YENİ BİR YÖNTEMLE FOTOVOLTAİK MODÜLLERİN İÇ VE DIŞ SICAKLIK KATSAYILARININ KARŞILIŞTIRILMASI

Bu calismada, fotovoltaik modullerin ic ve dis ortamda elde edilen sicaklik katsayilarinin karsilastirilmasi icin yeni bir yontem ortaya konmustur. Fotovoltaik modullerin/orgulerin gercek performanslarinin dogru bir sekilde simule edilebilmesi icin dogru sicaklik katsayilarinin kullanilmasi oldukca onemlidir. Bu yuzden, gercek performansin simulasyonunda hangi tip sicaklik katsayilarinin (ic veya dis) daha dogru sonuclar verecegi belirlenmelidir. Kisa devre akimi, acik devre gerilimi,  maksimum cikis gucu ve uretilen enerji gercek performans parametreleri olarak kabul edilmislerdir. Bu calismada ortaya konan yeni yontem, gercek performansi ic ve dis ortam sicaklik katsayilari icin simule etmekte ve hangi tip sicaklik katsayilarinin daha dogru oldugunu belirleyebilmek icin dis ortamda olculen gercek performansla karsilastirmaktadir.


INTRODUCTION
The photovoltaic (PV) phenomenon provides clean and efficient energy to all humanity. Forecasting the energy produced (E) by PV arrays is important for to analyze their economic viability and inspect their operation (Rodrigues et al., 2016). PV arrays are formed from identical PV modules that electrically connected in series-parallel combinations. Once knowing the PV module's performance, it is possible to calculate the PV array's one (Rus-Casas et al., In this study, a new method is presented to compare the indoor and outdoor TCs. The actual performance of PV module and array is simulated both for the indoor and outdoor TCs. Then simulated performances are compared with actual (measured) ones to clarify which TCs (indoor or outdoor) are more accurate in simulating the PV performance. The indoor TCs are taken from a PV module datasheet whereas the outdoor TCs are evaluated by means of shading procedure at field.

Actual Performance of PV Module
In this study, P M , I SC , V OC and E are considered as the actual photovoltaic performance parameters i.e. the actual performance (Hussein et al., 2004). The irradiation (G) and module temperature (T M ) dependence of P M , I SC , V OC and E are defined with well known expressions given below (Skoplaki and Palyvos, 2009). These expressions are valid for both PV module and array.
Where n is the ideality factor of individual solar cell, N S is the number of individual solar cells connected electrically in series within a PV module, k B is the Boltzmann constant, q is the charge of electron, G REF is the reference irradiation ( 1000 W/m 2 ), T REF is the reference module temperature (25 o C), P MREF is the reference peak power, I SCREF is the reference short-circuit current, V OCREF is the reference open-circuit voltage . In addition, G and T M are the irradiation and module temperature, respectively, which correspond to the operating conditions where a PV module is deployed outdoor.

Indoor and Outdoor Temperature Coefficients
Indoor TCs are evaluated by manufacturers at controlled laboratory conditions and are given in PV module datasheet (Dubey et al., 2015). On the other hand, outdoor TCs are evaluated at field considering particular constrains (Dubey et al., 2015;Emery et al., 1996). Because of many challenges in evaluating outdoor TCs, these constrains provides to obtain reliable and repeatable results (Dubey et al., 2015;Mihaylov et al., 2016). Outdoor TCs of any photovoltaic module (α OUT, β OUT , γ OUT ) are calculated from temperature dependent I-V curve measurements that conducted a day with conditions of stable sunshine around solar noon (high than 800 W/m 2 ) and at calm wind speed (less than 2 m/s). Shading procedure is utilized to create temperature gradient on a PV module. First of all, a PV module is shaded with an opaque cover until it's temperature reaches near the ambient temperature. Then, I-V curves of a PV module are scanned with sampling interval (1 or 5 minutes) as the module temperature (T M ) rises due to removing a cover until the T M reaches in thermal equilibrium with environment where a PV module is deployed (Emery et al., 1996). The I SC , V OC , and P M parameters are extracted from the T M dependent experimental I-V curves. After that, the normalized I SC , V OC , and P M parameters are sketched with respect to the normalized module temperature, according to Table 1. The linear functions are fitted to the scattered data. Finally, the slopes of these functions correspond directly to the outdoor TCs of these parameters (α OUT, β OUT , γ OUT ) (

New Comparison Method for Temperature Coefficients
The actual performance (P M , I SC , V OC and E) is simulated for same operating conditions (G and T M ), but for different type of the temperature coefficients; indoor and outdoor TCs, using Eqs. (1)-(4). Then, the indoor and outdoor performances are compared with the actual performance measured at field by means of root mean square error approximation (RMSE) described below.
where, F MEAS , F SIM , and N are actual (measured) values, simulated values and number of data, respectively. The new method proposed here to compare temperature coefficients is shown in Figure 2. The indoor performance and outdoor performance indicate which type of temperature coefficients; indoor or outdoor, respectively, are used to simulate the actual performance.

MATERIAL
Since the main actor of PV market is crystalline silicon (Si) based PV modules, the back contact single crystalline Si PV module was selected as device under test (DUT). Currentvoltage (I-V) curves of the DUT were traced using a multi-channel measurement system. Kipp-Zonnen CM11 model type pyranometer was used to sense the irradiation (G) that exerted on the DUT. The temperature of DUT (T M ) was sensed via pasting four probes thin film Pt-100 temperature sensor on the back surface of DUT with thermal conducting paste and the temperature sensor was covered with insulating tape. Datasheet values of the DUT are listed in Table 2. In this study, the ideality factor of DUT is considered as 1.2 which is valid for a single crystalline silicon based PV modules (Bellia et al., 2014). It is note to remember that, the temperature coefficient that supplied in PV module datasheet are called as indoor ones (α IN, β IN , and γ IN ).

RESULTS AND DISCUSSION
Averaged outdoor TCs of the DUT were calculated from numerous I-V measurements during annual period of 2014. In this study, these TCs are called as outdoor ones. The calculation procedure of outdoor TCs is well described in Section 2.2. The indoor and outdoor TCs are given in Table 3.    For all the 12 days, simulated (indoor TCs and outdoor TCs) curves match well with actual ones. The RMSE values of simulated parameters were calculated and shown in Table 4. The actual (E ACT ), indoor TCs (E INDOOR ) and outdoor TCs (E OUTDOOR ) energy values were calculated from actual P M -local time curves, indoor TCs P M -local time curves and outdoor TCs P M -local time curves, respectively, according to the Eq.(4). The calculated energy values (E ACT , E INDOOR , E OUTDOOR ) and corresponding error values are shown in Table 5. The PV array ( Figure 6) with 8.4 kW P rated output peak power that located in the campus of Muğla Sıtkı Koçman Üniversity is used also to verify the effectiveness of the new comparison method. The details of the PV array is well described elsewhere in (Eke and Senturk, 2012).  (Eke and Senturk, 2012).
Since the PV array is formed from 84 numbers of identical DUTs, the indoor and outdoor TCs are assumed valid for the PV array. The operating conditions (G and T M ) were taken from a data-logger that integrated into the PV array (Eke and Senturk, 2012). One day was selected to test the new method. Since data-logger does not store actual values of I SC and V OC of the PV array, only the P M and E values were simulated for both the indoor and outdoor TCs using Eq.(1) and Eq.(4), respectively, and corresponding operating conditions (G and T M ). The simulated (indoor and outdoor TCs) and actual values of P M were sketched versus local time and shown in Figure 7. To see difference of the indoor and outdoor TCs clearly on simulating the actual performance, the absolute differences (|Diff.| were calculated between the error values of the indoor TCs and outdoor TCs. The RMSE and absolute difference values, shown in Table 4,  Table 5 and Figure 7, indicate that there is not significant discrepancy between the indoor and outdoor TCs as simulating the actual performance of PV module and PV array at field. In some measurements, discrepancies were observed for absolute differences (|Diff.| of simulated parameters (see Table 4 and Table 5). These discrepancies could be attributed with the outdoor TCs evaluation procedure where operating conditions are not exactly invariant as the indoor procedure. Since the maximum absolute difference is 0.4% (marked with grey in Table 4 and  Table 5), these discrepancies are trivial.

CONCLUSION
In this study, a new method is presented to compare the indoor and outdoor temperature coefficients. Different from the conventional comparison method, the novelty of new method is to use the actual performance (I SC , V OC , P M and E) as decisive index to compare the indoor and outdoor TCs. The new method is validated for the back contact mono-crystalline Si PV module and PV array at field. It is concluded that both indoor and outdoor TCs could simulate the actual performance of PV module and PV array almost with same accuracy. Thus despite they have been evaluated at fixed laboratory conditions, the indoor temperature coefficients are quite enough to simulate the actual photovoltaic performance at field. Since manufacturers of PV modules always provide these temperature coefficients in PV module datasheet, it is not necessary to obtain and utilize the outdoor TCs as simulating the actual performance of PV module or PV array. Because obtaining outdoor temperature coefficients is cumbersome process where all external parameters vary with respect to time.