Calculation of sensitivity index using one-at-a-time measures based on graphical analysis

This work proposes different methods to calculate the sensitivity of system parameters based on graphic analysis. From percentages variations in base values of the parameters, one-at-a-time measures are taken to the construction of a graph that expresses the impact on the output caused by variations in the inputs. Two case studies are proposed, with simulated and experimental data, in which the proposed methods are applied and compared based on the values of the sensitivity indices for ranges of interest. All methods confirm quantitatively the most sensitive variables and the choice of one of the methods depends on the characteristics considered relevant for analysis, i.e., degree of sensitivity of a parameter in relation to another, the behavior of the curve and presence of peaks.