Structural Reliability Evolution And Life Prediction Considering The Influence Of Uncertainties And Performance Degradation

Civil structures generally have large geometric sizes,numerous components,large dispersion in material properties,complicated boundary conditions and external loads.Therefore,their safety is easily affected by many uncertain factors.Moreover,civil structures are prone to damage and performance degradation after a long-term service.Hence,how to achieve real-time assessment of structural performance,as well as service life prediction,presents a hard task that should be faced in practice.Due to it,under probabilistic and statistic frameworks,this thesis studies the reliability evolution process and the life prediction methods of RC beams.The main contents and achievements are described as follows:Firstly,long-term deformation monitoring has been carried out on three RC beams under different dead loads.The empirical deformation formula for RC beams considering shrinkage and creep is improved according to the experimental monitoring data.Since changes in external environment and dispersion in material properties will introduce extra errors to the deformation formula,this thesis develops a method that can realize real-time monitoring of structures,as well as updating of the deformation formula.The method defines the parameters of the formula as variables.Then the deformation curves of the beams are fit using the real-time monitoring data,by which means the updated formula parameters are obtained.After that,the updated formula can be used to forecast possible deformation of the beams in the future.Secondly,based on the Bayes theorem and the classic theories of probability and statistics,this thesis proposes a bearing capacity updating model that can consider the performance degradation and the external load history of RC beams.The updating model defines the bearing capacity as a random variable.After a certain validation load,the Bayes formula is adopted for updating the probability distributions of the bearing capacities of RC beams.On the other hand,performance degradation always occurs to some extent during the usage of civil structures,which cause the decrease of structural bearing capacity.Hence,a stiffness attenuation coefficient is introduced into the bearing capacity updating process.Six RC beams have been tested using static loads and the measured deflection and strain data are used for validation.The results show that the bearing capacity updating model possesses the ability of fault tolerance.Meanwhile,it is found that large errors exist in the updating results without considering the effects of performance degradation.And the estimation of bearing capacity presents larger values inducing a problem to structural safety.Thirdly,based on the estimation of future loads,this thesis links reliability indices with time in order to propose a reliability evolution curve model.Using the laboratory monitoring data of the aforementioned six beams,six reliability evolution curves are obtained for subsequent analysis.By comparing the model predictions with experimental data,it is observed that the proposed model shows certain rationality and fault tolerance.Lastly,this thesis proposes a long-term loading model based on some rational assumptions.Incorporated with the reliability method or the theory of probability and statistics,the model is used to estimate the residual lives of the six beams.The first incorporation adopts the allowable reliability value from the design code to define the failure criteria of a structure.On the other hand,the second incorporation estimates the residual lives through calculating the means of life probability distributions of the beams.Both methods may take into account uncertainties and provide quantitative results,which demonstrates their practicability.