动力阻尼特性及速度有限元法

为了节省动态问题的计算量和存贮量,本文在[16]的基础上,分析动态阻尼特性,研究阻尼与材料、有限元格网尺寸或频率等定量关系,给出了最大模的估计定理及不同情况的推论.结合数值分析实例,论证阻尼对于动拉应力的影响是按模有界的;其影响值存在正负.这说明,只认为阻尼会使应力“一律偏小”的观念是不确切的.文中,进一步说明了速度有限元法的特点.文末附录了必要的数值计算成果.     

A symplectic analytical wave propagation model for damping and steady state forced vibration of orthotropic composite plate structure

An analytical wave propagation model is proposed in this paper for damping and steady state forced vibration of orthotropic composite plate structure by using the symplectic method. By solving an eigen-problem derived in the symplectic dual system of free bending vibration of orthotropic rectangular thin plates, the wave shape of plate is obtained in symplectic analytical form for any combination of simple boundary conditions along the plate edges. And then the specific damping capacity of wave mode is obtained symplectic analytically by using the strain energy theory. The steady state forced vibration of built-up plates structure is calculated by combining the wave propagation model and the finite element method. The vibration of the uniform plate domain of the built-up plates structure is described using symplectic analytical waves and the connector with discontinuous geometry or material is modeled using finite elements. In the numerical examples, the specific damping capacity of orthotropic rectangular thin plate with three different combinations of boundary condition is first calculated and analyzed. Comparisons of the present method results with respect to the results from the finite element method and from the Rayleigh–Ritz method validate the effectiveness of the present method. The relationship between the specific damping capacity of wave mode and that of modal mode is expounded. At last, the damped steady state forced vibration of a two plates system with a connector is calculated using the hybrid solution technique. The availability of the symplectic analytical wave propagation model is further validated by comparing the forced response from the present method with the results obtained using the finite element method.     

A unified approach for elasto‐plastic FEM simulations,incorporating adaptive spatial and time discretization

In FEM calculations the discretization should be chosen in a way, that further mesh refinement does not change the results. Otherwise the discretization error might yield unphysically stiff material behavior. If elasto‐plasticity is considered, a second type of error due to the time discretization of the evolution equations has to be taken into account. Due to the non‐linearity of the underlying initial boundary value problem, large time increments often result in non‐converging solutions during equilibrium iteration. In our approach the time integration error resulting from a second order BDF2 time integration method is calculated and utilized in an automatic step size control. In conjunction with a DAE‐treatment of the initial boundary value problem, it allows in the average considerably larger time steps compared to classical ‘elastic predictor – plastic corrector’ schemes. To reduce the discretization error, adaptive mesh‐refinement based on a Z²‐error‐estimator is performed. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

Selection of intensity modulated radiation therapy treatment beam directions using radial basis functions within a pattern search methods framework

The selection of appropriate radiation incidence directions in radiation therapy treatment planning is important for the quality of the treatment plan, both for appropriate tumor coverage and for better organ sparing. The objective of this paper is to discuss the benefits of using radial basis functions within a pattern search methods framework in the optimization of the highly non-convex beam angle optimization (BAO) problem. Pattern search methods are derivative-free optimization methods that require few function value evaluations to converge and have the ability to avoid local entrapment. These two characteristics gathered together make pattern search methods suited to address the BAO problem. The pattern search methods framework is composed by a search step and a poll step at each iteration. The poll step performs a local search in a mesh neighborhood and assures convergence to a local minimizer or stationary point. The search step provides the flexibility for a global search since it allows searches away from the neighborhood of the current iterate. Radial basis functions are used and tested in this step both to influence the quality of the local minimizer found by the method and to obtain a better coverage of the search space in amplitude. A set of retrospective treated cases of head-and-neck tumors at the Portuguese Institute of Oncology of Coimbra is used to discuss the benefits of using this approach in the optimization of the BAO problem.     

Vibration and buckling analyses of FGM plates with multiple internal defects using XIGA-PHT and FCM under thermal and mechanical loads

In this paper, the vibration and buckling analyses of the FGM (functionally graded material) plates with multiple internal cracks and cutouts under thermal and mechanical loads are numerically investigated using the combined XIGA-PHT (extended isogeometric analysis based on PHT-splines) and FCM (finite cell method). Material properties are graded only in the thickness direction. The effective material properties are estimated by using either the rule of mixture or the Mori-Tanaka homogenization technique. The plate displacement field is based on the HSDT (higher-order shear deformation plate theory) without any requirement of the SCF (shear correction factor). The HSDT model can exactly represent the shear stress distribution and improve the accuracy of solutions. The PHT-splines can naturally fulfill the C¹-continuous requirement of the HSDT model. The representation of internal defects is mesh-independent. The discontinuous and singular phenomena induced by the cracks are captured using the enrichment pattern in the XIGA, and the influence of cutouts is implemented by the FCM. The geometries of cutouts are captured by means of adaptive quadrature procedure based on a simple unfitted structural mesh, which avoids the need for multiple patches to describe the complex geometry and eliminates the enforcement of C¹-continuity patch-coupling across the patch boundaries. The initial mesh density around the cracks and cutouts can be controlled flexibly utilizing the local refinement property of the PHT-splines. After validating the results of the developed approach with those available in the literature, the effects of material gradient index, side to thickness ratio, boundary conditions, cutout size and crack length on the normalized frequency and the critical buckling parameter are investigated. Numerical results illustrate the effectiveness and accuracy of the present approach.     

基于变体积约束的阻尼材料微结构拓扑优化研究北大核心CSCD

阻尼复合结构的抑振性能取决于材料布局和阻尼材料特性.该文提出了一种变体积约束的阻尼材料微结构拓扑优化方法,旨在以最小的材料用量获得具有期望性能的阻尼材料微结构.基于均匀化方法,建立阻尼材料三维微结构有限元模型,得到阻尼材料的等效弹性矩阵.逆用Hashin-Shtrikman界限理论,估计对应于期望等效模量的阻尼材料体积分数限,并构建阻尼材料体积约束限的移动准则.将获得阻尼材料微结构期望性能的优化问题转化为体积约束下最大化等效模量的优化问题,建立阻尼材料微结构的拓扑优化模型.利用优化准则法更新设计变量,实现最小材料用量下的阻尼材料微结构最优拓扑设计.通过典型数值算例验证了该方法的可行性和有效性,并讨论了初始微构型、网格依赖性和弹性模量等对阻尼材料微结构的影响.     

Multiscale Modelling of the Effective Viscoplastic Material Behaviour of Textile-Reinforced Polymers Using XFEM

In this contribution a modelling approach using numerical homogenisation techniques is applied to predict the effective nonlinear material behaviour of composites from simulations of a representative volume element (RVE). Numerical models of the heterogeneous material structure in the RVE are generated using the eXtended Finite Element Method (XFEM) which allows for a regular mesh. Suitable constitutive relations account for the material behaviour of the constituents. The influence of the nonlinear matrix material behaviour on the composite is studied in a physically nonlinear FE simulation of the local material behaviour in the RVE ­ effective stress-strain curves are computed and compared to experimental observations. The approach is currently augmented by a damage model for the fibre bundle. (© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

A relaxation-based approach to damage modeling

Common material models that take into account softening effects due to damage have the problem of ill-posed boundary value problems if no regularization is applied. This condition leads to a non-unique solution for the resulting algebraic system and a strong mesh dependence of the numerical results. A possible solution approach to prevent this problem is to apply regularization techniques that take into account the non-local behavior of the damage [1]. For this purpose a field function is often used to couple the local damage parameter to a non-local level, in which differences between the local and non-local parameter as well as the gradient of the non-local parameter can be penalized. In contrast, we present a novel approach [2] to regularization that no longer needs a non-local level but directly provides mesh-independent results. Due to the new variational approach we are also able to improve the calculation times and convergence behavior. (© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

A refined stress-strain analysis in the load transfer zone of flat specimens of high-strength unidirectional composites in uniaxial tension 1. Theoretical analysis

With the use of an analytical approach developed, the influence of distribution of a tangential load on the stress concentration in uniaxially tensioned flat specimens of high-strength unidirectional composites near the grips of a testing machine is evaluated. In view of singularity of the analytical solution derived at the points of discontinuity of boundary conditions, for estimating the stress concentration, it is suggested to employ the averaged value of longitudinal stresses, which is calculated by means of an improper integral across the thickness of a near-surface layer. __________ Translated from Mekhanika Kompozitnykh Materialov, Vol. 42, No. 6, pp. 787–796, November–December, 2006.     

Density gradient based regularization of topology optimization problems

In the present paper we focus on the regularization of topology design problems with regard to numerical instabilities such as the occurrence of optimal designs characterized by oscillating material density distributions, for example in the form of the well–known “checkerboard–like” patterns, as well as the dependence of the obtained designs on the used finite–element–mesh. In this context we discuss a regularization approach penalizing spatial oscillations of the material density by means of a penalty functional reflecting the global distribution of the density gradient. (© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Investigation of the Free-Surface Flow in Partially Filled Extruders

In this work we present the free-surface flow of high viscous Newtonian liquids in a simplified extruder model, namely the flat-plate model. Due to high density and viscosity ratios, a free-surface approach based on the volume of fluid method is used. The material distribution for different degree of fillings and the resulting power consumption is investigated. In addition, the obtained power characteristic is compared against analytical consideration and numerical simulation of a fully filled case. The investigation shows that the power characteristic of a partially filled extruder can be estimated by a fully filled computation. (© 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Surface effects on nonlinear free vibration of functionally graded nanobeams using nonlocal elasticity

In this paper, to consider all surface effects including surface elasticity, surface stress, and surface density, on the nonlinear free vibration analysis of simply-supported functionally graded Euler–Bernoulli nanobeams using nonlocal elasticity theory, the balance conditions between FG nanobeam bulk and its surfaces are considered to be satisfied assuming a cubic variation for the component of the normal stress through the FG nanobeam thickness. The nonlinear governing equation includes the von Kármán geometric nonlinearity and the material properties change continuously through the thickness of the FG nanobeam according to a power-law distribution of the volume fraction of the constituents. The multiple scale method is employed as an analytical solution for the nonlinear governing equation to obtain the nonlinear natural frequencies of FG nanobeams. The effect of the gradient index, the nanobeam length, thickness to length ratio, mode number, amplitude of deflection to radius of gyration ratio and nonlocal parameter on the frequency ratios of FG nanobeams is investigated.     

Material-Force-Based Refinement Indicators in Adaptive Strategies

A material-force-based refinement indicator for adaptive finite element strategies for finite elasto-plasticity is proposed. Starting from the local format of the spatial balance of linear momentum, a dual material counterpart in terms of Eshelby's energy-momentum tensor is derived. For inelastic problems, this material balance law depends on the material gradient of the internal variables. In a global format the material balance equation coincides with an equilibrium condition of material forces. For a homogeneous body, this condition corresponds to vanishing discrete material nodal forces. However, due to insufficient discretization, spurious material forces occur at the interior nodes of the finite element mesh. These nodal forces are used as an indicator for mesh refinement. Assigning the ideas of elasticity, where material forces have a clear energetic meaning, the magnitude of the discrete nodal forces is used to define a relative global criterion governing the decision on mesh refinement. Following the same reasoning, in a second step a criterion on the element level is computed which governs the local h-adaptive refinement procedure. The mesh refinement is documented for a representative numerical example of finite elasto-plasticity. (© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

A regularization approach for damage models based on a displacement gradient

Common material models that take into account softening effects due to damage encounter the problem of ill-posed boundary value problems if no regularization is applied. This condition leads to a non-unique solution for the resulting algebraic system and a strong mesh dependence of the numerical results. A possible solution approach to prevent this problem is to apply regularization techniques that take into account the non-local behavior of the damage [1]. For this purpose a field function is used to couple the local damage parameter to a non-local level, in which differences between the local and non-local parameter as well as the gradient of the non-local parameter can be penalized. In contrast, we present a novel approach to regularization in which no field function is needed [2]. Hereto, the regularization is carried out by means of the divergence of the displacements and no additional quantity is necessary since the displacements are already defined on a non-local level. The idea is that with an increasing value of the damage the element's volume will increase as well. This is a result of the softening due to the occurring damage. The increasing volume can be measured by the divergence of the displacements which can be penalized by an additional energy part. The lack of any field function and the regularization by the use of the divergence of the displacements entails several numerical advantages: the computational effort is considerably reduced and the convergence behavior is improved as well. Naturally, the numerical results are mesh independent due to the regularization. (© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Quantifying uncertainty with ensembles of surrogates for blackbox optimization

Blackbox optimization tackles problems where the functions are expensive to evaluate and where no analytical information is available. In this context, a tried and tested technique is to build surrogates of the objective and the constraints in order to conduct the optimization at a cheaper computational cost. This work introduces an extension to a specific type of surrogates: ensembles of surrogates, enabling them to quantify the uncertainty on the predictions they produce. The resulting extended ensembles of surrogates behave as stochastic models and allow the use of efficient Bayesian optimization tools. The method is incorporated in the search step of the mesh adaptive direct search (MADS) algorithm to improve the exploration of the search space. Computational experiments are conducted on seven analytical problems, two multi-disciplinary optimization problems and two simulation problems. The results show that the proposed approach solves expensive simulation-based problems at a greater precision and with a lower computational effort than stochastic models.

     

Prediction of fracture toughness of embrittled steels by means of the Small Punch Test

To characterise the randomly distributed strength and fracture toughness of brittle steels, many specimens have to be destroyed. Since the Small Punch Test (SPT) needs only little material, it is a well suited experiment, when only a small volume of material is available. In this study the cleavage fracture of a ferritic steel at low temperature was investigated using the Beremin model. The failure probability is described with a 2-parameter Weibull distribution in terms of the so-called Weibull stress, which is calculated using an elastic-plastic finite element stress analysis. While the transfer of Weibull parameters works well between similar geometries and loading conditions, it works bad in more general cases. Modifications of the Beremin model are necessary to overcome this problem. Recent publications consider a lower threshold value of the Weibull stress, which leads to a lower Weibull modulus and therefore to a stronger volume size effect of strength. The suitability of this approach to transfer cleavage fracture results from SPT to fracture mechanics specimens was investigated. (© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Layered elastic solid method for the generation of unstructured dynamic mesh

The layered elastic solid method (LESM), a modified elastic solid method (ESM) was put forward in the present study. In LESM, the computational zone is divided into several layers and material properties of these layers, which stay a certain value in ESM, are changed with mesh deformation. The deformation capability of mesh in LESM is better than ESM and the quality of the mesh generated by LESM is superior to that generated by ESM when they undergo the same deformation. In ESM, the main influence factors on mesh quality and deformation capability are Poisson’s ratio and single-step rotation angle. In LESM, mesh quality and deformation capability reach a largest value with an increase in Young’s modulus. Meanwhile, the mesh can achieve a larger deformation capability when single-step rotation angle is 0.25°. Finally, numerical simulation on a two-dimensional aerofoil using LESM was carried out. It is found that the results of LESM show a better agreement with experiment results.     

A homogenization method for heterogeneous materials using a multi-scale technique

For modern microelectronics solders lifetime and stability predictions are important. To perform such an analysis material properties are required. As electronic devices and the corresponding amount of matter used become smaller, the influence of a changing microstructure on mechanical properties must be considered. First some analytical methods were conducted for upper and lower bounds ignoring the exact geometric distribution of the solder phases. Second, analytical equations derived for geometries such as laminate structures were applied to examine the influence of the geometry on homogenized properties. Third, a multi-scale approach for periodic media was presented allowing for a more general analysis of structures. We assume that the solder material is composed of periodic cells, which represent the properties of the whole structure. Composite materials with periodic structures can be investigated by using at least two scales. A global scale is related to the whole piece of material whereas a local scale is related to the periodic cell only. The constitutive equations are stated and a homogenization technique for the elastic properties of arbitrary structures is derived. The resulting equations are solved numerically and results are presented. Again, for layered materials closed-form formulas are derived and compared to the numerical results. The method is also used to obtain effective mechanical properties for materials with linear hardening. (© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Modeling of microscopic failure in heterogeneous materials

The proper modeling of state-of-the-art engineering materials requires a profound understanding of the nonlinear macroscopic material behavior. Especially for heterogeneous materials the effective macroscopic response is amongst others driven by damage effects and the inelastic material behavior of the individual constituents [1]. Since the macroscopic length scale of such materials is significantly larger than the fine-scale structure, a direct modeling of the local structure in a component model is not convenient. Multiscale techniques can be used to predict the effective material behavior. To this end, the authors developed a modeling technique based on representative volume elements (RVE) to predict the effective material behavior on different length scales. The extended finite element method (XFEM) is used to model discontinuities within the material structure independent of the underlying FE mesh. A dual enrichment strategy allows for the combined modeling of kinks (material interfaces) and jumps (cracks) within the displacement field [2]. The gradual degradation of the interface is thereby controlled by a cohesive zone model. In addition to interface failure, a non-local strain driven continuum damage model has been formulated to efficiently detect localization zones within the material phases. An integral formulation introduces a characteristic length scale and assures the convergence of the approach upon mesh refinement [3]. The proposed method allows for an efficient modeling of substantial failure mechanisms within a heterogeneous structure without the need of remeshing or element substitution. Due to the generality of the approach it can be used on different length scales. (© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Goal-oriented mesh adaptation for FE-vibration analyses of shell-like structures

The well–known semidiscrete approach in structural dynamics consists of two steps. First the discretization of the spatial domain using finite elements is performed. In the second step the numerical integration of the resulting system of ordinary differential equations is executed. The numerical schemes in both discretization steps contribute to the total discretization error. In this contribution the focus is on the spatial discretization error and on efficient strategies for error assessment and error reduction by goal–oriented mesh adaptation schemes. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)