The Construction Static Analysis And Aerodynamic Stability Of Suspension Bridge With Steel Truss Stiffening Girder

Because of the structural characteristics,the steel truss stiffened suspension bridges in both construction and severice state have special static and aerodynamic issues.Typical issues are the overstress of the trusses when erected and the complex flow patterns around the girder,which make it more challenging when investigating its aerodynamic performance.In addition,some other problems,such as the determination of the reasonable state of the bridge and the large vibration of the main cable during construction,have become the basic concerns of the suspension bridge.In light of above mentioned issues,this study is to carry out the following subjects:(1)Determinning the finished state of suspension bridgesAn analytical algorithm is proposed to calculate the internal force and configuration of the cable system in the suspension bridge.Compared with the existing algorithms,the proposed algorithm can accurately consider the influence hangers' weight,and can be used for the two-and three-dimensional cable system with slight modification.In addition,an iterative algorithm for obtaining the finished state with the finite element method(FEM)is also proposed.(2)Optimizing the construction schemes for the suspension bridgesTo facilitate the optimizition process for the suspension bridges,an analytic method is proposed to calculate the deformation of the bridge with large displacement effect in consideration.On this basis,a primary construction scheme for the deck slab is efficiently determined.With the availiable errecting scheme for the steel truss subsection,this study systematically carried out the optimization process among the feasible construction schemes by comparing the construction stress,construction convenience,and wind resistance reliability,and eventually determined an appropriate scheme for the project.Furtherly,by slightly adjusting the length of the rigid-joint section in this scheme,the stress of the truss during the entire construction is greatly reduced.(3)Establishing the nonlinear galloping analysis method for the cable.In view of the shortcomings of the current linear galloping theory,a nonlinear analysis method suitable for suspension cable is established in this study.The analysis method can fully take into account of the geometrical nonlinearity and the vibration of the main cable in multiple directions(three translational freedom and one torsional freedom).With the availiable quasi-stationary wind load,this approach is applied to the main cable during construction,and the nonlinear dynamic behavior of the main cable at different wind velocity is analyzed and obtained.(4)Simulating the nonlinear aerodynamic loads of bridge sectionsFor the bridge deck with a bluff cross-section,the nonlinear,unsteady characteristics resulting from the flow separation,reattachment around the deck and three-dimensional(3D)nature of the ensuring wake preclude the formulation of an analytical expression of the bridge aerodynamics.Although the CFD schemes show a great promise of simulating nonlinear bluff-body aerodynamics,the computational effort is extremely high considering the three-dimensional nature of wakes and intensive flow separations from the structure.In this study,the newly proposed s-LSTM networks is used to develop a reduced-order model of the wind-bridge interaction system.The s-LSTM networks are trained using the high-fidelity input-output aerodynamics datasets,where the amplitude and frequency ranges of the input signals are appropriately selected.On this basis,a hybrid training scheme consisting of SGD(Stochastic Gradient Descent)and MBFGS(Modified Broyden–Fletcher–Goldfarb–Shanno)is proposed which can improve the efficiency about 30 times higher in the training process.The adopted numerical examples indicate the proposed framework can generate the corresponding reduced-order models with high-fidelity,and the bridge motion-induced nonlinear unsteady aerodynamics forces and hence the nonlinear post-flutter behaviors can be accurately and efficiently predicted with the trained s-LSTM networks.(5)Developping the efficient method for analyzing aerodynamic stability of bridge structures under nonlinear aerodynamic loadsThe deep learning technology is used to simulate the aerodynamic load of typical bridge sections.The trained neural network is used as the ?provider? of aerodynamic loads in the 3D bridge aeroelastic analysis,thereby eliminating the computational inefficient fluid calculation,which makes it possible to perform nonlinear aeroelastic analysis of 3D bridges on personal computers.In order to further improve the efficiency,the dynamic response of the bridge under arbitrary load is determined by the superposition of multiple convolution calculations based on the Duhamel integral principle.The proposed methodology is used to obtain the aeroelastic responses of two suspension bridges respectively with the steel box girder and steel truss girder,and the corresponding aerodynamic performance in the post-flutter is obtained.