Probabilistic deformation capacity models of reinforced concrete columns failed in flexural-shear based on Bayesian theory

In order to predict the deformation capacity of reinforced concrete columns (RC) failed in flexural-shear under seismic loading, a methodology to construct probabilistic deformation capacity models of columns is developed based on Bayesian theory. The prior models are proposed based on evaluating the biases of base deterministic models. And, the quasi-static test data of 20 rectangular RC columns failed in flexural-shear were collected from PEER-Structural Performance Database and used as objective knowledge. Two types of information, epistemic information and experimental information, are synthesized by Bayesian estimation method and inferred to update prior models. The probabilistic models for the deformation capacity of RC columns failed in flexural-shear under cyclic loading are constructed. Subsequently, the probabilistic models are simplified by using the stepwise procedure to delete unimportant terms and the dynamic updating of prior models is achieved. Then, the effect of different correction items is discussed. The results show that the approach inherits the completeness of prior models and the accuracy of experimental data information. Therefore, it could more accurately predict the deformation capacity of RC columns failed in flexural-shear, which is convenient to consider the capacity uncertainty in the aseismic design or in the evaluation of RC column.