Modeling and analysis of dust and temperature effects on photovoltaic systems' performance and optimal cleaning frequency: Jordan case study

This paper provides a regionally focused review of work conducted in the Middle East and North Africa (MENA) region related to the effects of dust accumulation and ambient temperature on PV performance. It proposes models to simulate these effects, and suggests a financial methodology to determine cleaning frequency for a case study in Jordan. Two models have been developed; the first utilizing Multivariate Linear Regression (MLR), and the second utilizing Artificial Neural Network (ANN), to estimate PV system conversion efficiency based on experimental data of exposure time to natural dust and ambient temperature. The methodology of building the two models is demonstrated and a comparison between them is made to discuss their validity and accuracy. In addition, estimating the losses due to dust accumulation and, consequently, optimizing cleaning frequency are presented. It is found that both models can predict conversion efficiency closely, with R2 values close to 90%. The two models are employed in calculating losses (represented in losses in system efficiency, η, and the monetary value of these losses) due to dust accumulation only; thus, finding an optimal cleaning frequency of the systems. During the 192 days spanning the duration of the study, the average efficiency reductions due to dust are 0.768%/day and 0.607%/day using MLR and ANN models, respectively. Consequently, energy losses for the duration of the study are 10.282 kWh/m2 and 8.140 kWh/m2, and economic losses are 3.76 US$/m2 and 2.98 US$/m2 using MLR and ANN models, respectively. The optimal cleaning frequency is calculated to be 12-15 days depending on the model and length of exposure time adopted in analysis. This is the first reported optimal cleaning frequency in Jordan, and agrees with most recommendations of other already published works by other researchers and techniques in the MENA region.