Assessment of soiling prediction strategies to improve the accuracy of simulation models for parabolic trough solar thermal power plants

Soiling significantly degrades the optical performance of concentrating solar power (CSP) plants, yet it is often oversimplified in utility-scale simulations. This work presents a computationally efficient modelling framework for parabolic trough plants that integrates a novel soiling prediction method requiring only satellite derived aerosol optical depth and precipitation data. Validated against a year of operational data from a 50 MW commercial plant, the model reduces the prediction error in electrical power output from 13.41% (constant cleanliness assumption) to 13.09%, corresponding to deviations in annual yield on the order of 500 MWh. Notably, the proposed modelling approach provides improved accuracy compared to simulations based on reflectance measurements in the plant (13.29%), highlighting that satellite-driven models can better capture the spatial and temporal representativeness of an entire solar field than sparse physical sensors. This modelling framework bridges the gap between detailed physical models requiring extensive data and the common industry practice of assuming a constant cleanliness factor. This framework can be used in future work to optimise cleaning schedules, directly enhancing the operational and economic performance of CSP projects.