Evaluation and probabilistic prediction of shear strength for RC beams without shear reinforcement
Date of Issue2016-05-16
School of Civil and Environmental Engineering
Over the past decades, a large number of shear strength models have been developed by numerous researchers. The majority of the developed shear strength models works in deterministic manner with a collected database and a simplified mechanical or semi-empirical representation. However, the uncertainty of shear strength was neglected in application. It is difficult for engineers to choose an appropriate prediction model in the engineering practice due to the large scatter among the deterministic predictions. Therefore, this report aimed to evaluate the accuracy of proposed deterministic shear strength prediction models and develop a probabilistic prediction model of shear strength for RC beams without shear reinforcement. Research on shear behavior in the past one hundred years was reviewed with attention paid to load transfer mechanisms development. Furthermore, evaluations of eight well-known shear strength models were carried out to investigate the reliability of deterministic prediction models of shear strength for RC beams without shear reinforcement based on the database of 127 tested beams. It concludes that the shear strength predicted by design provisions of CSA (2004) is conservative with relatively high accuracy. A probabilistic model to predict the shear strength of RC beams without shear reinforcement was proposed by both mechanical approach and data-driven approach. Specifically, a function giving the probabilistic shear strength was derived from the commonly known relationship between the shear forces and the rate of change in bending moment along the beam to reflect the combination of beam action and arch action. The GLUE method was then adopted to update the probability distribution of two unknown model parameters, specifically, k1 and k2, as posterior Weibull distribution from the prior uniform distribution. The mean prediction and standard deviation prediction models were proposed to give a prediction band of shear strength for each specimen for the purpose of facilitating use in engineering practice of this probabilistic prediction model.
Final Year Project (FYP)
Nanyang Technological University