Data assimilation and reliability estimation of existing rc structure

Abstract eng:
The clear line that separates aleatory and epistemic uncertainties does not exist, but it depends on the limit state model we adopt. In the process of reliability estimation of existing structures, epistemic uncertainty should be updated by the observation, but aleatory uncertainty should not. We should pay special attention to the difference of epistemic and aleatory uncertainty in reliability estimation of an existing structure. Reliability estimation of existing structures is however more difficult than that of newly constructed structures because probability density function (PDF) of the updated model parameters are not expressed in a simple form when a nonlinear observation equation or nonGaussian variable is involved. Monte Carlo (MC) based methods for non-linear filtering or Bayesian updating techniques have been developed since 1990's. This technique is applied to many fields recently such as space physics, geophysics, medical sciences and so on. From the standpoint of inverse problem, it is known as data assimilation. Though most applications with data assimilation by SMCS concentrate on estimation of model parameters, the methodology can be extended to reliability estimation quite easily. A new framework of reliability estimation of an existing structure is proposed based on SMCS in this study. The proposed method is demonstrated through a numerical example of reliability analysis as to deteriorating RC structures. Corrosion of rebar due to chloride attack is a major mechanism for the deterioration of RC structures. In the numerical example, the uncertainty types, aleatory and epistemic, are discussed. It shows the reliability estimation with epistemic uncertainty assumption gives larger updating magnitude of limit state probabilities.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket