Study On Wind Load And Wind-induced Response Of Large-span Roof

Components and claddings (C&C) on buildings often damage earlier comparing to the main wind-force resisting system (MWFRS) under wind load, especially for the long-span roof structures. The structural internal pressure will significantly be changed after the damage of C & C such as curtain walls, doors, exterior windows and roof coverings, and subsequently the characteristics of wind loads acting on MWFRS will accordingly be changed relative to the enclosed cases. Therefore, analysis on wind-induced internal pressures and responses of buildings with wall-opening or roof-opening is considered indispensable. In addition, due to the limits of experimental facilities, the arrangement of taps on the full-scale or modeling buildings cannot always meet the requirement for wind load analysis. Therefore, an appropriate interpolation method is required to extend the existing data acquired from wind tunnel tests or full-scale experiments to other unknown points.Based on wind tunnel tests and field measurements, this thesis presents the investigations on the characteristics of external pressures and wind-induced internal pressures on a long-span structure with openings on walls and roof center. Furthermore, theories of wind-induced internal pressures were also studied. Then the wind-induced responses and the optimization of structural wind resistance were investigated systematically. Lastly, combining the engineering example with field measurements, a wide range of wind pressure interpolation methods were discussed in this thesis. The main research contents and results are as follows:In the first part of this thesis, the wind-induced internal pressures of building with wall-openings and roof-opening were investigated in detail. A rigid wind tunnel model with multi-openings was designed, and a series of pressure testing experiments were conducted under ten cases. Discussions on the values, distributions and spectral characteristics of wind-induced internal pressures were made under different situations. Comparative analysis of statistical parameters, such as the power spectral density and spatial correlation, was taken between the internal and external pressures. Also, the flow exponents under different structure situation were studied based on the wind-induced internal pressure transmission theory. The results show that the internal pressure coefficients can be expressed with a constant value when the aperture ratio is less than 10% in the case without roof-opening; however, for the case with an opening in the roof center, wind-induced internal pressures cannot be suitably represented by a constant value. Large negative pressures occurred in the downstream areas of the hole in the situation with roof-opening. Flow exponents in the situation without roof-opening agree well with the results in related literatures. Moreover, the values of flow exponent in the situation with roof-opening were suggested in this thesis.In the second part, the wind-induced responses and structural optimization method were studied. Based on the wind tunnel test data, a time domain analysis was conducted using finite element analysis software. Taking the time-domain results of wind-induced responses as references, the applicability of two types of frequency-domain methods for calculating the wind-induced responses of long-span structures, namely Complete Quadratic Combination (CQC) method and Square Root of the Sum of the Squares (SRSS) method, was discussed. And then, the effects of mode shape selection and modal coupling on frequency-domain analysis methods were evaluated. The study on wind-induced response optimization was adopted on the basis of finite element structure's design parameters, including single objective optimization method and multi-objective optimization method. The results show that: the CQC method is appropriate to calculate structural wind-induced response of wind-sensitive structures, while the SRSS method underestimates the wind-induced response because of ignoring the influence of the modal coupling term for wind-sensitive structures, such as long-span roof structures. The influence of modal coupling term on the RMSs of wind-induced fluctuating displacements can be positive or negative. For instance, there is a positive effect in the windward and vortex (conical vortex) areas. Fifty-order computing modes can basically reflect the general trend of wind-induced responses, and more accurate results can be obtained by 150-order computing modes. State variables and objective function are incapable of achieving optimal by single objective optimization method. Nevertheless, a series of optimal solution can be draw with setting different weight coefficient for the objective function according to the design intent in multi-objective optimization method, recommended as effective method for structural optimization design consequently.Analysis on several interpolation theories and interpolation techniques were conducted in the last part of this thesis. Example in the aspect of wind pressure time history simulation and reconstruction of measuring points was given to validate different interpolation theory, such as spatial interpolation method, AR autoregressive model method, and neural network, concentrating on different interpolation methods in various interpolation theories concretely. The method by the combination of neural network and wavelet decomposition was proposed and applied on engineering project and field measurement for analysis and validation. The results show that the interpolation effect of fluctuating wind pressure coefficient generated by spatial interpolation method isn't ideal, for just taking the correlation of wind pressure distribution in the geometry space into account. By comparison, the interpolation of triangulation with linear interpolation method (TLI) and radial basis function method (RBF) have relatively better precision. The fitting curve of fluctuating wind pressure power spectrum generated by AR autoregressive model principle has good interpolation accuracy, based on wind pressure time history of raw data points. Neural network possesses ideal interpolation results of distribution, by training and interpolating on average wind pressure coefficient and power spectrum, while deviation exists in high frequency parts of wind power spectrum. The good precision of interpolation is shown in engineering project and field measurement, by adopting the method of combinating neural network and wavelet decomposition, which avoids the energy leaking problem, brought by cutting off frequency in FFT analysis.