Calculation of reliability index for in-plane shear failure of unreinforced masonry walls based on Gaussian process model

Abstract

The current methods for reliability index calculation are less effective for in-plane shear failure of unreinforced masonry (URM) walls because there is significant uncertainty in the walls’ mechanical parameters and resistance model. To address this issue, the authors propose a new method based on stochastic Gaussian process (GP) model. They conducted in-plane shear resistance experiments on six URM walls and converted the walls’ deterministic shear resistance models to prior GP models, which they updated to posterior models using the experimental results. The posterior GP models specified normal probability density functions (PDFs) for shear resistance of the walls, and the PDFs linearized the walls’ performance functions to the simplest form. With the linearized performance functions, the walls’ failure probability was calculated by direct integration and their reliability indices were determined accordingly. The reliability indices are in good agreement with those given by the Monte Carlo simulation (MCS) using a large number of samples. The proposed method does not deal with high dimensional limit state surfaces, it relieves the influence of uncertainty of the shear resistance model, and it is time-efficient. With these features, the proposed method can facilitate the reliability index calculation for in-plane shear failure of the URM walls.