Reducing Measurement Uncertainty of Temperature Coefficients of Photovoltaic Modules

Abstract

In temperature coefficient measurements of PV modules, a low level of agreement in results is observed between different laboratories, with deviations up to ±10% for Pmax temperature coefficients. Such deviations are considered as an uncertainty component in PV energy yield assessments, and are seen as financial risk by investors and purchasers. A ±10% error in the Pmax temperature coefficient can result in more than 1% error in estimated energy yield (and thus financial return on investment), which is significant especially for utility scale PV. The goal of this work is to reduce the uncertainty of the Pmax temperature coefficients from ±10% down to ±5%, which will reduce financial risk and therefore support the roll-out of utility scale PV. This thesis will focus on identification and reduction of measurement uncertainty of the temperature coefficients. A measurement setup for PV module temperature coefficients, integrating a solar simulator and a temperature chamber, was developed by Eternalsun Spire. This setup will be evaluated using a Monte Carlo based uncertainty analysis. Additionally, uncertainty drivers - relevant for current, typical industry practice - will be demonstrated, contributing to further understanding of observed deviations between temperature coefficient measurement results from different laboratories. The k=2 expanded measurement uncertainty for the relative temperature coefficient for a PV module measured in the Temperature Coefficient Lab Flasher is approximated at 4.05%. Three major uncertainty components are: 1. Temperature induced spectral mismatch (2.46%) 2. STC uncertainty of the solar simulator (1.09%); 3. Temperature correction of the monitor cell (0.41%). Additionally, it is demonstrated that: 1. silicon heterojunction PV modules can exhibit non-linear temperature behaviour and therefore coefficients that are dependent on the range of measurement; 2. significant measurement errors can occur when the cell temperature and backsheet temperature are not in thermal equilibrium, which typically applies for the currently widely applied ”hot potato” method for determination of temperature coefficients; 3. the simulator spectrum can significantly affect the temperature coefficient measurement even for solar simulators with A+ rating for spectral match. Based on the uncertainty analysis it is shown that the temperature coefficient measurement setup can be used to achieve a measurement uncertainty of PV module temperature coefficients below 5%. Additionally, the demonstrated uncertainty drivers encountered in typical industry practice can support awareness and thus mitigation of uncertainty drivers previously not aware of. For this purpose, recommendations from this work have been included in an international IEC standard for measurement of temperature coefficients.