IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 7, NO. 6, NOVEMBER 2017 1755 Long-Term Soiling Analysis for Three Photovoltaic Technologies in Santiago Region Pierre Besson , Constanza Mu ˜ noz, Gonzalo Ram´ırez-Sagner, Marcelo Salgado, Rodrigo Escobar, and Werner Platzer Abstract—The effect of soiling on PV systems impacts negatively the energy production. This phenomenon is highly dependant on the environmental context and conditions of operation. Indeed dirt, dust, and other air contaminants are site-specific and their accu- mulation on PV modules depends on the installation configura- tion. This study, conducted in Santiago de Chile over a period of two and half years, focuses on analyzing power production and soiling losses of three photovoltaic technologies, monocrystalline, polycrystalline, and thin-film Si. We present the method used for analyzing time series of production data, and demonstrate the sea- sonality of soiling rates in Santiago city. Indeed, soiling rate values for winter season, where contamination is high, are three times higher than values for summer season, where contamination is low. A yearly trend of soiling is determined and used to study the link between economical parameters and cleaning pattern applied. The results allow us to define an optimal cleaning period depending on the period of the year. Index Terms—Cleaning, data analysis, photovoltaic (PV) sys- tems, performance analysis, soil. I. INTRODUCTION T HE accumulation of dust, and other contaminants present in the air, on the surface of solar panels can cause a sig- nificant decrease of the system’s energy production [1], [2]. De- pending on the local environment and the cleaning frequency, annual losses of soiling vary from values below 3% to values as high as 30% [3]. Moreover, different panel properties, such as spectral response, angular response, or electrical architecture, may lead to different behavior when the module is facing soil- ing. Therefore, there is a growing interest in better understand- ing and quantifying this soiling process, in order to predict more accurately the energy yield of a power plant and plan in a cost- effective way cleaning events. Different field studies demon- strate this interest for a better characterization of soiling losses in Manuscript received May 5, 2017; revised August 2, 2017 and September 6, 2017; accepted September 10, 2017. Date of publication October 4, 2017; date of current version October 19, 2017. This work was supported in part by the FONDEF program from CONICYT through Grant D08i1097, in part by the CORFO institution under Grant 13CEI2-21803, and in part by DICTUC under Grant 13150011. (Corresponding author: Pierre Besson.) P. Besson, C. Mu˜ noz, G. Ram´ırez-Sagner, M. Salgado, and W. Platzer are with the Centre of Solar Energy Technologies, Fraunhofer Chile Research, San- tiago 7550296, Chile (e-mail: pierre.besson@fraunhofer.cl; constanza.munoz@ fraunhofer.cl; gonzalo.ramirez@fraunhofer.cl; marcelo.salgadobravo@gmail. com; werner.platzer@fraunhofer.cl). R. Escobar is with the Pontificia Universidad Catlica de Chile, Santiago 7820436, Chile (e-mail: rescobar@ing.puc.cl). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JPHOTOV.2017.2751752 harsh climate [4], [5]. Most of the research focuses on studying one technology at a time in an area suitable for large deployment (desertic conditions, for example), but there also exist studies on the impact of contaminated environment on PV production. In [6], Kalogirou et al. realized studies on modules of dif- ferent technologies installed on a roof in an urban coastal area of Cyprus, where they observed drop of more than 40% in ex- treme cases. Kaldellis et al. [5] worked on the problem of air pollution in Athens, and have artificially analyzed the impact of different substances on module performance. They showed that carbonaceous fly-ash particles, which are mainly produced by oil and coal combustion, are impacting to a lesser extent the energy generation than red soil. More recently, Urrejola et al. [7] have performed a study on the performance ratio (PR) of modules in Santiago de Chile, a city with levels of contamination similar to Mexico city [8]. The results show a seasonal trend in soiling, with higher soiling rates in winter months (June to October), period characterized by high level of contamination, than summer months (December to February), months with low level of contamination. Knowing such tendencies allows us to optimize cleaning schedules and a better estimate of total annual soiling losses [9]. The approach proposed in this paper consists of 1) establish- ing an adapted method for extracting reliable soiling rates from inverter production data; 2) evaluating the results and quantify- ing the seasonality of the different soiling rates observed; and 3) apply the results obtained in order to propose an optimized cleaning schedule, and quantify economically the relevance of such procedure. II. EXPERIMENTAL PROTOCOL A. Experimental Setup The panels under test used for this study correspond to three PV systems of different technologies, installed on a three floor building’s rooftop of the Pontificial Catholic University of Chile. The orientation and inclination of the different systems is simi- lar: the tilt angle is 32 ◦ and the azimuth is 350 ◦ , corresponding to a North orientation shifted 10 ◦ toward the West. The building location corresponds to a highly densified urban environment, and its GPS coordinates are: latitude 33.50 ◦ S, longitude 70.61 ◦ W, and altitude 577 m. In order to have a representative study of soiling impact on photovoltaic (PV) modules, we work on market available PV technologies based on monocrystalline silicon (Mono c-Si), 2156-3381 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.