Bayesian Assessment Of The In-service Durability Status Of PC Beams Based On Corrosion Potential

As the scale of highway bridge construction increases year by year,the deterioration of the structural performance and durability state of built bridges will become more and more serious in the future,and research in the direction of health monitoring and structural performance assessment needs to be emphasized.Therefore,it is of great importance to clarify the principle and correction method of nondestructive testing of bridge structure durability index,to consider the influence of randomness and subjective uncertainty on the comprehensive assessment of the durability state of in-service structures,and to carry out research tests on the development of accelerated deterioration of structural durability and the law of structural performance decline,etc.for the rational decision of normal use and modification and expansion of bridges.The main research contents and conclusions of this thesis are as follows:(1)Briefly describe the electrochemical principle of reinforcement corrosion within the concrete structure and the evaluation criteria of each country’s specification,discuss the influencing factors of the half-cell potential detection results and propose correction methods,and carry out the accelerated corrosion test of steel bar depassification.The results show that: the half-cell potential detection value decreases with the increase of ambient temperature and concrete surface humidity,and the concrete surface humidity is the main factor affecting the potential test results;the corrosion current density threshold is proposed as the judgment criterion for the occurrence of depassivation of reinforcing steel.(2)A Bayesian model describing the interrelationship between NDT electrochemical indicators was established,and an integrated a priori model for judging the corrosion stage of reinforcing steel was established by integrating all NDT electrochemical indicators.The results show that the rust potential Ecorr and corrosion current density log(icorr)and rebar polarization resistance log(Rρ)satisfy the log-linear relationship,log(icorr)and concrete resistance log(ρ)also satisfy the log-linear relationship;the evaluation effect of the integrated system is better than that of the single "weak classifier".(3)To update the parameters of the Bayesian model describing the interrelationship between NDT electrochemical indicators using the MCMC method;to complete the training of network update for each "weak classifier";to propose a method to reduce the misjudgment rate of the integrated model by critical transfer probability,and to discuss the critical NDT reference value for the integrated model when judging the steel bar deactivation.The reference values of electrochemical indexes are discussed.The results showed that the model prediction effect was not significantly affected by different updating methods,and the updated Bayesian model was better than the a priori model in reasoning about the test values;the accuracy of each "weak classifier" in determining the corrosion stage I and II of steel bars was improved in the ranges of 1.72%～11.25% and23.96%～43.08%,respectively,after network updating training.The reference values of the NDT electrochemical indexes for the determination of the critical state of bar depassivation by the a posteriori integrated model are given,and the false judgment rate of the integrated system model is reduced to less than 3% by means of the critical transfer probability.(4)The traditional theoretical prediction model and probabilistic prediction model establishment for the time of steel deactivation,the interval estimation of instantaneous corrosion rate and corrosion amount of steel reinforcement are realized through the actual measurement data,and the variability and spatial variability under different corrosion forms are discussed.The results show that: the Bayesian-based probabilistic prediction model for the time to de-passivation of reinforcing steel is better than the traditional deterministic prediction model;the difference between the upper and lower limits of the interval prediction of the corrosion rate of reinforcing steel under uniform corrosion is controlled within the range of Δη<3%;the corrosion of reinforcing steel exhibits significant variability in time and space.(5)In-service PC hollow slab beams durability state assessment and accelerated decay tests were carried out.The results show that: the current durability state of the structure is good,but some areas are close to the critical depassivation state;the interval estimation of the corrosion rate of the steel strands indicates that the corrosion rate of each strand varies greatly in space;the overall corrosion rate varies significantly between different strands.