Prediction of the shear strength of reinforced masonry walls using a large experimental database and artificial neural networks

Abstract

This paper analyses the accuracy of a selection of expressions currently available to estimate the in-plane shear strength of reinforced masonry (RM) walls, including those presented in some international masonry codes. For this purpose, predictions of such expressions are compared with a set of experimental results reported in the literature. The experimental database includes specimens built with ceramic bricks and concrete blocks tested in partially and fully grouted conditions, which typically present a shear failure mode. Based on the experimental data collected and using artificial neural networks (ANN), this paper presents alternative expressions to the different existing methods to predict the in-plane shear strength of RM walls. The wall aspect ratio, the axial pre-compression level on the wall, the compressive strength of masonry, as well as the amount and spacing of vertical and horizontal reinforcement throughout the wall are taken into consideration as the input parameters for the proposed expressions. The results obtained show that ANN-based proposals give good predictions and in general fit the experimental results better than other calculation methods.