Abstract:
The performance and degradation rate of photovoltaic (PV) modules primarily depend on the module design, technology type, field operating conditions and climate. With the increasing penetration of both grid connected and standalone photovoltaic (PV) systems to cater the shortfall of energy, it is necessary to determine the performance and durability of installed PV systems in different climates. The identification of the underlying performance degradation mechanisms of different technologies over the years of operation in different climatic regions can lead to better design and reduce financial risk for investors. In order to understand the degradation behavior of PV systems, there is a need for highly accurate performance modeling and real-time data analysis. The best-known method so far for the PV performance evaluation is the I-V method; however, its downside is that the PV plant has to be taken offline, hence incurring the downtime cost. This research aims to determine the most optimal method for the online computation of the degradation rate of a PV system while being actively tied with the grid. The literature usually offers three metrices to estimate the performance of an in-field PV system: performance ratio (PR), performance index (PI) and kilowatt-hour (KWh). However, these metrices are not enough to reflect the real performance of system when system logged performance and environmental data are biased due to irregular shutdowns, poor calibration of sensors and non-uniform irradiance. Purpose of this research is to make a model that incorporate these metrices with real time statistical analysis and modelling of system logged data to mitigate the bias and enhance the accuracy of evaluated performance and degradation rates.
In this research grid connected PV power plants from the hot-humid climate of Islamabad, Pakistan and hot-dry climate of Tempe, Arizona are analyzed by using above mentioned model. Pakistan’s power plant degradation rates calculated by using only performance metrices are very high; however, these uncongenial results was due to biased data that modified with statistical models (ARIMA and Winter’s) to extract the exact information. These methods then validated by using python RdTOOLS from NREL to obtain optimal online degradation rate calculation model.
