Study on System Reliability of Key Structures of Long-Distance Water Transfer Project
|Keywords||System Reliability Failure Hypersurface Important Sampling SVM RSM Aqueduct Slope Genetic Algorithms Poisson Stochastic Process|
Many uncertainties are inherent in engineering practices. It is significant to quantify the influence of those uncertainties on safety of engineering projects. The long-distance water transfer project is of a large scale and a long route, and is affected by more complex uncertainties, so more attention is paid to its safety. In a point of macroview, the long-distance water transfer project is a series system which mainly consists of trenches and constructions, therefore safety of key structures of the series system has important influence on the safety of the whole system. System reliability of key structures of long-distance water transfer project is analysised based on the middle route of South-North Water Transfer Project (SNWTP). Aqueduct has a larger size and a more complex structure among the constructions of the middle route of SNWTP, and the water is transferred mainly by trenches, so the system safety of the large aqueducts and the slope stability of trenches reflect the safety level of the whole long-distance water transfer project at a certain extent. As a key problem of the system safety of long-distance water transfer projects, the system reliability analysis of the large aqueducts and the slope stability should be solved. In this paper, three aspects were studied: the fundamental theory of reliability analysis, the system reliability assessment method of aqueduct structures, the system reliability assessment method of slope stability of trenches. The major contents are summarized as follows:(1) A partly analytical complex Monte-Carlo method in structural failure hypersurface is proposed. Samples are mapped to the structural failure hypersurface by separating a random variable or an expression of the limit state function and evaluating the cumulative distribution function of it analytically, then the structural failure probability is estimated in combination with importance sampling method. Approaches for the original X-space and the rotated orthogonal normal V-space are presented respectively. Samples are mapped to the hyperplane of structural failure, so the availability of sampling is guaranteed and the proposed method is more suitable for highly nonlinear limit state functions. The dimension of sampling may be reduced and a more accurate and stable result is achieved in virtue of adopting analyticalresolution. This method has been proved to be more efficient by theoretical analysis and numerical examples.(2) Response Surface Method (RSM) can deal with the issue with implicit performance functions well for its famous accuracy and good efficiency. The key problem of RSM is the reconstruction of the response surface. The amount of sample set used to reconstruct the response surface usually is small. According to the excellent learning and generalization ability of support vector machine (SVM) even with small samples, the response surface of the structure is reconstructed by SVM, and then a new RSM based on SVM for structural reliability analysis is proposed. Furthermore, an improved RSM based on SVM is developed. A new response surface and a new design point are achieved by adding the old design point to sample set in iterative process of the improved method, so no any result of FEM calculation is abnegated and the amount of FEM calculations is reduced dramatically. It is proved by examples that the calculation efficiencies and accuracy can been increased by proposed methods.(3) Cracks of aqueduct structures could decrease the durability and safety of the aqueduct. Sometimes, the serviceability limit state may be as the control condition in design. A three-dimensional finite element model of a large prestressed aqueduct is constructed to calculate the maximal tensile stresses of selected mainly components, and reliabilities of crack resistance of the selected main components are calculated by SVM response surface method and the corresponding sensitivity analyses are performed. Then the correlation coefficients between failure modes are evaluated and system reliability of crack resistance is computed by wide bounds method and Ditlevsen bounds method. At last, the deflection reliability of aqueduct is calculated by SVM response surface method.(4) Based on local average method of random field, Discrete spatial average variance functions of soil parameter over two-dimensional slip arc and three-dimensional slip surface are developed. The proposed methods are simple and suitable to characteristics of slices method of slope stability analysis, and can take the effect of spatial correlation of soil parameters on reliability of slope stability into consideration efficiently. Influence of spatial relative distance of soil parameters on the reliability of slope stability is investigated.(5) Actually there should be a critical three-dimensional slip surfaces which give the highest failure probability if spatial variabilities of soil parameters are taken into consideration. The existent reasons of the minimal three-dimensional reliability index and the corresponding critical slip surface are investigated and illustrated by examples. This makes it possible to get three-dimensional reliability index of soil slope by means of optimization techniques. Therefore, a model of spatial variability of soil parameters is adopted and Genetic Algorithm (GA) method is used to search for the reliability index directly, then three-dimensional reliability analysis method for slope stability, based on GA method, is proposed. The minimal three-dimensional reliability index and the corresponding three-dimensional slip surface are achieved by the proposed method.(6) A new system reliability method of a given length slope is proposed based on Poisson stochastic process theory and by measuring the whole length of the slope with the width of three-dimensional failing soil mass. The system reliability of an example trench slope of middle route of SWTP is assessed.(7) Computer program for SVM reliability analysis method and program for three-dimensional reliability analysis of slope are developed in this paper.This research work is sponsored by Technical Innovation Fund of the Ministry of Water Resources of PRC.