Damage assessments of reinforced concrete shear walls using visual features of surface crack patterns

The purpose of this study is to quantify the damage of reinforced concrete shear walls after the earthquakes. One of the most important tasks after earthquakes is to assess the buildings and make sure that they are safe to occupy. So far, this assessment is mostly conducted by visual inspection and measuring the crack width that is prone to error. Thus, automation of the damage assessment using image processing techniques can significantly improve the assessment accuracy. In this paper, a database comprising 236 images of 72 damaged walls at various drift levels is collected. The database covers a broad range of structural and geometric properties. The crack pattern associated with each wall and each drift ratio is one by one extracted manually, and the corresponding geometry indices of each crack pattern are calculated. The indices are assessed and the correlation with drift ratio is investigated. Finally, using symbolic regression, empirical equations are proposed that can predict the peak drift ratio that the wall has experienced during an earthquake using the crack pattern. Four scenarios are considered for obtaining the empirical equations. The predicted drift ratio along with fragility curves can finally be used to estimate the damage level of the walls.