Scour Risk Analysis Of Existing Bridge Based On Statistical Inversion Theory

The bridge is one of important joints in the transportation system.In recent years,bridge accidents have caused great impacts on social safety and economic development.Nowadays,most of the risk assessment methods are quantitative,meanwhile with a lot of subjective factors.So it is required to further improve the current risk assessment theory and methods.Based on the statistical inversion theory,this paper puts forward a risk assessment method.Based on this method,we analyze the scour risk scene,and then analyzing the different influencing factors which effect on failure probability.The main works of this paper are as follows:(1)Introduce the model of the depth of scouring,and establish an analytical model and the finite element model to calculate the critical load under the scouring scene as the important part of the failure function of bridge piers.According to the applied model and the actual risk scene,the risk sources in the process of scouring are summarized.(2)Propose a method to the assessment of scour risk based on theory of statistical inversion.In this method,based on Bayes update,we update the distribution of the priori information according to the measured values,The result of the failure probability can be estimated by Monte Carlo sampling in the forward model.(3)Use the priori information and the posterior information to calculate failure probability of the piers of the northern approach bridge in Sutong Bridge,and along with the variance coefficient of the scour depth reducing,the reliability of the risk assessment is improved.(4)Compare the assessment results by using models suggested in the Chinese hydrological specification and the American hydrological norms.The results show that when using American hydrological norms,assessment result is more conservative.The effects of different distributions of scouring parameters on the assessing results are analyzed.After taking different distributions,it can be shown that the influence of the model distributions is significant.And the influence of the correlation in parameter information on the risk assessment results is analyzed.The findings show that different evaluation results will be obtained after the different measured information is updated.Risk assessment is a dynamic process,and should be based on the specific environment to assess the risk.