Estimation of multiple crack propagation pattern in concrete using Voronoi tessellation method

In this study, the multiple crack propagation is examined using a machine learning algorithm developed based on Voronoi tessellation method. In this way, a novel classification and geometrical dimensions-properties estimation approach, and a future-behavior-prediction method are proposed for all multiple crack formations on concrete surfaces. The crack properties, such as width, depth, length, volume, formation, etc., are measured using digital images for multiple crack formations. Then, the cracks are divided into parts to evaluate the capability of estimating the crack pattern by the use of methodology proposed in this study. Real crack formations in a construction site are observed and used in the experimental studies. The results of the experiments show that the technique developed is precise and effective for estimating multiple crack propagation on concrete surfaces. The details of the methodology developed, the crack formations observed and the estimation results obtained are presented in this paper.