Soiling investigation for PV and CSP system: experimental and ANN modelling analysis in two sites with different climate

• Published in: International journal of sustainable energy 2022-07, Vol.41 (6), p.629-645
• Main Authors: Maftah, Anas, Azouzoute, Alae, El Ydrissi, Massaab, Oufadel, Ayoub, Maaroufi, Mohamed
• Format: Article
• Language: English
• Subjects: Air temperature; ANN model; Artificial neural networks; Climate; Climatic conditions; Concentrated Solar Power; Environmental conditions; environmental parameters; Neural networks; Parameters; Photovoltaic; Photovoltaic cells; Photovoltaics; Rainfall; Reflectors; Regression analysis; Regression models; Relative humidity; soiling; Soils; Statistical analysis; Wind; Wind direction; Wind speed
• ISSN: 1478-6451; 1478-646X
• DOI: 10.1080/14786451.2021.1965605

Photovoltaic and CSP development are related to site specification. In addition to high levels of insolation, climate and environmental conditions have a great impact on the efficiency of these technologies. Soiling stands as a major parameter influencing the degradation of transmittance and reflectance for PV glass and CSP reflectors respectively, which decreases electrical performance. In this paper, the PV system and TraCS system (Tracking Cleanliness System) are used to evaluate the soiling effect in two sites with different climate conditions. Environmental variables such as air temperature, relative humidity, wind speed, wind direction, and precipitation, are used as inputs to model soiling. The relationship between different input and output variables are analysed using two linear regression models in a simple form, and with the interaction between model predictors. An in-depth analysis is performed with an artificial neural network (ANN) model, to assess its ability to resolve the complexity of the soiling problem in relation to environmental parameters, and also for comparison with the linear models. The results show that in the region with relatively low rainfall soiling is significantly high and the soiling loss reach up in the summer season -1.07%/day. However, in the region with significant rainy days, the soiling loss is in the range of -0.18%/day in the same period of the year. On the other hand, the ANN model performed significantly in estimating the soiling ratio in comparison to the linear models in terms of R 2 values and statistical indicators.