几何非线性样条有限元法

根据样条有限元方法的基本原理,拓展建立了一种几何非线性样条有限元方法.该方法充分利用了样条函数乘积的性质,将样条函数的积分显式地表达出来,使得传统有限元中需要花费大量时间的刚度矩阵计算可以很简单的得到,具有精度高、输入量小、连续性强等优点.它与普通有限元法使用高斯积分计算的精确结果均在高斯积分点上不同,样条有限元所得到...     

非线性精细积分方法及其在拟动力试验中的应用

将精细积分方法和预估-校正Adams-Bashforth-Moulton多步法相结合,构造了一种避免状态矩阵求逆、隐式预估-校正、四阶精度的精细积分多步法,可用于多自由度结构体系的非线性地震反应分析。基于精细积分多步法,构造了一种实用的显式拟动力试验数值积分方法,该方法在成倍地增大时间步长后的计算精度比中心差分法高,稳定性较好,试验工作可大量减少。最后,将本文显式方法应用于组合筒体结构拟动力试验中。     

结构地震反应分析的Newmark精细耦合级数显式方法

将Newmark-β法中常平均加速度法的基本假定与精细积分法耦合,对积分项的计算引入指数矩阵的Taylor级数展开式,提出了动力方程的显式耦合级数解,设计了相应的时程积分算法,并应用在结构的地震反应分析中。分析表明,由于该方法是显式的,在质量矩阵为对角矩阵时,不需要计算耦联的方程组,因此可以有效地减少内存占用和机时耗费。该方法的稳定性条件显然满足,其精度可根据Taylor级数展开式的截断阶数进行灵活控制。算例表明该方法对地震作用的有效性和适应性。     

改进的结构系统可靠度的条件边缘乘积法

分析了结构系统可靠度计算与多维正态积分之间的转化关系, 以及原有的用于多维正态积分计算的条件边缘乘积法的不足之处, 从条件正态分位数计算和相关系数矩阵修正两方面给出改进措施, 形成一种改进的条件边缘乘积法。该方法采用数值积分方法进行二维正态分布函数的计算, 提高原有方法中条件正态分位数计算的精度, 并在此基础上对各相关系数矩阵进行修正, 从而有效提高了原有方法在串联系统可靠度计算中的精度。同时, 将这一改进方法与单失效模式下的结构可靠度显式迭代算法相结合, 提出从单元到系统的结构系统可靠度分析流程。数值算例表明了这一改进方法的有效性和实用性。     

一种工程实用的数值积分方法

在实际工程结构动力反应分析中,往往由于结构型式十分复杂,常用的两种直接积分方法,即显式积分方法和隐式积分方法,在使用中都存在着一定的局限性,如何将这两种积分方法合理有效地结合起来,是一个十分有意义的研究课题。针对实际工程问题中整体结构计算时间步长的选择往往受局部区域的材料特性、尺寸大小等因素影响的这一现象,提出了一种对结构局部区域进行隐式积分、对其余区域进行显式积分的工程实用的显隐式数值积分方法,这种积分格式相对于显式积分格式而言,能显著提高整体结构的计算速度。最后采用三个数值计算实例对这一方法进行验证。     

二维压电材料动强度因子的扩展有限元计算

采用扩展有限元求解二维弹性压电材料动断裂问题。扩展有限元的网格独立于裂纹,因此网格生成可大大地简化,且裂纹扩展时不需重构网格。采用相互作用积分技术计算动强度因子。比较了标准的力裂尖加强函数和力-电裂尖加强函数对动强度因子的影响,结果表明标准的力裂尖加强函数能有效地分析压电材料动断裂问题。分析了极化方向对动强度因子的影响。数值分析表明采用扩展有限元获得的动强度因子与其他数值方法解吻合得很好。     

工程结构非线性动力响应数值分析中零能量模态控制对响应的影响

在工程结构非线性动力响应数值分析中,通常采用显式积分格式。这样可避免大量的刚度矩阵求逆叠代运算。但显式积分有条件稳定的,从而时间步长△t的取值受到限制,这样必导致进行大量的小时间步长循环计算。为了减少计算量,通常采用高斯积分一类的数值积分。单点高斯积分的采用可引起hourglass模态。控制hourglass模态是结构非线性动力响应数值分析中必须解决的问题之一。本文以八节点六面体实体单元单点高斯积分为例详细分析了hourglass模态产生的机理,并在此基础上讨论了hourglass模态的控制措施及具体应用。通过数值算例分析发现hourglass模态控制的采用对正常动力响应模态造成一定的影响。     

基于三次样条插值的精细积分法

结合指数矩阵的精细算法,提出了一类基于三次样条插值的精细积分方法。针对结构动力学方程一般解中的积分项,考虑在一个时间步长内激励为线性和正余弦两种变化形式,通过对积分项中的指数矩阵进行三次样条插值函数模拟,得到一组新的被积函数,最后通过多次分部积分,构造了一类新的高精度计算格式。在三次样条插值函数构造过程中引入了指数矩阵的精细算法,有效避免了中间过程中有效数字的丢失,同时还有效解决了HPD-F算法中涉及的矩阵求逆问题,大大增加了算法的数值稳定性。数值算例显示了该方法的有效性。     

波谱单元法在空间桁架地震响应分析中的应用

摘 要：针对传统波谱单元法（SEM）只能用于求解节点集中荷载作用下结构动力响应问题的不足,提出了一种通过计算地震等效波谱节点荷载求解桁架结构地震响应的方法。基于虚功原理,利用波谱形函数积分得到地震等效波谱节点荷载的显式表达式,通过修改波谱单元法中单元刚度矩阵的波数,考虑了阻尼对结构动力特征的影响,采用数值拉普拉斯（Laplace）变换替换快速傅立叶（FFT）变换,回避了传统波谱单元法中FFT的周期性问题。利用地震荷载等效后的波谱节点荷载对三维空间桁架结构进行地震响应分析,结果表明,采用本文的方法能方便的计算桁架结构的地震等效波谱节点荷载,精确求解结构的地震响应,与传统有限元法（FEM）相比,大大减少计算单元数量,提高计算精度,且便于编程计算。     

基于Wilson-θ显式算法的动载荷快速定位及重建方法EI北大核心CSCD

通过将Wilson-θ数值算法转变为一种等效的显式表达式,由此建立了一种基于Wilson-θ显式算法的动载荷识别方法。在选取合适的参数θ值以后,该算法是无条件稳定的,并且避免了以往隐式识别算法中由于递推计算产生误差累积导致识别结果严重发散的问题。通过引入响应系数矩阵,该显式算法可以灵活地选取响应类型用于载荷识别计算。同时结合分离变量方法,将载荷位置信息从建立的模型矩阵中提取出来,由此减少矩阵求逆的次数,提高载荷定位计算的效率。结合简支梁的仿真算例与试验测试对该载荷识别方法进行验证。研究结果表明,基于Wilson-θ显式算法的动载荷快速定位及重建方法能够有效地识别得到载荷位置及相应的时间历程。相对于传统载荷识别方法,该方法计算结果的准确性更好,且运算效率更高。     

Time dependent crack tip enrichment for dynamic crack propagation

We study several enrichment strategies for dynamic crack propagation in the context of the extended finite element method and the effect of different directional criteria on the crack path. A new enrichment method with a time dependent enrichment function is proposed. In contrast to previous approaches, it entails only one crack tip enrichment function. Results for stress intensity factors and crack paths for different enrichments and direction criteria are given.     

New crack‐tip elements for XFEM and applications to cohesive cracks

An extended finite element method scheme for a static cohesive crack is developed with a new formulation for elements containing crack tips. This method can treat arbitrary cracks independent of the mesh and crack growth without remeshing. All cracked elements are enriched by the sign function so that no blending of the local partition of unity is required. This method is able to treat the entire crack with only one type of enrichment function, including the elements containing the crack tip. This scheme is applied to linear 3‐node triangular elements and quadratic 6‐node triangular elements. To ensure smooth crack closing of the cohesive crack, the stress projection normal to the crack tip is imposed to be equal to the material strength. The equilibrium equation and the traction condition are solved by the Newton–Raphson method to obtain the nodal displacements and the external load simultaneously. The results obtained by the new extended finite element method are compared to reference solutions and show excellent agreement. Copyright © 2003 John Wiley & Sons, Ltd.     

An XFEM/Spectral element method for dynamic crack propagation

A high-order extended finite element method based on the spectral element method for the simulation of dynamic fracture is developed. The partition of unity for the discontinuous displacement is constructed by employing p order spectral element. This method shows great advantages in the simulations of moving crack and mixed mode crack. The numerical oscillations are effectively suppressed and the accuracy of computed stress intensity factors and crack path are improved markedly. Furthermore the simulation results show that p-refinement is more effective in improving the stress contour near the crack tip than h-refinement. The well known form of the explicit central difference method is used and the critical time step for this method is investigated. We find that by using lumped mass matrix the critical time step Δt _c for this high-order extended finite element is almost independent of the crack position.     

On generation of lumped mass matrices in partition of unity based methods

The partition of unity based methods, such as the extended finite element method and the numerical manifold method, are able to construct global functions that accurately reflect local behaviors through introducing locally defined basis functions beyond polynomials. In the dynamic analysis of cracked bodies using an explicit time integration algorithm, as a result, huge difficulties arise in deriving lumped mass matrices because of the presence of those physically meaningless degrees of freedom associated with those locally defined functions. Observing no spatial derivatives of trial or test functions exist in the virtual work of inertia force, we approximate the virtual work of inertia force in a coarser manner than the virtual work of stresses, where we inversely utilize the ‘from local to global’ skill. The proposed lumped mass matrix is strictly diagonal and can yield the results in agreement with the consistent mass matrix, but has more excellent dynamic property than the latter. Meanwhile, the critical time step of the numerical manifold method equipped with an explicit time integration scheme and the proposed mass lumping scheme does not decrease even if the crack in study approaches the mesh nodes — a very excellent dynamic property. Copyright © 2017 John Wiley & Sons, Ltd.     

Mass lumping strategies for X‐FEM explicit dynamics: Application to crack propagation

This paper deals with the numerical modelling of cracks in the dynamic case using the extended finite element method. More precisely, we are interested in explicit algorithms. We prove that by using a specific lumping technique, the critical time step is exactly the same as if no crack were present. This somewhat improves a previous result for which the critical time step was reduced by a factor of square root of 2 from the case with no crack. The new lumping technique is obtained by using a lumping strategy initially developed to handle elements containing voids. To be precise, the results obtained are valid only when the crack is modelled by the Heaviside enrichment. Note also that the resulting lumped matrix is block diagonal (blocks of size 2 × 2). For constant strain elements (linear simplex elements) the critical time step is not modified when the element is cut. Thanks to the lumped mass matrix, the critical time step never tends to zero. Moreover, the lumping techniques conserve kinetic energy for rigid motions. In addition, tensile stress waves do not propagate through the discontinuity. Hence, the lumping techniques create neither error on kinetic energy conservation for rigid motions nor wave propagation through the crack. Both these techniques will be used in a numerical experiment. Copyright © 2007 John Wiley & Sons, Ltd.     

A semi‐implicit integration scheme for a combined viscoelastic‐damage model of plastic bonded explosives

This paper presents a new implementation of a constitutive model commonly used to represent plastic bonded explosives in finite element simulations of thermomechanical response. The constitutive model, viscoSCRAM, combines linear viscoelasticity with isotropic damage evolution. The original implementation was focused on short duration transient events; thus, an explicit update scheme was used. For longer duration simulations that employ significantly larger time step sizes, the explicit update scheme is inadequate. This work presents a new semi‐implicit update scheme suitable for simulations using relatively large time steps. The algorithm solves a nonlinear system of equations to ensure that the stress, damaged state, and internal stresses are in agreement with implicit update equations at the end of each increment. The crack growth is advanced in time using a sub‐incremental explicit scheme; thus, the entire implementation is semi‐implicit. The theory is briefly discussed along with previous explicit integration schemes. The new integration algorithm and its implementation into the finite element code, Abaqus, are detailed. Finally, the new and old algorithms are compared via simulations of uniaxial compression and beam bending. The semi‐implicit scheme has been demonstrated to provide higher accuracy for a given allocated computational time for the quasistatic cases considered here. Published 2014. This article is a US Government work and is in the public domain in the USA.     

Modeling of dynamic crack branching by enhanced extended finite element method

The conventional extended finite element method (XFEM) is enhanced in this paper to simulate dynamic crack branching, which is a top challenge issue in fracture mechanics and finite element method. XFEM uses the enriched shape functions with special characteristics to represent the discontinuity in computation field. In order to describe branched cracks, it is necessary to set up the additional enrichment. Here we have developed two kinds of branched elements, namely the “element crossed by two separated cracks” and “element embedded by a junction”. Another series of enriched degrees of freedom are introduced to seize the additional discontinuity in the elements. A shifted enrichment scheme is used to avoid the treatment of blending element. Correspondingly a new mass lumping method is developed for the branched elements based on the kinetic conservation. The derivation of the mass matrix of a four-node quadrilateral element which contains two strong discontinuities is specially presented. Then by choosing crack speed as the branching criterion, the branching process of a single mode I crack is simulated. The results including the branching angle and propagation routes are compared with that obtained by the conventionally used element deletion method.     

An enriched element‐failure method (REFM) for delamination analysis of composite structures

This paper develops an enriched element‐failure method for delamination analysis of composite structures. This method combines discontinuous enrichments in the extended finite element method and element‐failure concepts in the element‐failure method within the finite element framework. An improved discontinuous enrichment function is presented to effectively model the kinked discontinuities; and, based on fracture mechanics, a general near‐tip enrichment function is also derived from the asymptotic displacement fields to represent the discontinuity and local stress intensification around the crack‐tip. The delamination is treated as a crack problem that is represented by the discontinuous enrichment functions and then the enrichments are transformed to external nodal forces applied to nodes around the crack. The crack and its propagation are modeled by the ‘failed elements’ that are applied to the external nodal forces. Delamination and crack kinking problems can be solved simultaneously without remeshing the model or re‐assembling the stiffness matrix with this method. Examples are used to demonstrate the application of the proposed method to delamination analysis. The validity of the proposed method is verified and the simulation results show that both interlaminar delamination and crack kinking (intralaminar crack) occur in the cross‐ply laminated plate, which is observed in the experiment. Copyright © 2009 John Wiley & Sons, Ltd.     

Extended meshfree methods without branch enrichment for cohesive cracks

An extended meshless method for both static and dynamic cohesive cracks is proposed. This new method does not need any crack tip enrichment to guarantee that the crack closes at the tip. All cracked domains of influence are enriched by only the sign function. The domain of influence which includes a crack tip is modified so that the crack tip is always positioned at its edge. The modification is only applied for the discontinuous displacement field and the continuous field is kept unchanged. In addition to the new method, the use of Lagrange multiplier is explored to achieve the same goal. The crack is extended beyond the actual crack tip so that the domains of influence containing the crack tip are completely cut. It is enforced that the crack opening displacement vanishes along the extension of the crack. These methods are successfully applied to several well-known static and dynamic problems.