System Reliability Research On Long-Span Steel Cable-Stayed Bridges In Construction And Service Processes

With the flourishing development of long-span cable-stayed bridges in recent years, the structural components of the bridges become more large-scale, complex and slender. The increase in scales has raised greater challenges to structural safety, and also makes the bridge system reliability issues in construction and service periods more prominent. Aiming at the inherent difficulty in theory and practice of system reliability analysis, this dissertation takes the system limit state surface issue as the starting point of research. By the introduction of support vector classification (SVC) algorithm and Latin Hypercube Sampling (LHS) migration strategy, a new system reliability evaluation method based on SVC migration sampling is developed. Togethor with this and further investigation of various reliability index calculation methods (e.g. FORM, SORM and MCS), the system reliability theory is improved. Applying all the aforementioned methods and obtained random variable models, a system reliability research on long-span steel cable-stayed bridge is carried out, the results profoundly reveal the evolution law and characteristics of system reliability in different construction and service phases. The main aspects of this dissertation are listed as follows:1. Further research on structural reliability index evaluation algorithm. Aiming at the convergence and efficiency drawbacks of traditional reliability index evaluation algorithm, the fundamental mechanism of various algorithms are investigated. Based on the findings, a search direction modification algorithm for FORM is developed. Thereafter, the line sampling Monte Carlo method is modified and improved to adapt to the system reliability evaluation. Finally, the numerical sampling and regression analysis methods are introduced to analyze the second-order expansion of the modified performance function, which ultimately leads to a new regression analysis-prediction algorithm of SORM. The numerical examples validate that the presented algorithms possess good superiority and effectiveness.2. Research on structural system reliability evaluation algorithm based on SVC migration sampling. According to the inherent defects in bounding, accuracy and efficiency of traditional system reliability theory, a concept of system limit state surface is herein introduced. Employing the SVC algorithm to divide the safe and failure domains above, the reconstitution of system limit state surface can then be achieved. Taking all the aforementioned ideas as basis, a new system reliability evaluation algorithm is developed via a combination of the sampling strategy of LHS migration and the SVC algorithm. In order to testify the validity of proposed algorithm, two widely used illustrative examples are introduced and analyzed by different methods for comparison. The results show that the present algorithm has obvious advantages in accuracy, efficiency and stability, which lays a solid foundation for the practical application to the actual structure assessment.3. Research on load and resistance probabilistic models for long-span cable-stayed bridges with steel girder. In order to obtain the load and resistance models for bridge reliability assessment, a large amount of survey and data collection work is carried out. Based on the girder segment weighing result and other existing data, the dead load models are established firstly. Thereafter, the bridge live load models in construction and service periods are obtained by massive statistical survey. Finally, with a comprehensive summary of existing research, the pylon and girder resistance models for long-span steel cable-stayed bridges are formulated, which provides a favorable basis for the following bridge system reliability assessment.4. Research on parallel wire stay-cable resistance model. According to the composition and mechanism of parallel wire stay-cables, the statistical theory as well as wire strength test results is employed to investigate the law of wire tensile resistance distribution. Based on this, the cable parallel system (Daniels model) and Monte Carlo sampling method are adopted to develop the Daniels model based resistance algorithm for parallel wire cables. Using this algorithm and component reliability analysis method, the cable reliability research is finally carried out.5. System reliability assessment for the construction and service periods of long-span cable-stayed bridges with steel box girder. According to the established load and resistance models, a typical super long-span cable-stayed bridge——Sutong bridge is deeply investigated by the use of SVC migration sampling algorithm. The system reliability indexes as well as primary failure modes in 8 working conditions during construction and service periods are systematically studied and discussed, which comprehensively reveals the evolution law of bridge system reliability in different key phases.