The internal attack by sulfates in concrete is basically characterized by the formation of deleterious byproducts, such as delayed ettringite formation, which occur due to the presence of the sulfur element (S) in its chemical structure. The byproducts have expansive properties, which, after the cure of the concrete, enhance the occurrence of internal strains, promoting crack propagation, mass loss, decrease of mechanical resistance and durability of the structure. Although the subject is currently studied, there is little research related to data optimization and statistical study of the morphology and size of these byproducts. This article describes the adaptation of the hierarchical analysis to attribute statistical weights in the qualitative results obtained by the scanning electron microscope (SEM). To obtain the main objective of this research, roller-compacted concrete was extracted with possible occurrence of internal sulfate attack in a dam located in the state of Paraná, in southern Brazil. Then, the samples were mapped in laboratory and observed the presence of gypsum, in addition, measurements of the size of the ettringite were carried out with SEM-FEG/EDX and qualitative and semi-quantitative chemical microanalyses by dispersive energy spectroscopy, EDX. The adaptation of the hierarchical analysis to attribute statistic weights proved to be an option for the transformation of the qualitative data of the ettringites observed in the SEM-FEG.