结构可靠性分析的自适应子集模拟方法

结构可靠性分析需要精确高效的失效概率计算方法。为解决高维非线性可靠性分析问题中的失效概率计算问题,本文提出了两种以失效概率估计精度为停机控制参数的自适应子集模拟方法。理论分析和数值算例表明:(1) 两种自适应子集模拟方法能根据失效概率的估计精度要求自适应调整样本量;(2) 考虑样本量优化的自适应子集模拟方法能进一步减少总样本量,提高计算效率。本文所提方法为研究者对结构进行精确高效的可靠性分析提供了一条可行途径。     

基于子集模拟法的风机叶片可靠度分析

风机叶片是风机机组中最重要的组件之一,其可靠性对整个发电系统的健康运行十分关键。传统的蒙特卡罗法可靠度计算效率较低,本文将子集模拟法应用到风机叶片的可靠度计算中,通过合理选取中间失效事件临界值和建议概率密度函数,采用Metropolis算法模拟产生一系列的样本,极大地提高了抽样效率。结果表明,子集模拟法在风机叶片可靠度计算中精度较高,且结构的失效概率越小,相比于传统蒙特卡罗法效率更高,可以极大地减轻计算压力。     

基于子集模拟和重要抽样的可靠性灵敏度分析方法

针对工程实际中大量存在的小失效概率问题,提出了基于子集模拟和重要抽样的可靠性灵敏度分析方法. 在子集模拟重要抽样可靠性分析方法中,通过引入合理的中间失效事件,将小的失效概率表达为一系列较大的条件失效概率的乘积,而较大的条件失效概率则可通过构造中间失效事件的重要抽样密度函数来高效求解. 基于子集模拟重要抽样可靠性分析的思想,论文将可靠性灵敏度转化为条件失效概率对基本变量分布参数的偏导数形式,推导了基于子集模拟和重要抽样的可靠性灵敏度估计值及估计值方差的计算公式,并采用算例对所提方法进行了验证. 算例结果表明所提方法具有较高的计算精度和效率,并且适用单个和多个失效模式系统.      

基于不同MCMC抽样子集模拟的模型修正方法对比研究

子集模拟方法作为结构可靠度分析方法,也可应用于工程优化问题,诸如优化设计、模型修正等．为研究基于不同蒙特卡洛马尔可夫链(MonteCarloMarkovChain,MCMC)抽样的子集模拟优化方法(Subset Simulation Optimization, SSO),以有限元模型修正作为优化背景问题,开展其精度和效率的对比研究．介绍标准SSO和子集模拟(Subset Simulation, SS)常见的MCMC抽样方法,并基于上述不同MCMC抽样的SSO开展某局部损伤悬臂梁（10维变量）的有限元模型修正,修正结果与基于遗传算法(Genetic Algorithm, GA)的模型修正方法进行对比;而后将上述不同MCMC抽样的SSO修正方法应用于某四层钢框架有限元模型修正中（11维变量）．结果表明,采用随机游走的延迟拒绝修正M-H方法(MMH algorithm with Delayed Rejection, MMHDR)和自适应条件抽样方法(Adaptive Conditional Sampling, ACS)的SSO有限元模型修正具有较好的精度和效率,在工程结构有限元模型修正中更具...     

基于改进蛙跳算法的结构可靠指标计算

为了克服传统的结构可靠指标求解方法存在的模型描述和计算缺陷,引入改进混合蛙跳算法用于求解可靠指标。针对标准蛙跳算法求解组合优化问题时后期收敛速度慢、容易陷入局部最优的缺点,新算法引入蚁群算法的信息素调整策略,对蛙跳算法后期局部搜索进行改进,扩大算法后期搜索范围,使蛙跳算法跳出局部最优。最后,将改进后的算法应用到结构可靠指标的求解过程,算例验证了改进算法的有效性。     

高速流场气动光学RANS/DSMC混合数值模拟算法研究

提出了适用于高速流场气动光学数值模拟的RANS/DSMC混合算法.通过RANS对全局时均流场进行数值模拟,再对其中局部流场的脉动量采用DSMC进行数值模拟,以Maxwellian速度分布实现RANS宏观量信息向DSMC微观量信息的传递.采用超声速环境下尖劈模型对该混合算法进行校验,对比试验结果论证了算法的正确性.在计算...     

OpenSees 混合模拟试验技术发展与应用

将现代结构地震性能模拟与试验分为整体结构模拟试验和子结构模拟试验两类. 在此基础之上详细 讨论了混合模拟试验的基本概念背景及其应用与发展现状. 结合面相对象有限元程序OpenSees 的高层程序 架构和并行计算程序架构, 详细论述了基于OpenSees 的4 种混合模拟试验系统, 并简要介绍了多尺度一体化 混合建模的概念与应用. 证明基于商业或非商业有限元软件平台开展混合模拟试验是可行的; 混合模拟试验 所要处理的核心问题是如何将不同的有限元软件平台以及各个试验站点的设备联系起来, 并实现相互之间的 协同工作. 混合模拟试验在强震作用下结构连续倒塌问题的研究中具有一定的优势. 指出混合模拟试验这一 新技术的实质性发展离不开目前所强调的多学科之间的无缝融合.      

非结构混合网格在气动热标模模拟中的应用

非结构混合网格消除了结构网格节点的结构性限制,可以较好地处理边界,同时兼顾了粘性边界层模拟的需求,具有灵活性大、对复杂外形适应能力强和生成耗时短等优点,在飞行器气动特性模拟中得到广泛应用.本文针对非结构混合网格的特点,把前期针对非结构混合网格气动力高精度模拟发展改进的梯度计算方法和Roe格式熵修正方法推广应用到气动热流的数值模拟.以典型钝锥标模外形的高超声速绕流为研究对象,开展了不同网格形式和第一层网格不同间距的影响研究.结果 表明,热流计算时,头部物面网格最好采用四边形或四边形交叉剖分得到的三角形网格,物面法向的网格雷诺数取20左右,为热流计算时非结构混合网格的生成提供了指导,同时验证了计算方法的有效性和可靠性.     

波浪破碎的一种混合湍流模拟模式

近岸波浪的变形与破碎,一方面影响水体标识码运输,另一方面对消波护岸具有指导意义.本文提出一种三维混合湍流模拟模式方法,将求解区域分为造波区、波浪传播区和消波区. 造波区采用层流模式,通过基于 Fluent 的二次开发的 UDF 方法在边界进行速度造波.这种方法在给定入口速度的条件标识码据已知波高进行 精准插值,从而控制水的体积分数. 在波浪传播区域,采用大涡标识码行模拟研究,在消波区,采用 RANS 模型并利用多孔介质消波法进行消波. 模标识码 VOF 方法捕捉波浪破碎过程中的自由面变化.本文对波高为 5.5cm 的规则波 (M1)、波高为 13.5cm 的 规则波 (M3)、有效标识码 7.75cm 的 TMA 谱单向不规则波 (U1) 和有效波高为 19cm 的 TMA 谱多向不规则波 (B5) 展开标识码研究,并与前人的相关标识码果作比较,各条件下模拟结果与实验结果吻合.模拟结果说明本文提出的模型能够准确模拟出波浪传播过程中的折射和标识码象,并且能够捕捉的波浪破碎过程中的自由面变化,为三维波浪的传播与破碎的数值模拟提供一种模拟方法.      

气体化学反应流动的DSMC／EPSM混合算法研究

发展了平衡粒子模拟方法(EPSM)，建立了与高温气体化学反应动力学理论相匹配的：EPSM耦合模型，并通过混合参数进行流区的自动识别，将：EPSM方法与蒙特卡罗直接模拟方法(OSMC)结合，构造了可模拟化学反应流动的DSMC／EPSM混合算法。应用该算法对汲及化学反应的二维高超音速竖板绕流流场进行模拟，将结果与DSMC方法的结果进行比较，验证了新算法对求解化学反应流动的可行性。将混合算法的计算效率与DSMC方法的计算效率进行比较，发现混合算法能够大大提高计算效率。     

结构动力可靠度的重要抽样法

重要抽样法是蒙特卡洛数值模拟方法中的一种重要的方差缩减技术,目前重要抽样法在工程结构可靠度计算中的应用主要集中于静力问题。本文分析了动力可靠度蒙特卡洛方法的特点,提出了在结构动力可靠度问题中应用重要抽样法的方法,并针对白噪声荷载,给出了选择重要抽样函数的方法和重要抽样函数的具体表达式。理论和数值分析表明,本文所提出的重要抽样法应用于结构动力可靠度计算是可行的。     

一种用于蠕变损伤失效概率计算的分层抽样方法

根据蠕变损伤试验和相关的研究资料,确定出影响材料蠕变损伤的随机参数;建立了蠕变损伤失效的概率模型.其次根据分层抽样法的基本思想,建立了一种考虑对蠕变损伤失效概率区间分情况抽样的蒙特卡洛分层抽样方法.并且在蠕变损伤模型中选择等效应力作为一个分层抽样变量.随后根据工程实际情况,提出了一种在统计意义下确定抽样变量的安全区域和失效区域的分层抽样的策略,以减少抽样次数.并推导出确定分层抽样区间的计算方法.然后对选择区间大小进行了讨论,可知如果所选择的分层抽样区间太小,会漏掉许多导致蠕变损伤失效的等效应力抽样值,从而使计算结果误差增大;但是如果分层抽样区间选择太大,会增加无效抽样次数,因而降低抽样效率.最后的算例表明了该方法在同样的计算精度下,其计算效率比直接抽样法的效率要高.     

结构随机分析的Monte Carlo加权残值法

本文提出一种结构随机分析的Monte Carlo加权残值法。文中建立了这种方法的基本列式,并通过静力挠度、固有频率和屈曲荷载等算例,表明本文方法理论简捷,计算工程量少,精度较高,是随机结构数值分析的有效方法。     

Spectral simulation and shock absorber identification

In vehicle dynamics shock absorbers are used for the optimization of driving comfort and driving safety. Therefore, it is necessary to identify characteristics of shock absorbers under real conditions. This paper introduces the use of hardware-in-the-loop simulations for the identification of shock absorbers involving stochastic models of the road roughness. For this purpose a dynamic hydraulic test stand is used replacing the classical mechanical test stands which allow only sinusoidal excitation. For the Monte Carlo simulation with a real shock absorber in the loop, the random excitation of ground roughness is generated using a modified spectral representation method based on the famous contributions of Shinozuka. Motion and force of the shock absorber are measured and fed back to the Monte Carlo simulation of a car model in real time. The characteristic of the shock absorber is identified using the classical least squares method and a correlation-based method. A piecewise linear model for the characteristic relating the damping force and the velocity of the piston is applied for the shock absorber identification.     

Eulerian-Lagranigan simulation of aerosol evolution in turbulent mixing layer

The formation and evolution of aerosol in turbulent flows are ubiquitous in both industrial processes and nature. The intricate interaction of turbulent mixing and aerosol evolution in a canonical turbulent mixing layer was investigated by a direct numerical simulation (DNS) in a recent study (Zhou, K., Attili, A., Alshaarawi, A., and Bisetti, F. Simulation of aerosol nucleation and growth in a turbulent mixing layer. Physics of Fluids, 26, 065106 (2014)). In this work, Monte Carlo (MC) simulation of aerosol evolution is carried out along Lagrangian trajectories obtained in the previous simulation, in order to quantify the error of the moment method used in the previous simulation. Moreover, the particle size distribution (PSD), not available in the previous works, is also investigated. Along a fluid parcel moving through the turbulent flow, temperature and vapor concentration exhibit complex fluctuations, triggering complicate aerosol processes and rendering complex PSD. However, the mean PSD is found to be bi-modal in most of the mixing layer except that a tri-modal distribution is found in the turbulent transition region. The simulated PSDs agree with the experiment observations available in the literature. A different explanation on the formation of such PSDs is provided.     

三维非结构网格DSMC方法的实现及其应用

研究了三维非结构网格DSMC方法实现的过程。将Bird位置元方案中的子网格思想引入到非结构网格上来,只存储子网格的总体标识号,利用较少的计算网格提高了分子的分辨率与计算精度。提出了将体积元坐标搜索算法与交替数字二叉树搜索算法(ADT)相结合的方法来跟踪模拟分子在网格之间的迁移,使用ADT方法判别分子与物面是否作用,避免了分子表面反射的非确定论判据。利用Fortran 90的动态分配内存技术编制了通用计算程序。最后对高超声速过渡流域航天飞机头部外形绕流进行了数值模拟,数值结果初步验证了算法的可行性。     

Monte Carlo simulation of self-avoiding lattice chains subject to oscillatory shear flow in polymer solution

 This paper has introduced a pseudo-potential in bond-fluctuation model to simulate oscillatory shear flow of multiple self-avoiding chains in three dimensions following our previous work under simple shear flow. The oscillatory flow field was reasonably reproduced by lattice Monte Carlo simulation using this pseudo-potential neglecting hydrodynamic interaction. By sampling the configuration distribution functions, the macroscopic viscoelasticity of semi-concentrated polymer solution was determined. Both Newtonian and non-Newtonian regimes were studied. The complex modulus and dynamic viscosity exhibit a reasonable power relation with oscillatory frequency, which is consistent with present theories and experiments. Consequently, lattice Monte Carlo simulation has been extended to model free-draining self-avoiding multi chains subject to oscillatory shear flow and to investigate associated viscoelasticity on the molecular level. Received: 1 October 1999 Accepted: 19 October 1999     

Wave scattering method for dynamic response analysis of structures with stochastic parameters

The wave scattering method is presented to analyze dynamic response of frameworks with stochastic parameters. First, with the uncertain physical, geometric, and loading properties in consideration, the stochastic waveguide equations containing the axial, torsional and flexural wave modes are established. Second, the stochastic wave scattering equation and wave translation matrix are derived to obtain the wave modes. Third, the methodology to extract the generalized displacements and forces from stochastic wave modes is proposed. Finally, a cantilever beam, a planar framework, and a space framework have been presented as numerical examples to illustrate the e?ciency of the proposed method. It is found that the results obtained by the proposed method with higher computational e?ciency show an excellent agreement with those by Monte Carlo simulation method. Furthermore, the influences of stochastic parameters on dynamic response are revealed.     

STOCHASTIC ALGORITHM AND NUMERICAL SIMULATION FOR DROP SCAVENGING OF AEROSOLS

The time evolution of aerosol size distribution during precipitation, which is founded mathematically by general dynamic equation (GDE) for wet removal, describes quantitatively the process of aerosol wet scavenging. The equation depends on aerosol size distribution, raindrop size distribution and the complicated model of scavenging coefficient which is induced by taking account of the important wet removal mechanisms such as Brownian diffusion, interception and inertial impaction. Normal numerical methods can hardly solve GDE, which is a typical partially integro-differential equation. A new multi-Monte Carlo method was introduced to solve GDE for wet removal, and then was used to simulate the wet scavenging of aerosols in the real atmospheric environment. The results of numerical simulation show that, the smaller lognormal raindrop size distribution and lognormal initial aerosol size distribution, the smaller geometric mean diameter or geometric standard deviation of raindrops can help scavenge small aerosols and intermediate size aerosols better, though large aerosols are prevented from being collected in some ways.