Parameter sensitivity of CFRP retrofitted substandard joints by stochastic computational mechanics

Abstract

The response of both substandard as-built and CFRP retrofitted RC beam-column joints was investigated by a stochastic study to identify the effect of inherent uncertainties in material constitutive models. Since the scatter of the capacity is inevitably influenced by material properties, the relative impact of each material property on the global response was measured by the sensitivity analysis. It was conducted by evaluating the partial correlation coefficient between material properties and simulated response. First, experimentally validated deterministic nonlinear numerical models were developed in FE environment. After that, they were evolved to the stochastic level, which considers the randomness in prominent material parameters. The basic statistical characteristics and probability density functions of response variables were then provided by the probabilistic assessment. Finally, the most influential material parameters characterizing the quasi-static cyclic behavior of the as-built and retrofitted joints were outlined in accordance with the results of sensitivity analyses. In addition, the hysteric response of the as-built and retrofitted specimens was not only well-characterized by the numerical model but also local damages, such as large diagonal cracks in the as-built specimen and shear cracks after CFRP rupture in the retrofitted specimen, were adequately reproduced in the finite element environment.