Field measurement uncertainty is > 8 % Summary of :

Two testing methods for photovoltaic (PV) plants, an infrared (IR)-measurement aerial system and a monitoring system on module level, are compared with respect to their capabilities for identifying irregularities of PV panels in a PV plant. For the first method, a hypothesis is tested that infrared temperature and module power can be correlated with an empirical linear correlation using the raw measurement data of the second method. A workaround is proposed how to quantify power losses for unknown PV plants showing at the same time how power losses depend on ambient conditions. Also, the data is used to improve results from the IR-measurements. The second method is explained to compare the two methods. It is shown that each method offers different advantages, helping to properly assess and schedule O&M measures. Janine Teubner et al. / Energy Procedia 124 (2017) 560–566 561 ScienceDirect฀ ScienceDirect฀ as on the safety of all modules in the same string. Drone-mounted infrared (IR) measurement systems (like aIR-PVcheck) present already a well-known, fast and easy method to detect modules with irregular temperature patterns, a reliable indicator of defects (see [1], [2] for further explanations). But the nature and severity of the so-found suspicious sites are not perfectly clear up to now, although a lot of research has already been conducted in that direction [3]. If it was possible to quantify all modules’ power losses, maintenance of solar power plants could be optimized even further. The module-resolved SunSniffer® measurement system is a new promising method which continuously measures the voltage and junction box temperature of every single module. This study compares the two testing methods with respect to their capabilities for identifying irregularities of PV panels in a PV plant and combines their relevant findings. For the SunSniffer system, the measurement data is validated by comparison to whether or not thermal irregularities are found. For aIR-PV-check, an empirical approach is followed for correlating the module temperatures measured via IR to their respective electrical power generation losses. According to the law of energy conservation, the mean module temperature should be the higher the lower the power output is. 2. Method and experimental procedure 2.1. Explanation of the SunSniffer system The SunSniffer system consists of the measuring units attached to each module which measure each module’s voltage and junction box temperature with a high temporal resolution as well as the evaluation engine and an online portal which makes the measurements accessible [4], [5]. The main aim of the SunSniffer engine is to define the health status of modules thus identifying weak modules and defects like PID or LID while distinguishing these from shadowing in order to be able to eventually suggest module replacements. As SunSniffer directly measures electrical data, it is known which defects really are power-relevant. Thus, the measurement data act as a reference for the aIRPV-check data. Analysis of the data quality shows that 500 out of 510 modules yielded usable data during the measurement period. Generally, modules may have a different instantaneous power reduction depending on the ambient conditions as can be seen in Fig. 1 which shows the continuously measured electrical data from SunSniffer: The module voltage changes depending on the ambient conditions. Thus, also the power loss changes. Fig. 1. The measured module voltages of a power-reduced module and its reference as well as the irradiance on 04.08.2016 2.2. Processing data for correlation of module power and temperature with aIR-PV-check For assessing module powers by their temperatures by aIR-PV-check, the total energy balance of a PV module is evaluated which allows calculating the module power from the mean module temperature and the ambient conditions as well as their distribution. But especially the wind speed distribution governing the heat losses is 562 Janine Teubner et al. / Energy Procedia 124 (2017) 560–566 complex to calculate (see [6]). Thus, an empirical correlation for a given set of environmental parameters may prove easier in practice. Additionally, in order to minimize IR measurement uncertainties, the module temperature is preprocessed by calculating the module temperature difference using the mean module temperatures of a powerreduced module T�D and a suitable non-defective reference module of the same string T�ND using aIR-PV-check: �� � ��D � ��ND (1) The index D indicates a defective, ND a non-defective module. The mean module temperatures are the IR temperatures of one module averaged over the whole module area. The module power varies depending on the severity of the defect. Hence, it is normalized by using a reference module of the same string; so that the relative module power Prel can be obtained by dividing the corresponding module voltages (see Eq. (2)). In order to attain reliable data, a roof-mounted PV plant equipped with the module-resolved SunSniffer® system was inspected by aIR-PV-check. The plant is located in Cadolzburg, Middle Franconia (plant C), with a total power of 100 kWp and 510 modules. It was measured on 04.08.2016 from 12:00-12:11 at an irradiation of 950 W/m and an ambient temperature of 26°C. As shown in Fig. 1, the module voltages change and since aIR-PV-check is an instantaneous measurement method, the module voltages used are averaged over the IR measurement period: During the aIR-PV-check period, relatively constant ambient conditions were ensured. The corresponding measurement data taken from the SunSniffer system was averaged over the same period to make it comparable to aIR-PV-check, described with ���� � ��,�� � ��,�� with ��,�� being the time beginning and ��,�� being the end of the IR measurement: �rel����IR�, �U��IR�, ���IR�� � �D �ND��IR � �D �ND � ∑ �D��� ���e,IR ���b,IR �D ∑ �ND��� ���e,IR ���b,IR �ND � �D �ND���IR (2) �D is the voltage of the defective module, �� the number of measuring points of �D during ��IR , �ND is the voltage of the non-defective reference module while ��� indicates the number of measuring points of �ND during ��IR. � is the module’s power. To facilitate the manual data processing, the pre-analyzed power loss data supplied by the SunSniffer software was used to filter the modules in advance. To determine a sensitive power loss limit, some strings’ module powers were analyzed which appeared to be perfectly fine. One string differed by 4% in power losses, another one showed a variation of 3%. Also, the uncertainty of the modules’ power amounts to 3% as given by the manufacturer. Thus, a sensitive power loss threshold of 3% is considered sufficiently high for filtering the naturally occurring power fluctuations whilst being low enough to be a sensitive power loss threshold. The relative power for each pre-filtered module was calculated according to equation (2) and correlated to the module temperature differences obtained from the IR images according to equation (1). Fig. 2 (left) depicts the IR image taken of one module exhibiting an inhomogeneous temperature pattern. A superimposed pattern due to convection can clearly be seen as the lower row of modules is noticeably colder, so the reference module has to be chosen with care. It should be in a similar convective environment, thus, a module in the same row next to the defective module was chosen in this example. In Fig. 2 (right), the module voltage is found to be lower than its reference equaling a mean relative power loss of 13% averaged over the aIR-measurement time period. The mean module temperature amounts to 51.9 °C while the maximum module temperature is 63.2 °C. The mean reference module temperature is 50.9 °C resulting in a mean module temperature difference of 1 K. As can be seen, even a seemingly minor temperature difference of the mean module temperatures of 1 K can be related to a non-negligible power loss of over 10%. Δ Δ linear@950W linear @440W known measurement points Janine Teubner et al. / Energy Procedia 124 (2017) 560–566 563 T� T��� � �� � �� Prel ���� � ��,�� � ��,�� ��,�� ��,�� �rel����IR�, �U��IR�, ���IR�� � �D�ND��IR � �D�ND � ∑ D���b,IR�D∑ �ND���e,IR���b,IR�ND � �D�ND���IR �D �� �D ��IR �ND��� �ND��IR � Fig. 2. Left: The IR image shows the module from Fig. 1 which exhibits temperature irregularities (white rectangle). Right: Voltage over time for the same module and the reference module next to it during the aIR-measurement interval from the SunSniffer system. Then, in order to gain a correlation for a second set of ambient parameters, a plant on an industrial building in Nuremberg (plant N) was measured on 01.09.2016 from 14:00-16:00 with an irradiation of 440 W/m and an ambient temperature of 30°C. It consists of several roofs with 4552 solar modules amounting to 1092 kWp. The latter is not fully equipped with SunSniffer and allows no allocation of modules. Therefore, the results from plant C are applied and modified to assess the modules found via aIR-PV-check in plant N.