Comparative Evaluation of Sensitivity Analysis Methods Applied to Distributed Photovoltaic Systems

The use of statistical techniques, such as stochastic processes, Monte Carlo simulations and analysis of variance (ANOVA), has been widely used in power system analysis and are essential tools for developing studies and models with high methodological rigor. Sensitivity analysis, on the other hand, is widely recognized for its importance in robust modeling and uncertainty assessment, and although it is receiving increasing attention, it is still underused. For this reason, this study aims to compare sensitivity analysis methods in the context of power systems, assessing their effectiveness. To this end, a methodology that determines the technical impacts of the insertion of photovoltaic distributed generation on a feeder was adapted to evaluate the effects of inverter, system and location factors on average voltage violations. The results show the superiority of the global sensitivity analysis, which, by using Monte Carlo simulations, associates uncertainties in the inputs with the outputs, revealing previously unknown correlations or confirming existing ones, such as the well-known influence of inverters' reactive power on bus voltages and its potential application in voltage control.