abstract
This work presents the results of an investigation into the viability of several models that are commonly used to simulate the PV module temperature. In the past, these models, namely, Faiman-, King- and NOCT-model, were validated mostly for rooftop PV and in hotter climates. We test these models for façade PV in moderate climates and show that the annual average deviation between the PV module temperature simulated with those models and the measured PV module temperature is below 3.1 K between the hours of 09:00 -- 16:00 for all models used. With exception of the NOCT-model all models include fitting coefficients. Fitting those coefficients to measurement data obtained in moderate climate and for façade PV does not yield an improvement (i.e. no smaller annual average deviation between simulated and measured module temperature within measurement precision). Applying a simple 10 min. rolling average to the simulated temperature data yields similar or better results compared to the fitting of the model coefficients to the dataset. Even though the NOCT-model does not include the effect of wind speed, in contrast to the other models mentioned, it showed a very competitive performance.