h- and p-Adaptive boundary element methods

Adaptive boundary element methods and their application to the 2-D potential and elasticity problems are presented in the paper. Based on the theory of approximation, a-posteriori error estimates for the boundary element solution have been developed. Both h-version and p-version are studied. The linear elements are used in the h-version, on the other hand, the hierachical shape functions are employed for the implementation of the p-version. A number of numerical examples are given, demonstrating the success of the techniques developed in this paper.     

Adaptive h-p procedures for high accuracy finite element analysis of two-dimensional linear elastic problems

Two adaptive procedures of the h-p version for a high accuracy finite element analysis of two-dimensional elastic problems are studied. These are based on a strategy of first using an h-version to predict a nearly optimal mesh up to a certain accuracy and then following up with a p-version to achieve a higher accuracy. The h-version, using linear triangular elements, is developed by coupling a code, ADMESH, with an error estimation in energy norm. Following the h-version, two alternative procedures of non-uniform p-refinements are then performed. In procedure I, p-refinements are made in one step by selectively adding hierarchical shape functions of order p = 2 and 3 based on the estimated error in energy norm. In procedure II, p-refinements are made in a step-by-step way by which, in the kth step of the p-refinements, hierarchical shape functions of order p = k + 1 are selectively introduced. In the first step of p-refinements of procedure II, hierarchical variables are selected by means of the estimated errors in energy norm, whereas in the later steps, they are selected with a guidance of an error estimate which evaluates the local average error of stresses. The performances of both procedures and the rate of convergence are studied in numerical examples. Numerical tests for the error estimates being used are also made. Obtained results indicate that both procedures can achieve a high accuracy (say, error below 5% measured in energy norm) in an exponential rate of convergence.     

Adaptive hierarchical boundary elements

A residual definition of error indicators and estimators is used. A p-and a h-implementation of an adaptive hierarchical boundary element method is presented.

The direct version of boundary elements based on the collocation method is reviewed and the way in which the residual boundary function is obtained is presented. The hierarchical definition of the interpolation and its advantages are discussed. Numerical interpolation to compute error estimators and indicators is established and its non-dimensionality is defined.     

A new residue and nodal error evaluation in h-adaptive boundary element method

An error indicator and evaluation of nodal error are proposed in the h-adaptive boundary element method. Error indicator is defined based on a new residue calculated at collocation points only. Nodal error is evaluated from the approximated residue by solving a matrix equation. Convergence of the solution is estimated with the nodal error. Efficiency of the proposed strategy is proved through numerical examples.     

Efficient error estimation and adaptive meshing method for boundary element analysis

This paper describes a new and efficient error estimator by using the Direct Regular Method and h or h-r adaptive meshing for BEM analysis. This posteriori error estimator correctly indicates the discretization errors on each element. Based on the error distribution, and the adaptive meshing is generated automatically. The accuracy and convergence of this method are demonstrated by the numerical results on the stress concentration problem and the crack problem.     

An adaptive discontinuous Galerkin method for viscoplastic analysis

This paper describes an h-adaptive, space-time discontinuous Galerkin finite element method for quasi-static viscoplastic response in a time-varying domain. The equations of equilibrium and the evolution equations in the viscoplastic material model comprise an elliptic/hyperbolic system. We focus here on two aspects of the model: stabilization of the hyperbolic subproblem and residual-based error estimates and adaptive algorithms for viscoplastic analysis.     

Adaptive multilevel BEM for acoustic scattering

We study the h- and p-versions of the Galerkin boundary element method for integral equations of the first kind in 2D and 3D which result from the scattering of time harmonic acoustic waves at hard or soft scatterers. We derive an abstract a-posteriori error estimate for indefinite problems which is based on stable multilevel decompositions of our test and trial spaces. The Galerkin error is estimated by easily computable local error indicators and an adaptive algorithm for h- or p-adaptivity is formulated. The theoretical results are illustrated by numerical examples for hard and soft scatterers in 2D and 3D.     

The p-adaptive boundary integral equation method

This work is concerned with the development and application of the p-adaptive boundary integral equation method (BIEM) in practical elastostatics engineering situations. Some basic concepts inherent to the p-adaptive technique are summarized and discussed. A pseudocode which illustrates the way for generating the p-adaptive system of equations in microcomputers is also provided.

Two numerical examples, which show the accuracy of the method discussed herein are included.     

Coupled model- and solution-adaptivity in the finite-element method

Modeling of elastic thin-walled beams, plates and shells as ID- and 2D-boundary value problems is valid in undisturbed subdomains. Disturbances near supports and free edges, in the vicinity of concentrated loads and at thickness jumps cannot be described in a sufficient way by 1D- and 2D-BVPs. In these disturbed subdomains dimensional (d)-adaptivity and model (m)-adaptivity have to be performed coupled with h- and/or p-adaptivity using hierarchically expanded test spaces in order to guarantee reliable and efficient overall results. The expansion strategy is applied for enhancing the spatial dimension and the model which is more efficient and evident for engineers than the reduction method.

Using local residual error estimators of the primal problem in the energy norm by solving Dirichlet-problems on element patches, an efficient integrated adaptive calculation of the discretization—and the dimensional error is possible and reasonable, demonstrated by examples.

We also present an error estimator of the dual problem, namely a posterior equilibrium method (PEM) for calculation of the interface tractions on local patches with Neumann boundary conditions, using orthogonality conditions. These tractions are equilibrated with respect to the global equilibrium condition of the stress resultants. An upper bound error estimator based on differences between the new tractions and the discontinuous tractions calculated from the stresses of the current finite element solution. The introduction of new element boundary tractions yields a method which can be regarded as a stepwise hybrid displacement method or as Trefftz method for local Neumann problems of element patches.

An important advantage of PEM is the coupled computation of local discretization, dimensional- and model errors by an additive split.     

A posteriori error estimation and adaptive mesh refinement for combined thermal-stress finite element analysis

Local and global error estimators and an associated h-based adaptive mesh refinement schemes are proposed for coupled thermal-stress problems. The error estimators are based on the “flux smoothing” technique of Zienkiewicz and Zhu with important modifications to improve convergence performance and computational efficiency. Adaptive mesh refinement is based on the concept of adaptive accuracy criteria, previously presented by the authors for stress-based problems and extended here for coupled thermal-stress problems. Three methods of mesh refinement are presented and numerical results indicate that the proposed method is the most efficient in terms of number of adaptive mesh refinements required for convergence in both the thermal and stress solutions. Also, the proposed method required a smaller number of active degrees of freedom to obtain an accurate solution.     

Accuracy estimation in the finite element analysis of transverse bending of thin flat plates

As a basic study for the establishment of an accuracy estimation method in the finite element method, this paper deals with the problems of transverse bending of thin, flat plates. From the numerical experiments for uniform mesh division, the following relation was deduced, ε ∝ (h/a)^k, k 1, where ε is the error of the computed value by the finite element method relative to the exact solution and h/a is the dimensionless mesh size. Using this relation, an accuracy estimation method, which was based on the adaptive determination of local mesh sizes from two preceding analyses by uniform mesh division, was presented.

A computer program using this accuracy estimation method was developed and applied to 28 problems with various shapes and loading conditions. The usefulness of this accuracy estimation method was illustrated by these application results.     

Multipoint flux mixed finite element methods for slightly compressible flow in porous media

In this paper, we consider multipoint flux mixed finite element discretizations for slightly compressible Darcy flow in porous media. The methods are formulated on general meshes composed of triangles, quadrilaterals, tetrahedra or hexahedra. An inexact Newton method that allows for local velocity elimination is proposed for the solution of the nonlinear fully discrete scheme. We derive optimal error estimates for both the scalar and vector unknowns in the semidiscrete formulation. Numerical examples illustrate the convergence behavior of the methods, and their performance on test problems including permeability coefficients with increasing heterogeneity.     

Guaranteed and Fully Robust a posteriori Error Estimates for Conforming Discretizations of Diffusion Problems with Discontinuous Coefficients

We study in this paper a posteriori error estimates for H ¹-conforming numerical approximations of diffusion problems with a diffusion coefficient piecewise constant on the mesh cells but arbitrarily discontinuous across the interfaces between the cells. Our estimates give a global upper bound on the error measured either as the energy norm of the difference between the exact and approximate solutions, or as a dual norm of the residual. They are guaranteed, meaning that they feature no undetermined constants. (Local) lower bounds for the error are also derived. Herein, only generic constants independent of the diffusion coefficient appear, whence our estimates are fully robust with respect to the jumps in the diffusion coefficient. In particular, no condition on the diffusion coefficient like its monotonous increasing along paths around mesh vertices is imposed, whence the present results also include the cases with singular solutions. For the energy error setting, the key requirement turns out to be that the diffusion coefficient is piecewise constant on dual cells associated with the vertices of an original simplicial mesh and that harmonic averaging is used in the scheme. This is the usual case, e.g., for the cell-centered finite volume method, included in our analysis as well as the vertex-centered finite volume, finite difference, and continuous piecewise affine finite element ones. For the dual norm setting, no such a requirement is necessary. Our estimates are based on H(div)-conforming flux reconstruction obtained thanks to the local conservativity of all the studied methods on the dual grids, which we recall in the paper; mutual relations between the different methods are also recalled. Numerical experiments are presented in confirmation of the guaranteed upper bound, full robustness, and excellent efficiency of the derived estimators.     

Stable Evaluation of Jacobian Matrices on Highly Refined Finite Element Meshes

In the adaptive finite element method, the solution of a p.d.e. is approximated by finer and finer meshes, which are controlled from error estimators. So, starting from a given coarse mesh, some elements are subdivided a couple of times. We investigate the question of avoiding instabilities which limit this process from the fact that nodal coordinates of one element coincide in more and more leading digits. To overcome this problem we demonstrate a simple mechanism for red subdivision of triangles (and hanging nodes) and a more sophisticated technique for general quadrilaterals.     

Geometry representation issues associated with p-version finite element computations

This paper addresses issues related to accurate geometry representation for p-version finite elements on curved three-dimensional domains. Specific options to account for domain geometry information during element-level computation are identified. Accuracy requirements on the geometry related approximations to preserve the optimal rate of finite element error convergence for second-order elliptic boundary value problems are given. An element geometric mapping scheme based on blending the exact shape of the domain boundary is described that can either be used directly during element integrations, or used to construct element-level geometric approximations of required accuracy. Smoothness issues of the rational blends on simplex topologies are discussed and a numerical example based on the solution of Poisson's equation in three dimensions is presented to illustrate the impact of the rational blends on the optimal rate of finite element error convergence.     

Deterministic parallel backtrack search

The backtrack search problem involves visiting all the nodes of an arbitrary binary tree given a pointer to its root subject to the constraint that the children of a node are revealed only after their parent is visited. We present a fast, deterministic backtrack search algorithm for a p-processor COMMON CRCW-PRAM, which visits any n-node tree of height h in time O((n/p+h)(logloglogp)²). This upper bound compares favourably with a natural Ω(n/p+h) lower bound for this problem. Our approach embodies novel, efficient techniques for dynamically assigning tree-nodes to processors to ensure that the work is shared equitably among them.     

A variational multiscale a posteriori error estimation method for mixed form of nearly incompressible elasticity

This paper presents an error estimation framework for a mixed displacement–pressure finite element method for nearly incompressible elasticity. The proposed method is based on Variational Multiscale (VMS) concepts, wherein the displacement field is decomposed into coarse scales that can be resolved by a given finite element mesh and fine scales that are beyond the resolution capacity of the mesh. Variational projection of fine scales onto the coarse-scale space via variational embedding of the fine-scale solution into the coarse-scale formulation leads to the stabilized method with two major attributes: first, it is free of volumetric locking and, second, it accommodates arbitrary combinations of interpolation functions for the displacement and pressure fields. This VMS-based stabilized method is equipped with naturally derived error estimators and offers various options for numerical computation of the error. Specifically, two error estimators are explored. The first method employs an element-based strategy and a representation of error via a fine-scale error equation defined over element interiors which is evaluated by a direct post-solution evaluation. This quantity when combined with the global pollution error results in a simple explicit error estimator. The second method involves solving the fine-scale error equation through localization to overlapping patches spread across the domain, thereby leading to an implicit calculation of the local error. This implicit calculation when combined with the global pollution error results in an implicit error estimator. The performance of the stabilized method and the error estimators is investigated through numerical convergence tests conducted for two model problems on uniform and distorted meshes. The sharpness and robustness of the estimators is shown to be consistent across the test cases employed.     

A posteriori error estimates for nonconforming approximations of Steklov eigenvalue problems

This paper deals with a posteriori error estimators for the non conforming Crouzeix-Raviart finite element approximations of the Steklov eigenvalue problem. First, we define an error estimator of the residual type which can be computed locally from the approximate eigenpair and we prove the equivalence between this estimator and the broken energy norm of the error with constants independent of the corresponding eigenvalue. Next, we prove that edge residuals dominate the volumetric part of the residual and that the volumetric part of the residual terms dominate the normal component of the jumps of the discrete fluxes across interior edges. Finally, based on these results, we introduce two simpler equivalent error estimators. The analysis shows that these a posteriori error estimates are optimal up to higher order terms and that may be used for the design of adaptive algorithms.     

Fast robust estimation of prediction error based on resampling

Robust estimators of the prediction error of a linear model are proposed. The estimators are based on the resampling techniques cross-validation and bootstrap. The robustness of the prediction error estimators is obtained by robustly estimating the regression parameters of the linear model and by trimming the largest prediction errors. To avoid the recalculation of time-consuming robust regression estimates, fast approximations for the robust estimates of the resampled data are used. This leads to time-efficient and robust estimators of prediction error.     

Curved shell elements for heat conduction with p-approximation in the shell thickness direction

A finite element formulation is presented for the curved shell elements for heat conduction where the element temperature approximation in the shell thickness direction can be of an arbitrary polynomial order p. This is accomplished by introducing additional nodal variables in the element approximation corresponding to the complete Lagrange interpolating polynomials in the shell thickness direction. This family of elements has the important hierarchical property, i.e. the element properties corresponding to an approximation order p are a subset of the element properties corresponding to an approximation order p + 1. The formulation also enforces continuity or smoothness of temperature across the inter-element boundaries, i.e. C⁰ continuity is guaranteed.

The curved shell geometry is constructed using the co-ordinates of the nodes lying on the middle surface of the shell and the nodal point normals to the middle surface. The element temperature field is defined in terms of hierarchical element approximation functions, nodal temperatures and the derivatives of the nodal temperatures in the element thickness direction corresponding to the complete Lagrange interpolating polynomials. The weak formulation (or the quadratic functional) of the three-dimensional Fourier heat conduction equation is constructed in the Cartesian co-ordinate space. The element properties of the curved shell elements are then derived using the weak formulation (or the quadratic functional) and the hierarchical element approximation. The element matrices and the equivalent heat vectors (resulting from distributed heat flux, convective boundaries and internal heat generation) are all of hierarchical nature. The element formulation permits any desired order of temperature distribution through the shell thickness.

A number of numerical examples are presented to demonstrate the superiority, efficiency and accuracy of the present formulation and the results are also compared with the analytical solutions. For the first three examples, the h-approximation results are also presented for comparison purposes.