节点文献
斜拉双层柱面网壳结构可靠度计算及设计方法探讨
Reliability Calculation and Discussion of Design Method of Cable-Stayed Double-Layer Cylindrical Latticed Shells
【作者】 蒋友宝;
【作者基本信息】 东南大学 , 结构工程, 2006, 博士
【摘要】 斜拉网格结构是斜拉桥技术和预应力技术综合应用到空间结构而形成的一种形式新颖的大跨度空间结构体系,以其良好的受力性能和新颖的建筑造型适应现代建筑发展的要求,具有广阔的应用前景。本文以斜拉双层柱面网壳结构为研究对象,对可靠度计算方法、风致响应分析、敏感性分析和基于可靠性的设计方法等方面进行了研究。针对经典的一次二阶矩法在计算可靠指标时可能不收敛的不足,提出了一种线性可行方向算法。该算法采用求解约束极值问题的可行方向策略,在迭代过程中综合考虑了目标函数和极限状态曲面方程的影响。实例验证具有较高的效率和精度。针对SORM方法在求解可靠指标时存在着对设计点求解精度较敏感的不足,本文提出了基于失效曲面样本点修正几何可靠指标的方法。该方法考虑了失效曲面的几何特性和响应面法的逼近误差。算例分析表明该方法对设计点的求解精度不敏感,且能得到较高精度的分析结果。在前人研究的基础上,本文提出了两种改进的响应面方法:状态空间响应面方法和极限状态响应面方法。它们为在一般情况下(无需用到功能函数值)如何应用响应面方法分析结构可靠度进行了有益的探索。模式识别技术和自适应技术的引入提高了可靠度分析的计算效率。算例分析结果表明,两种方法具有较高的精度,同时减少了有限元分析的次数,非常适合于大跨空间结构的可靠度分析。基于AR法和谐波合成法原理,本文编写了模拟多点互相关脉动风速时程的程序。通过算例模拟谱与目标谱的比较,验证了编制的模拟脉动风速程序的正确性。比较模拟结果可以得出:AR法模拟速度快,但精度稍差一些,而谐波叠加法虽耗费较多的机时,但模拟精度要比AR法好。然后采用模拟得到的风速时程曲线对斜拉双层柱面网壳结构进行了风致响应分析。对斜拉双层柱面网壳结构考虑位移失效和承载能力失效的可靠度分析表明:采用FORM法计算可靠指标具有较高的精度。随后采用FORM法翔实讨论了各参数变化时两种失效模式可靠指标的变化情况。接着,对斜拉双层柱面网壳结构进行了可靠度对随机变量的敏感性分析。敏感性分析包含两方面的内容:分布参数的敏感性和极限状态方程参数的敏感性。对极限状态方程参数的敏感性分析表明:相对位移失效模式来说,为使可靠指标的计算结果具有较高的精度,承载能力失效模式需考虑较多数目的随机变量。另外提出了可靠指标对随机变量标准值的敏感性系数的概念。它能考虑均值和标准差等比变化时,随机变量对可靠指标的影响程度。最后本文探讨了斜拉双层柱面网壳结构基于整体可靠度的设计方法。首先讨论了结构目标可靠指标的确定依据,然后在此基础上提出一种实用的极限承载能力设计方法。它通过在整体层次上改进设计表达式中的抗力和荷载分项系数,以最大可能地实现与目标可靠指标相一致。这些研究成果为斜拉双层柱面网壳结构的设计提供了参考资料。
【Abstract】 Cable-stayed grid structure is an original long-span spatial structure with modern prestressing techniques and cable-stayed bridge techniques. With good structural performance and original architectural shape, it can meet the requirements of modern architecture and is suit for modern sports, public and industrial building.The research work involves reliability calculation method, wind-induced analysis, sensitive analysis and reliability-based design method of the cable-stayed double-layer cylindrical latticed shell. A linear feasible direction algorithm is proposed to calculate reliability index. It overcomes the problem that the calculation of reliability index is probably not converged with the first order reliability method (FORM) when limit state equation is nonlinear. Based on the feasible direction strategy, the effect of objective function and limit state equation on calculation of reliability index is considered in the iteration of the algorithm. Numerical examples show that the algorithm is of higher precision and efficiency For the reliability index result with the second order reliability method (SORM) is sensitive to the design point solution, a method is proposed to improve reliability index based on the sampling point on the real limit state surface, which considers the geometrical characteristic of the limit state surface and approximate inaccuracy of the response surface method. Numerical examples show that the method is of high precision and not sensitive to the design point solution.Based on the previous research achievement, two improved response surface methods are proposed, called the state-space response surface method and the limit state response surface method. They provide a preliminary useful study for general case that how to apply the response surface method to analyze structural reliability problem when the performace function value is not used. The application of pattern recognition techniques and adaptive sampling techniques improves the reliability analysis efficiency. Numerical examples show that the two methods are of high precision with fewer finite element analysis times and suitable for the reliability analysis of the long-span spatial structures.Based on the auto-regressive (AR) method and the harmony superposition method, the programme on multi-nodal wind velocity time series simulation is compiled. Through the comparison of stimulant power spectrum and target power spectrum, the validation of the programme is proved. From the comparison it shows that the AR method simulates wind velocity fast but imprecisely, and the harmony superposition method simulates wind velocity accurately but time-consumingly. Then the wind-induced analysis is performed with the stimulant wind velocity time series.