Uncertainty Analysis Of Wind-Induced Resopnses Of Long-Span Bridges

With the improvement of society and technology,long-span bridges continuously emerge all over the world,and play a more and more important role in modern transportation.As the span of bridge increases,the structural frequence and damping ratio decrease so that bridges become more sensitive to wind loads.Therefore,studies carried on wind-induced responses of bridges become more important for wind resistant design.In past decades,many theoretical systems and technical means were proposed towards wind-induced responses of long-span bridges,however,most of them are based on deterministic analysis.Wind is a natural phenomenon with randomness,which influences the parameters involved in response analysis,e.g.,wind loads,aerodynamical characteristics and so on,leading to the uncertainty of windinduced response of long-span bridges.Hence,for providing theoretical references and methods of reliability-and performance-based design of wind resistance,it is necessary to incorporate uncertainties in wind-induced response analyses of long-span bridges.Bridge flutter and buffeting are two essential issues in wind resistance design of bridges.Flutter analysis concerning the free vibration of bridge due to self-excited force judges the instability state of bridge by determining the critical wind speed.Flutter derivatives are important parameters describing the aerodynamical characteristics of bridge sections and influence the final judgement of instability state of long-span bridges.However,due to existance of various factors,e.g.,identification technique,sampling error and randomness of vabriaion,etc.,the identified flutter derivatives surely have uncertainties which leads to the uncertainty of flutter critical wind speed.Bridge buffeting belongs to forced vibration due to the action of turbulent flow,so that turbulence characteristics of wind have important effects on buffeting analysis.The randomness of wind will cause uncertainties in turbulence parameters involved in buffeting analysis,for example turbulence intensity,integral length scale and others,and then lead to uncetain buffeting analysis results.Besides,the wind directionality has influences on the estimations of bridge buffeting under different return periods.Estimating the peak value of wind-induced response is one of the core tasks in performance-based design.Typically,the peak value can be estimated using Davenport's formula based on Gaussian hypothesis of bridge response.However,wind loads and windinduced responses of bridges may show non-Gaussianlity in some complex terrains such as mountainous area.Moreover,due to uncertainties of wind,the peak values of wind-induced responses of bridges will also be uncertain.Considering the uncertainty of peak value can improve the reliability of wind resistant design of bridgesIn the view of the above key issues,the dissertation will study the uncertainties of flutter derivative in flutter analysis,wind parameter in buffeting analysis and peak value of windinduced response.The main works of the dissertation is summarized as follows.(1)Propose flutter analysis methods considering uncertainties of flutter derivatives.Firstly,closed-form and simplified closed-form methods for flutter analysis are presented,and the mechnism of bridge instability is explained.Secondly,the uncertainties of flutter derivatives are discussed.Based on multivariate simulation,two simulation schemes are proposed to incorporate the flutter derivative uncertainty in flutter analysis.Next,based on simplified closed-form,an analytical approach is developed.The proposed methods are illustrated by an example in which probability distribution can be obtained.An assumed example is used to verified the analytical approach based on simplified closed-form.(2)Propose a buffeing analysis method that considers both of uncertainties of turbulence parameters and wind directionality effects.Firstly,multimode buffeting analysis of long-span bridge is presented.Secondly,a probabilistic 3-D turbulence model is introduced to consider turbulence parameter uncertainties.Thirdly,based on statistical imformation of turbulence intensity factors,integral length scales and exponential decay coefficients,etc.,uncertainties of turbulence parameters are considered by multivariate simulation.Next,Gaussian copula function is adopted to establish a probabilistic model of wind speed and direction to consider the wind directionalty effect.Then uncertainties of turbulence parameters are incorporated to calculate the peak buffeting responses of briges for any different return period.An example of Sutong Yangtze river bridge is used to illustrate the proposed approach.(3)Propose uncertainty analysis approaches for peak value of wind-induced response.Firstly,HPM-based translation process,a peak value estimation method,is presented.Secondly,based on the presented method,empirical formulae are developed by multiple linear regression analysis.The efficiency of empirical formulae is examined.Next,via empirical formulae,two analytical approaches are proposed to evaluate the peak value uncertainty,one is based on transformation of random variables and the other is based on Gaussian inference for the peak value at any fractile.Long data measured from wind tunnel test are used to exame the approaches in evaluating uncertainties of peak values of wind pressure coefficient and wind-induced response.