Tridimensional microtomographic image segmentation of lightweight concretes using automatic thresholding techniques for apparent porosity calculation

Abstract

Construction industry demands sustainable processes and the addition of lightweight aggregates presents as an effective solution to obtain different concrete mixtures by recycling production outputs from other industries, as the Ethylene-Vinyl Acetate (EVA), raw material mostly utilized at footwear industry. Despite all the sustainable appeal in the recycle of EVA or other lightweight aggregates into the concrete mixtures, the loss of resistance is a frequent issue that can be prevented by adding natural fibers, such as piaçava, to the composition. Aiming the usage of these new materials, it is necessary to know their characteristics. In materials science the techniques of non-destructive tests stand out for maintaining the integrity of the material in question promoting savings and sustainability. In this work, X-ray computerized microtomographic images from samples of different lightweight concrete compositions were analyzed. These analyzes were carried out by image processing techniques, such as the Otsu and k-means, aiming to characterize the microstructure of the cementitious matrix of lightweight concrete mixture with EVA and piaçava fiber. The three-dimensional analysis of the microtomographic images offers descriptors that make it possible to characterize these samples. The results were satisfactory when compared with mercury porosimetry tests performed on the samples.