Bredow Rafał, Kamiński Marcin
Department of Structural Mechanics, Lodz University of Technology, 90-924 Lodz, Poland.
Materials (Basel). 2022 May 18;15(10):3587. doi: 10.3390/ma15103587.
This work aimed to analyze the reliability of a steel hall that was recently erected in central Poland subjected to dynamic wind excitation using the stochastic finite element method. Reliability analysis was completed using the relative entropy concept delivered by Bhattacharyya and contrasted with the first-order reliability method recommended by the engineering design codes. Bhattacharyya probabilistic relative entropy was additionally rescaled in this study to fit the demands and recommended admissibility intervals given in Eurocode 0. The finite element method study was carried out thanks to a discrete model created in the system ABAQUS 2019, while all further statistical and probabilistic computations were programmed and completed in the symbolic environment of MAPLE 2019. Contrary to most engineering analyses in steel structure areas, this study included the important warping effect while designing the hall ridges and the purlins. Dynamic structural responses were determined via the Hilber-Hughes-Taylor algorithm and their series were numerically obtained for a series of input uncertainty parameters representing several mechanical and environmental quantities. The generalized 10th order iterative stochastic perturbation technique was contrasted in this context with statistical estimators from the Monte Carlo simulations and numerical integration resulting from the semi-analytical approach. The key research finding of this study was an extremely good coincidence between the FORM indices and the rescaled relative probabilistic entropies for the given stochastic excitations, which additionally did not depend on a choice of one of the three proposed numerical approaches.
这项工作旨在使用随机有限元方法分析波兰中部一座最近建成的钢屋在动态风激励下的可靠性。可靠性分析是利用Bhattacharyya提出的相对熵概念完成的,并与工程设计规范推荐的一阶可靠性方法进行对比。在本研究中,Bhattacharyya概率相对熵还进行了重新缩放,以符合欧洲规范0给出的要求和推荐的容许区间。有限元方法研究是借助在ABAQUS 2019系统中创建的离散模型进行的,而所有进一步的统计和概率计算是在MAPLE 2019的符号环境中编程并完成的。与钢结构领域的大多数工程分析不同,本研究在设计屋架和檩条时考虑了重要的翘曲效应。通过Hilber-Hughes-Taylor算法确定结构的动态响应,并针对一系列表示多个机械和环境量的输入不确定性参数,数值获得其响应序列。在此背景下,将广义十阶迭代随机摄动技术与蒙特卡罗模拟的统计估计器以及半解析方法产生的数值积分进行了对比。本研究的关键研究发现是,对于给定的随机激励,FORM指标与重新缩放后的相对概率熵之间存在极佳的一致性,而且这并不取决于所提出的三种数值方法中的哪一种。