Bayesian approach for optimal PV system sizing under climate change

This paper proposes a novel statistical approach for optimally sizing a stand-alone photovoltaic (PV) system under climate change. Traditionally, the irradiation profile of a typical day or year is used to size PV systems. However, facing the global warming crisis as well as the fact that no two years would have the same weather condition for a single site, this often makes the traditional way failed in the extreme weather conditions. This paper presents a method to statistically model the trend of climate change year by year and put it into the sizing formula, so that the results are optimal for the current weather condition and confidential for the future as well. Hence, the suitable sizes for the PV array and the number of batteries are obtained by pure computation. This is different from the traditional simulation-based sizing curve method. An economic optimization procedure is also presented. In addition to the capital and maintenance costs, a penalty cost is introduced when service fails. A new statistic-based reliability index, the loss of power probability, in terms of threshold-based Extreme Value Theory is presented. This index indicates the upper bound reliability for applications and provides rich information for many extreme events. A technological and economic comparison among the traditional daily energy balance method, sizing curve method and the proposed approach is conducted to demonstrate the usefulness of the new method.