A Partitioned finite element method for dynamical systems

A new analytical method is described for deriving the equations of motion of dynamical systems. The concept is to consider the displacements of the domain to be composed of rigid and elastic components. In contrast to other reduction methods, the domain modeled by finite number of degrees of freedom is discretized into two distinctive types of subdomains. Rigid and elastic subdomains are generated by consistent lumping of the domain properties under unique kinematic constraint relations. Equations of motion of the disjoint subdomains are derived by Lagrange's equations, in conjunction with the shape function matrix represented in partitioned form. This allows reduced sizes of matrices and avoids their possible singularities. Based on the invariance of energies under a compatible partitioned procedure, a simple analytical method is introduced for building the equations of motion of the whole domain from those of the subdomains. The dynamic analysis of a two-node domain with application to a blade-shaft combination is presented to illustrate the method.     

Adaptive error control for multigrid finite element

We consider the problem of adaptive error control in the finite element method including the error resulting from, inexact solution of the discrete equations. We prove a posteriori error estimates for a prototype elliptic model problem discretized by the finite element with a canomical multigrid algorithm. The proofs are based on a combination of so-called strong stability and, the orthogonality inherent in both the finite element method can the multigrid algorithm.     

Adaptive multigrid for finite element computations in plasticity

The solution of the system of equilibrium equations is the most time-consuming part in large-scale finite element computations of plasticity problems. The development of efficient solution methods are therefore of utmost importance to the field of computational plasticity. Traditionally, direct solvers have most frequently been used. However, recent developments of iterative solvers and preconditioners may impose a change. In particular, preconditioning by the multigrid technique is especially favorable in FE applications.The multigrid preconditioner uses a number of nested grid levels to improve the convergence of the iterative solver. Prolongation of fine-grid residual forces is done to coarser grids and computed corrections are interpolated to the fine grid such that the fine-grid solution successively is improved. By this technique, large 3D problems, invincible for solvers based on direct methods, can be solved in acceptable time at low memory requirements. By means of a posteriori error estimates the computational grid could successively be refined (adapted) until the solution fulfils a predefined accuracy level. In contrast to procedures where the preceding grids are erased, the previously generated grids are used in the multigrid algorithm to speed up the solution process.The paper presents results using the adaptive multigrid procedure to plasticity problems. In particular, different error indicators are tested.     

A parallel algebraic multigrid solver for finite element method based source localization in the human brain

Time plays an important role in medical and neuropsychological diagnosis and research. In the field of Electro- and MagnetoEncephaloGraphy (EEG/MEG) source localization, a current distribution in the human brain is reconstructed noninvasively by means of measured fields outside the head. High resolution finite element modeling for the field computation leads to a sparse, large scale, linear equation system with many different right hand sides to be solved. The presented solution process is based on a parallel algebraic multigrid method. It is shown that very short computation times can be achieved through the combination of the multigrid technique and the parallelization on distributed memory computers. A solver time comparison to a classical parallel Jacobi preconditioned conjugate gradient method is given. Received: 13 July 2001 / Accepted: 19 December 2001 RID="*" ID="*"Offprint requests: Carsten Wolters, MPI für neuropsychologische Forschung, MEG-Gruppe, Muldentalweg 9, 04828 Bennewitz, Germany (E-mail: wolters@cns.mpg.de) Communicated by G. Wittum     

A Nitsche-extended finite element method for earthquake rupture on complex fault systems

The extended finite element method (XFEM) provides a natural way to incorporate strong and weak discontinuities into discretizations. It alleviates the need to mesh discontinuities, allowing simulation meshes to be nearly independent of discontinuity geometry. Currently, both quasistatic deformation and dynamic earthquake rupture simulations under standard FEM are limited to simplified fault networks, as generating meshes that both conform with the faults and have appropriate properties for accurate simulation is a difficult problem. In addition, fault geometry is not well known; robustness of solution to fault geometry must be determined. Remeshing with varying geometry would make such tests computationally unfeasible. The XFEM makes a natural choice for discretization in these crustal deformation simulations on complex fault systems. Here, we develop a method based upon the XFEM using Nitsche’s method to apply boundary conditions, enabling the solution of static deformation and dynamic earthquake models. We compare several approaches to calculating and applying frictional tractions. Finally, we demonstrate the method with two problems: an earthquake community dynamic code verification benchmark and a quasistatic problem on a fault system model of southern California.     

On the robustness of a multigrid method for anisotropic reaction-diffusion problems

In this paper, we consider a reaction-diffusion boundary value problem in a three-dimensional thin domain. The very different length scales in the geometry result in an anisotropy effect. Our study is motivated by a parabolic heat conduction problem in a thin foil leading to such anisotropic reaction-diffusion problems in each time step of an implicit time integration method [7]. The reaction-diffusion problem contains two important parameters, namely ε >0 which parameterizes the thickness of the domain and μ >0 denoting the measure for the size of the reaction term relative to that of the diffusion term. In this paper we analyze the convergence of a multigrid method with a robust (line) smoother. Both, for the W- and the V-cycle method we derive contraction number bounds smaller than one uniform with respect to the mesh size and the parameters ε and μ.      

A real-time multigrid finite hexahedra method for elasticity simulation using CUDA

We present a multigrid approach for simulating elastic deformable objects in real time on recent NVIDIA GPU architectures. To accurately simulate large deformations we consider the co-rotated strain formulation. Our method is based on a finite element discretization of the deformable object using hexahedra. It draws upon recent work on multigrid schemes for the efficient numerical solution of partial differential equations on such discretizations. Due to the regular shape of the numerical stencil induced by the hexahedral regime, and since we use matrix-free formulations of all multigrid steps, computations and data layout can be restructured to avoid execution divergence of parallel running threads and to enable coalescing of memory accesses into single memory transactions. This enables to effectively exploit the GPU’s parallel processing units and high memory bandwidth via the CUDA parallel programming API. We demonstrate performance gains of up to a factor of 27 and 4 compared to a highly optimized CPU implementation on a single CPU core and 8 CPU cores, respectively. For hexahedral models consisting of as many as 269,000 elements our approach achieves physics-based simulation at 11 time steps per second.     

A finite element method for flow separation

The finite element model has been developed in order to solve separation pattern of the flow past an obstruction in a two-dimensional flow field. The Helmholtz-Poisson form of the Reynolds equations are solved alternately until a stable flow separation in the neighbourhood of the obstruction is obtained. In order to check the results of the finite element model, an experimental separation pattern using Pitot-tube measurements has been conducted. The computed and the experimental flow separation patterns show a good agreement.     

A hybrid method for finite element ordering

Nodal ordering for the formation of suitable sparsity patterns for stiffness matrices of finite element meshes are often performed using graph theory and algebraic graph theory. In this paper a hybrid method is presented employing the main features of each theory. In this method, vectors containing certain properties of graphs are taken as Ritz vectors, and using methods for constructing a complementary Laplacian, a reduced eigenproblem is formed. The solution of this problem results in coefficients of the Ritz vectors, indicating the significance of each considered vector.The present method uses the global properties of graphs in ordering, and the local properties are incorporated using algebraic graph theory. The main feature of this method is its capability of transforming a general eigenproblem into an efficient approach incorporating graph theory. Examples are included to illustrate the efficiency of the presented method.     

A smoothed finite element method for shell analysis

A four-node quadrilateral shell element with smoothed membrane-bending based on Mindlin-Reissner theory is proposed. The element is a combination of a plate bending and membrane element. It is based on mixed interpolation where the bending and membrane stiffness matrices are calculated on the boundaries of the smoothing cells while the shear terms are approximated by independent interpolation functions in natural coordinates. The proposed element is robust, computationally inexpensive and free of locking. Since the integration is done on the element boundaries for the bending and membrane terms, the element is more accurate than the MITC4 element for distorted meshes. This will be demonstrated for several numerical examples.     

A finite element method for localized failure analysis

A method is proposed which aims at enhancing the performance of general classes of elements in problems involving strain localization. The method exploits information concerning the process of localization which is readily available at the element level. A bifurcation analysis is used to determine the geometry of the localized deformation modes. When the onset of localization is detected, suitably defined shape functions are added to the element interpolation which closely reproduce the localized modes. The extra degrees of freedom representing the amplitudes of these modes are eliminated by static condensation. The proposed methodology can be applied to 2-D and 3-D problems involving arbitrary rate-independent material behavior. Numerical examples demonstrate the ability of the method to resolve the geometry of localized failure modes to the highest resolution allowed by the mesh.     

A finite element method for consolidation of clay

We consider a model for consolidation of clay in the case of an elasto-plastic soil skeleton. We prove existence of a solution and we prove an error estimate for a finite element method for finding approximate solutions of the problem.     

A parallel finite element solution method

New parallel computer architectures have revolutionized the design of computer algorithms, and promise to have significant influence on algorithms for structural engineering computations. In this paper, a parallel finite element solution method is presented. The solution method proposed does not require the formation of global system equations, but computes directly the element distortions, as opposed to solving a system of nodal equations. An element or substructure is mapped on to a processor of an MIMD multiprocessing system. Each processor stores only the information relevant to the element or substructure for which the processor represents. The finite element computations can be performed in parallel, in that a processor generates the local stiffness, computes the element distortions and determines the stress-strain characteristics for the element or substructure associated with the processor.     

An efficient algebraic multigrid method for solving optimality systems

An algebraic multigrid method (AMG) for solving convection-diffusion optimality systems is presented. Results of numerical experiments demonstrate robustness of the AMG scheme with respect to changes of the weight of the cost of the control and show that the computational performance of the proposed AMG scheme is comparable to that of AMG applied to single scalar equations.     

A finite element quasi-Eulerian method for three-dimensional fluid-structure interactions

A quasi-Eulerian approach is used in the development of a three-dimensional hydrodynamic finite element. With this approach the motion of the computing grid may be different from the motion of the material. Within each element the field variables are represented by trilinear interpolation functions and the pressure field is assumed to be constant. This leads to a set of simple relations for internal nodal forces that are easily coded and computationally efficient. Because the formulation is based upon a rate approach it is applicable to problems involving large displacements. The technique of degeneration is applied to the hexahedron to generate a pentahedral element. The use of the hour-glass dissipating nodal-forces for mesh stabilization is discussed. The procedure used to couple the fluid and structural domains of a problem is presented. The above method is applied to a three-dimensional, fluid-structure interaction problem in the area of reactor safety.     

A multiscale extended finite element method for crack propagation

In this paper, we propose a multiscale strategy for crack propagation which enables one to use a refined mesh only in the crack’s vicinity where it is required. Two techniques are used in synergy: a multiscale strategy based on a domain decomposition method to account for the crack’s global and local effects efficiently, and a local enrichment technique (the X-FEM) to describe the geometry of the crack independently of the mesh. The focus of this study is the avoidance of meshing difficulties and the choice of an appropriate scale separation to make the strategy efficient. We show that the introduction of the crack’s discontinuity both on the microscale and on the macroscale is essential for the numerical scalability of the domain decomposition method to remain unaffected by the presence of a crack. Thus, the convergence rate of the iterative solver is the same throughout the crack’s propagation.     

A finite element method for investigating general elastic multi-structures

A finite element method is proposed for investigating the general elastic multi-structure problem, where displacements on bodies, longitudinal displacements on plates, longitudinal displacements and rotational angles on rods are discretized using conforming linear elements, transverse displacements on plates and rods are discretized respectively using TRUNC elements and Hermite elements of third order, and the discrete generalized displacement fields in individual elastic members are coupled together by some feasible interface conditions. The unique solvability of the method is verified by the Lax–Milgram lemma after deriving generalized Korn’s inequalities in some nonconforming element spaces on elastic multi-structures. The quasi-optimal error estimate in the energy norm is also established. Some numerical results are presented at the end.     

A quasioptimal finite element method for elliptic interface problems

In [1] Babu?ka proposed perturbed variational methods for elliptic problems, with discontinuous coefficients. However, these methods are not quasioptimal, i.e. the approximate solutions generated by such methods do not reproduce the properties of “best approximation” possessed by the subspace of admissable approximants. In this paper we consider certain extrapolates obtained by use of a particular method of [1] and obtain “optimal” asymptotic error estimates. Our approach is similar to that of [7] where we proposed extrapolation methods for elliptic problems with smooth coefficients.     

A mixed finite element method for boundary flux computation

Most finite element schemes for thermal problems estimate boundary heat flux directly from the derivative of the finite element solution. The boundary flux calculated by this approach is typically inaccurate and does not guarantee a global heat balance.In this paper we present a mixed finite element method for calculating the boundary flux and show the superiority of this method through numerical examples of both diffusion and advection-diffusion problems.