling rates of 4.6–5.5% per day. A dynamic cleaning schedule, considering weather and soiling conditions, was developed to reduce these losses. Economic evaluation demonstrated that manual cleaning following the proposed dynamic schedule is cost effective, with profit margins of 48–77%, comparing energy gain and cleaning cost. Compared with fixed-interval cleaning, the proposed method maintained the same average SR but yielded 25–49% higher profitability across inverters.