Search

The concrete shrinkage strain and creep coefficient calculation method in CEB - FIP MC90 model is briefly introduced. For application convenience in bridge design, a simplified formula for calculation of the concrete shrinkage strain and creep coefficient is derived. Considering the varying range of every influence factor, a correlation analysis for the factor is carried out. And the weights of those factors related to concrete shrinkage and creep coefficient are discussed. Considering the climate characteristics and the bridge structure features in the Pearl River Delta area, the variation ranges of concrete nominal shrinkage strain and nominal creep coefficient are calculated respectively.

A model is proposed to simulate multivariate weakly stationary Gaussian stochastic processes based on the spectral representation theorem. In this model, the amplitude, phase angle, and frequency involved in the harmonic function are random so that the generated samples are real stochastic processes. Three algorithms are then adopted to improve the simulation efficiency. A uniform cubic B-spline interpolation method is employed to fit the target factorized power spectral density function curves. A recursive algorithm for the Cholesky factorization is utilized to decompose the cross-power spectral density matrices. Some redundant cosine terms are cut off to decrease the computation quantity of superposition. Finally, an example involving simulation of turbulent wind velocity fluctuations is given to validate the capability and accuracy of the proposed model as well as the efficiency of the optimal algorithms.

The modern prestressed concrete bridges have the characteristics of large-span, high concrete strength, reinforced volume prestress and linear shape complexity. The structure bearing capacity and durability are considerable impacted by concrete creep, such as stresses redistribution and deflections increase. Some of the prediction concrete creep models that are being widely used at present are briefly introduced in this paper. Based on the analysis of existing research achievement, for the aim of improving the calculation accuracy of creep effects in bridge structures, several main problems that should be paid attention to in concrete creep analysis are pointed out from four aspects: (1) in design stage, the reasonable creep prediction model should be selected in accord with the local building materials and environment conditions; (2) during the analysis of creep effects, the initial stresses state and the stresses redistribution should be attended; (3) the reliability of prestress loss calculation should be considered in analysis of concrete creep effects; (4) the differences between normal construction materials and that of new types should be taken into account in concrete creep analysis.

Earthquake input energy and structural energy dissipation are key indicators to assess the seismic performance of structures. To study the rules of distribution of hysteretic energy within structures, a 6-storey regular reinforced concrete frame structure model is analyzed through elasto-plastic time-history dynamic analysis using the El Centro and Northridge accelerograms. Based on the comparison between numerical results for the earthquake input energy and structural hysteretic energy under the minor, moderate and major earthquakes of Grade 8 and 9, the distribution of the ratio of the storey hysteretic energy to the total hysteretic energy through the height was further studied. It shows that the computed results corresponding to the two earthquake records are in good agreement under different ground motion severity. And the percentage of structural hysteretic energy to input energy is basically stable. The distribution pattern of storey hysteretic energy through the height is that the value of the upper stories is smaller than the value of the lower stories. And the ground motion severity has a minor influence on the distribution pattern when the plasticity of structure develops more sufficiently.

Wind loading is a dominant factor for design of a cable-membrane structure. Three orthogonal turbulent components, including the longitudinal, lateral and vertical wind velocities, should be taken into account for the wind loads. In this study, a stochastic 3D coupling wind field model is derived by the spectral representation theory. The coherence functions of the three orthogonal turbulent components are considered in this model. Then the model is applied to generate the three correlated wind turbulent components. After that, formulae are proposed to transform the velocities into wind loads, and to introduce the modified wind pressure force. Finally, a wind-induced time-history response analysis is conducted for a 3D cable-membrane structure. Analytical results indicate that responses induced by the proposed wind load model are 10%∼25% larger than those by the conventional uncorrelated model, and that the responses are not quite influenced by the modified wind pressure force. Therefore, we concluded that, in the time-history response analysis, the coherences of the three orthogonal turbulent components are necessary for a 3D cable-membrane structure, but the modified wind pressure force can be ignored.

In consideration of construction correlation between segments in cantilever erection practice of bridges, Gaussian random process model is introduced in this paper to represent the distribution of segmental parameters. Subsequently, through system rebuilding and maximum a posteriori criteria, a Bayesian formula is derived for sequentially estimating the mean values and standard deviations of random segmental parameters. It is shown that both the common-used least square estimation method and the Kalman filter estimation method are special cases of the proposed method. Finally, application of the proposed method in a long-span PC cable-stayed bridge for estimating self-weights of segmental girders is presented, and effectiveness of the proposed method in updating and predicting parameters shown.

In the application of the standard Levenberg-Marquardt training process of neural networks, error oscillations are frequently observed and they usually aggravate on approaching the required accuracy. In this paper, a modified Levenberg-Marquardt method based on variable decay rate in each iteration is proposed in order to reduce such error oscillations. Through a certain variation of the decay rate, the time required for training of neural networks is cut down to less than half of that required in the standard Levenberg-Marquardt method. Several numerical examples are given to show the effectiveness of the proposed method.