Reliability analysis of a shear-critical beam

Abstract

The response of a reinforced concrete beam constructed without transverse reinforcement to achieve shear failure was investigated by experimental and numerical methods. Due to inherent uncertainties in material constitutive models, a nonlinear finite element method (FEM) was combined with a suitable stochastic sampling technique to propose a more advanced model for estimating the response of a shear-critical beam. For this purpose, the specimen was first tested under monotonic loading up to shear failure by a four-point bending test. Then, the stochastic model was developed by using Latin Hypercube Sampling (LHS) including statistical correlation among the prominent material parameters. Random parameters of concrete and reinforcement steel were defined in accordance with the material test results and code recommendations. The constituent outcomes of the stochastic model including a set of load-displacement curves are presented. The results of the stochastic approach matched well with the behavior of the specimen observed during the experimental test. The probability density function for ultimate load was obtained. After that, the reliability of the member for the ultimate limit state was compared with the code requirements to ensure the safe loading range. The design load, which corresponds the failure probability related to ultimate limit state was computed. Moreover, a simplified ECOV (Estimation of Coefficient of Variation) method was carried out to estimate the design load. It is found that the load obtained from reliability analyses for design load was reasonably in good agreement with the code recommended value.