A machine learning-based formulation for predicting shear capacity of squat flanged RC walls

Abstract The squat flanged reinforced concrete (RC) walls have been widely utilized in nuclear power plant and building structures. Nevertheless, the empirical equations in current design codes and published studies show a significant discrepancy in calculating the shear strength of the walls. The purpose of this study is to develop an effective machine learning model, namely artificial neural network (ANN), for predicting the shear strength of squat flanged RC walls. A total of 369 test results of squat flanged RC walls were collected from the literature and used to develop the ANN model. The results of the proposed model were compared with those of existing design codes and published studies. The comparisons emphasized that the developed ANN model in this paper can predict the shear capacity of squat flanged RC walls more accurately than the existing equations. Moreover, the effect of input parameters on the predicted shear capacity of the walls was sufficiently investigated. A predictive formula based on the ANN model, which can cover thirteen input parameters, was then proposed to compute the shear strength of the squat flanged walls. Additionally, an efficient graphical user interface (GUI) platform has been established for facilitating the practical design process of the squat flanged RC walls.