Christian Rey

Non-overlapping domain decomposition methods in structural mechanics

SummaryThe modern design of industrial structures leads to very complex simulations characterized by nonlinearities, high heterogeneities, tortuous geometries... Whatever the modelization may be, such an analysis leads to the solution to a family of large ill-conditioned linear systems. In this paper we study strategies to efficiently solve to linear system based on non-overlapping domain decomposition methods. We present a review of most employed approaches and their strong connection. We outline their mechanical interpretations as well as the practical issues when willing to implement and use them. Numerical properties are illustrated by various assessments from academic to industrial problems. An hybrid approach, mainly designed for multifield problems, is also introduced as it provides a general framework of such approaches.

Iterative accelerating algorithms with Krylov subspaces for the solution to large-scale nonlinear problems

Discrete solution to nonlinear systems problems that leads to a series of linear problems associated with non-invariant large-scale sparse symmetric positive matrices is herein considered. Each linear problem is solved iteratively by a conjugate gradient method. We introduce in this paper new solvers (IRKS, GIRKS and D-GIRKS) that rely on an iterative reuse of Krylov subspaces associated with previously solved linear problems. Numerical assessments are provided on large-scale engineering applications. Considerations related to parallel supercomputing are also addressed.

A Rayleigh–Ritz preconditioner for the iterative solution to large scale nonlinear problems

The approximation to the solution of large sparse symmetric linear problems arising from nonlinear systems of equations is considered. We are focusing herein on reusing information from previous processes while solving a succession of linear problems with a Conjugate Gradient algorithm. We present a new Rayleigh–Ritz preconditioner that is based on the Krylov subspaces and superconvergence properties, and consists of a suitable reuse of a given set of Ritz vectors. The relevance and the mathematical foundations of the current approach are detailed and the construction of the preconditioner is presented either for the unconstrained or the constrained problems. A corresponding practical preconditioner for iterative domain decomposition methods applied to nonlinear elasticity is addressed, and numerical validation is performed on a poorly-conditioned large-scale practical problem.

Fast estimation of discretization error for FE problems solved by domain decomposition

This paper presents a strategy for a posteriori error estimation for substructured problems solved by non-overlapping domain decomposition methods. We focus on global estimates of the discretization error obtained through the error in constitutive relation for linear mechanical problems. Our method allows to compute error estimate in a fully parallel way for both primal (BDD) and dual (FETI) approaches of non-overlapping domain decomposition whatever the state (converged or not) of the associated iterative solver. Results obtained on an academic problem show that the strategy we propose is efficient in the sense that correct estimation is obtained with fully parallel computations; they also indicate that the estimation of the discretization error reaches sufficient precision in very few iterations of the domain decomposition solver, which enables to consider highly effective adaptive computational strategies.

On the initial estimate of interface forces in FETI methods

The balanced domain decomposition (BDD) method and the finite element tearing and interconnecting (FETI) method are two commonly used non-overlapping domain decomposition methods. Due to strong theoretical and numerical similarities, these two methods are generally considered as being equivalently efficient. However, for some particular cases, such as for structures with strong heterogeneities, FETI requires a large number of iterations to compute the solution compared to BDD. In this paper, the origin of the poor efficiency of FETI in these particular cases is traced back to bad initial estimates of the interface stresses. To improve the estimation of interface forces, a novel strategy for splitting interface forces between neighboring substructures is proposed. The additional computational cost incurred is not significant. This yields a new initialization for the FETI method and restores numerical efficiency which makes FETI comparable to BDD even for problems where FETI was performing poorly. Various simple test problems are presented to discuss the efficiency of the proposed strategy and to illustrate the so-obtained numerical equivalence between the BDD and FETI solvers.

Total and selective reuse of Krylov subspaces for the resolution of sequences of nonlinear structural problems

This paper deals with the definition and optimization of augmentation spaces for faster convergence of the conjugate gradient method in the resolution of sequences of linear systems. Using advanced convergence results from the literature, we present a procedure based on a selection of relevant approximations of the eigenspaces for extracting, selecting and reusing information from the Krylov subspaces generated by previous solutions in order to accelerate the current iteration. Assessments of the method are proposed in the cases of both linear and nonlinear structural problems.

Study of the strong prolongation equation for the construction of statically admissible stress fields: implementation and optimization

This paper focuses on the construction of statically admissible stress fields (SA-fields) for a posteriori error estimation. In the context of verification, the recovery of such fields enables to provide strict upper bounds of the energy norm of the discretization error between the known finite element solution and the unavailable exact solution. The reconstruction is a difficult and decisive step insofar as the effectiveness of the estimator strongly depends on the quality of the SA-fields. This paper examines the strong prolongation hypothesis, which is the starting point of the Element Equilibration Technique (EET). We manage to characterize the full space of SA-fields satisfying the prolongation equation so that optimizing inside this space is possible. The computation exploits topological properties of the mesh so that implementation is easy and costs remain controlled. In this paper, we describe the new technique in details and compare it to the classical EET and to the flux-free technique for different 2D mechanical problems. The method is explained on first degree triangular elements, but we show how extensions to different elements and to 3D are straightforward.

A strict error bound with separated contributions of the discretization and of the iterative solver in non-overlapping domain decomposition methods

This paper deals with the estimation of the distance between the solution of a static linear mechanic problem and its approximation by the finite element method solved with a non-overlapping domain decomposition method (FETI or BDD). We propose a new strict upper bound of the error which separates the contribution of the iterative solver and the contribution of the discretization. Numerical assessments show that the bound is sharp and enables us to define an objective stopping criterion for the iterative solver.

Substructured formulations of nonlinear structure problems – influence of the interface condition

Summary nWe investigate the use of non-overlapping domain decomposition (DD) methods for nonlinear structure problems. The classic techniques would combine a global Newton solver with a linear DD solver for the tangent systems. We propose a framework where we can swap Newton and DD so that we solve independent nonlinear problems for each substructure and linear condensed interface problems. The objective is to decrease the number of communications between subdomains and to improve parallelism. Depending on the interface condition, we derive several formulations that are not equivalent, contrarily to the linear case. Primal, dual and mixed variants are described and assessed on a simple plasticity problem. Copyright

A monolithic strategy based on an hybrid domain decomposition method for multiphysic problems Application to poroelasticity

A monolithic strategy based on an hybrid domain decomposition method for the numerical simulation of multiphysic problems is presented. It relies on a “physical” choice of primal interface unknowns. First numerical assessments are described for poroelasticity problems.