Study On The Modification Method Of NSM FRP-to-concrete Bonded Joints Strength Models

The NSM-FRP reinforcement technology is a concrete structure reinforcement method flourished in recent years.In the NSM-FRP reinforcement system,the FRP-to-concrete bonded joint is a weak link,the performance of which decides the reinforcement effect.The NSM-FRP reinforcement technology has attracted great conc-erns of researchers as its significant advantages.A bond bearing capacity prediction model has been proposed to predict the bond mechanical properties of FRP-concrete interface in research process.The model is based on limited experimental data,incomplete research parameters and idealized calculation methods,therefore uncertainty about the model exists,which can cause errors in the prediction model.The security hidden danger exists in actual reinforcement project if the error exceeds the allowable range.In view of this,this paper presents a model updating method for the bond bearing capacity of FRP-concrete interface considering the uncertainty in model.The ma-in research contents are as follows:Firstly,the prediction model and its uncertainties are analyzed.Based on the collected data,the selected bearing capacity model is calculated,and the uncertainty coefficient is defined through the calculated bearing capacity value and the measured bearing capacity value.It is shown that the calculation value of bearing capacity isdiscrete and deviates greatly from the measured value by analyzing the statistical characteristics of the uncertain coefficient.On this basis,explicit a strong correlation bet-ween the uncertain coefficients and the input parameters using Spearman correlation coefficient methodSecondly,the relationship between each input parameter and the uncertain coefficient is analyzed on this basis.A function between uncertain coefficients and input parameters is established to describe the correlation between uncertain coefficients and input parameters.The function is solved based on the multiple regression analysis method.The uncertainty factors in model is removed by this function to modify the model,the calculated value of the model is close to the measured value and the discreteness is greatly reduced.Thirdly,the JC method is used to analyze the reliability of the model.The results by comparing reliability index in three cases indicate that the design reliability without considering the model uncertainty is dangerous when the model uncertainty is seriously underestimated.At the same time,the reliability index can converge to a unified and reasonable level by removing the dependence of uncertainty coefficient on design parameters.