1710 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 7, NO. 6, NOVEMBER 2017 Detecting Photovoltaic Module Failures in the Field During Daytime With Ultraviolet Fluorescence Module Inspection Arnaud Morlier , Michael Siebert, Iris Kunze, Gerhard Mathiak, and Marc K¨ ontges Abstract—We present the potential of ultraviolet fluorescence imaging for the detection of function and safety failures of pho- tovoltaic modules in the field. We apply this method to detect hotspots, mismatched cells, power-loss-inducing cracks, and we show how we use it to evaluate the crack history of a module. We present a device able to acquire fluorescence images in the field in the daytime without disconnecting the modules and with a through- put of more than 200 modules per hour for a single operator. Fur- thermore, we show how the technique allows for the discrimination of specific damages caused to the photovoltaic modules by sudden events such as hailstorms. We demonstrate that this method pos- sesses an informative potential comparable to both thermography and electroluminescence together with less practical limitations. Index Terms—Fluorescence, nondestructive testing, solar module reliability. I. INTRODUCTION D ETECTING flawed photovoltaic (PV) modules which cause power losses in a PV array is a common issue for a PV power plant owner. A single module defect can severely affect the output of a whole module string or may cause a seri- ous safety issue such as fire. On large field power plants, various methods are employed to find potentially defective modules. On- line monitoring of PV modules or I–V-curve testing of strings allows detecting the occurrence of defects causing a power loss in operation. Nevertheless, the localization of a single defect module in a string requires a more detailed investigation. The nowadays common methods at the disposal of the reli- ability expert for the inspection of PV are thermography and electroluminescence (EL). EL has been first used by Fuyuki et al. [1] and has become widespread for laboratory quality con- trols [2]. EL can also be applied in the field to capture whole strings [3]. The possibility to embark an EL camera on un- manned aerial vehicles for higher measurement throughputs is Manuscript received June 9, 2017; revised August 4, 2017 and September 12, 2017; accepted September 18, 2017. Date of publication October 10, 2017; date of current version October 19, 2017. This work was supported by the German Federal Ministry for Economic Affairs and Energy under contract number 0325735D. (Corresponding author: Arnaud Morlier.) A. Morlier, M. Siebert, I. Kunze, and M. K¨ ontges are with the Institute for Solar Energy Research Hamelin, Emmerthal 31860, Germany (e-mail: morlier@isfh.de; siebert@isfh.de; kunze@isfh.de; koentges@isfh.de). G. Mathiak is with T ¨ UV Rheinland Energy GMBH, Cologne 51105, Germany (e-mail: gerhard.mathiak@de.tuv.com). 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.2017.2756452 also promising [4]. However, for module inspection with this camera-based technique, all daylight noise needs to be omitted. Although advanced EL techniques relying on gallium detector cameras allow nowadays also for daytime measurements [5], EL can almost exclusively be performed during the night or at twilight. Furthermore, EL imaging requires disconnecting the modules from the plant circuit and connecting them to a cum- bersome power source. The thermography technique is simpler to implement and is useful to reveal safety issues such as hotspots [6], [7]. Never- theless, this method requires the modules to be in operation and is dependent on the solar irradiation conditions. For glass/backsheet modules, it is possible to rely on the flu- orescence of ethylene vinyl acetate (EVA) under UV irradiation to detect the presence of cracks in silicon solar cells [8]. This method has been successfully used to detect cracked cells in modules in the field without the need for electrical contacting or demounting of the modules. For measuring, a PV module is illuminated by a UV light and the resulting fluorescence signal is measured by a charge-coupled device (CCD) camera. Flu- orophores present in the module encapsulation formed during weathering are excited and emit light in the visible range, mainly between 400 and 600 nm [9]. During field exposure and weath- ering, the fluorophores are degraded (i.e., photobleached) by oxygen, which diffuses through the polymer backsheets and the encapsulating polymer. The cells protect the polymer comprised between the cells and the front glass from oxygen diffusion. Nev- ertheless, cracks in the cells allow for the diffusion of oxygen to this intermediary polymer layer resulting in a local extinction of the fluorescence emission on the crack and its surrounding [10]. Schlothauer et al. have studied the spatial resolution of the fluorescence and also shown that the luminescence of EVA can be correlated with other material characteristics such as oxy- gen diffusion induced postcrosslinking [11], [12]. In this work, we show which information can be extracted from fluorescence images additionally to the already reported crack detection ap- plication and its potential for reliability assessment. For this, we compare fluorescence images with EL images or tempera- ture mappings. We acquire these images with an experimental device called fluorescence & electroluminescence outdoor in- spection system that we developed at ISFH. This device allows acquiring both the fluorescence and the EL images of an indi- vidual module by means of a consumer digital camera under daylight conditions. 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.