Wind Induced Fatigue Analysis Using T-B Spectrum Method And Its Application On Buffeting Fatigue Of Bridge Girder

The construction of long-span bridges(e.g.,suspension bridge and cable-stayed bridge)has brought out great convenience to human life and economic development.However,with the requirements of economic and social development,modern bridge undergoes a rapidly development towards long-span and high-rise,which in turn leads to an increase of structural wind sensitivity.Currently,wind resistance research of long-span bridges mainly focuses on the structural strength,stiffness and stability,while the research on fatigue damage under long-term effect of ambient wind is relatively rare.Therefore,it is particularly important to develop wind-induced fatigue analysis method for structural durability design and reliability evaluation.This study focuses on the simulation of stationary wind field,random fatigue analysis and bridge buffeting fatigue.The main work and result are summarized as follows:Wind field simulation can provide load input for time domain fatigue analysis.At present,the most commonly used simulation method is classical spectral representation,and its simulation efficiency mainly depends on two aspects,namely the decomposition of cross power spectral density matrix and the summation of trigonometric functions.In order to improve the efficiency of the former,interpolation-based simulation methods are applied in wind field simulation.However,since the traditional interpolation-based simulation algorithms need a large number of trial calculations beforehand to determine the interpolation point distribution,these algorithms lack practicability to some extent.On the other hand,for the sake of enhancing the efficiency of the latter,the simulation algorithm based on Taylor expansion is also used in this field.However,this algorithm does not clarify the process in determination of the required number of expansion items.Based on these problems,this study firstly proposes an adaptive interpolation-based simulation algorithm with uneven distribution of interpolation points.Then,the comprehensive analysis on Taylor expansion-based simulation method is conducted and the criteria for determining the number of expansion items are provided.The results showed that the proposed method not only enhanced simulation efficiency,but also improved model practicability.As far as average fatigue damage estimation is concerned,existing methods mainly include time domain and frequency domain methods.The time domain methods generally have high accuracy,but their heavy workload is not conducive to practical engineering application.On the other hand,the frequency domain methods are simple,but most of them belong to semi-theoretical and semi-empirical models with low precision.To this end,taking alongwind,crosswind and their coupled fatigue response of high-rise building as example,the performances of time domain rainflow method and frequency domain T-B spectrum method are evaluated,respectively.Firstly,the influence of sample duration on rainflow method is analyzed,and the corresponding modification is provided.Secondly,the comparison between rainflow method and T-B spectrum method are performed.By this comparison,an improved T-B spectrum method is developed.The results showed that the sample duration had a great influence on the performance of rainflow method;the improved T-B spectrum method had satisfactory accuracy.Based on improved T-B spectrum method and rainflow method(the influence of sample duration has been considered),the disturbance of uncertainties from model,damping and material parameters on wind-induced fatigue damage estimation is analyzed systematacially,where rainflow method is regarded as benchmark.Firstly,the uncertainty introduced by fatigue damage calculation model is evaluated.Based on the existing estimation method of model uncertainty for narrow-band Gaussian stochastic process,a suitable method for assessing model uncertainty caused by broadband Gaussian stochastic process is proposed.Secondly,30% damping variability is used as an example to discuss its disturbance on fatigue damage estimation.Thirdly,the effects of material parameters uncertainty on fatigue damage estimation are described by the homoscedastic variance hypothesis and heteroscedastic hypothesis models,respectively.Finally,the study that considers all the above uncertain factors is carried out.The results presented that compared with model uncertainty using narrowband Gaussian assumption,the method proposed in this study had higher accuracy;the fatigue damage estimation results were related to all the above uncertain factors;the improved T-B spectrum method was still effective to evaluate fatigue damage in the above cases.Wind-induced fatigue analysis is also important for bridge structure.This study adopts Puli Great Bridge as the engineering background to study the two aspects of wind-induced fatigue,namely mean wind speed distribution and bridge buffeting fatigue analysis.The existing bridge buffeting fatigue analysis often ignores the influence of wind direction,which is regarded to be blown from the direction perpendicular to bridge axis.In order to consider its influence on buffeting fatigue damage,KDE-GPD is firstly proposed to describe the joint distribution of wind speed and wind direction based on the 33 months wind speed measured data at the bridge site.Secondly,based on improved T-B spectrum method,comparative analysis of main beam buffeting fatigue damage between considering and ignoring the influence of wind directionality is carried out.Finally,the study that considers the effects of model,damping and material parameters uncertainties on fatigue damage estimation is implemented.The results indicated that KDE-GPD could better reflect actual wind speed distribution;model uncertainty may be neglected in the bridge buffeting fatigue analysis;the considerations of damping and material parameters uncertainties increased fatigue cumulative damage;on the other hand,the influence of wind directionality decreased the damage.