Modeling and simulation of the soiling dynamics of frequently cleaned reflectors in CSP plants

• Autor(a) principal: Radia, AIT EL CADI
• Data de Publicação: 2018
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10174/31755; https://doi.org/S. Bouaddi, A. Fernández-García, A. Ihlal, R. Ait El Cadi, L. Álvarez-Rodrigo, Modeling and simulation of the soiling dynamics of frequently cleaned reflectors in CSP plants, Sol. Energy. 166 (2018) 422–431. https://doi.org/10.1016/J.SOLENER.2018.03.070.; https://doi.org/10.1016/j.solener.2018.03.070
• topic: CSP; Silvered-glass mirror; Reflectance; SoilingMarkov switching; TRNSYS

The task of cleaning reflectors is a widely adopted strategy to handle the soiling in concentrated solar power (CSP) plants. The dynamics of the soiling of frequently cleaned mirrors is dependent on many factors such as the cleaning method, its frequency and the seasonal soiling rate. This study proposes a new approach to model the soiling of regularly cleaned reflectors instead of the inaccurate methods undermining the CSP yield by using fixed reflectance assumptions. Markov switching (MS) regime, which is a non linear time series approach where parameters are allowed to switch between the regimes, are applied on the reflectance data of second surface silvered-glass mirrors, exposed at the Plataforma Solar de Almería (PSA) in Spain during two years and cleaned biweekly using high pressure demineralized water. The nonlinearities typically exhibited by such reflectance time series are successfully modeled by accounting for two regimes, a soiled and a clean regime, and incorporating various parameters (rain, washing cycle, and lagged autoregressive terms) in the suggested MS models. Based on the information criteria and the diagnostic of the models residuals, the best model has a switching normalized reflectance mean of 0.944 during the clean regime versus 0.688 during the soiled regime, in addition to one fixed lag autoregressive term. In order to evaluate the adequacy of the proposed model versus the traditional approach, which uses fixed reflectance as input for estimating CSP plants yield, four reflectance scenarios were studied by simulating the output of a 30 MWe plant using TRNSYS© (TRaNsientSYstem Simulation) program. The first and the second scenarios used the time series of the measured and fitted model reflectance data, while the third and the forth scenarios used fixed inputs (a maximum and a yearly average reflectance). The comparison of the simulation results showed that adopting the innovative proposed concept of switching regimes results in very good performance, especially in the soiled regime during which the simplistic reflectance considerations, which ignore the soiling dynamics and the applied washing cycle, undermine the generated power.