The coupled use of finite element (FEM) and boundary element (BEM) methods is an old dream in structural analysis: Very obviously, the specific advantages of the BEM - easy input and high accuracy of the results, comparatively simple extension to adaptivity - and the FEM - natural tackling with inhomogeneous or nonlinear, e.g. plastic, material, and fast computation - combine well to produce a very comprehensive and powerful tool for practical purposes. In spite of this, until now, the nonsymmetric, fully populated and non-positive-definite BEM matrices generated by the usual collocation BEM have prevented a strong shift to coupled analysis. However, none of these drawbacks remain if the symmetric Galerkin method is applied. This method not only provides a very straightforward coupling algorithm, but offers the possibility to generate positive definite BEM stiffness matrices and thus provides an access to standard iterative equation resolution and domain decomposition by substructuring. The problems associated with the hypersingular kernel functions that are inherent to the symmetric method, as well as the drawback of considerable computational cost, may be successfully overcome employing analytical integration. At least for two-dimensional polygonal domains, i.e. those that are most frequently to be analyzed, the method poses no serious problems. It is the aim of the present article to demonstrate that the method is open to engineering applications, including the coupled BEM/FEM analysis of elastoplastic problems.
«
The coupled use of finite element (FEM) and boundary element (BEM) methods is an old dream in structural analysis: Very obviously, the specific advantages of the BEM - easy input and high accuracy of the results, comparatively simple extension to adaptivity - and the FEM - natural tackling with inhomogeneous or nonlinear, e.g. plastic, material, and fast computation - combine well to produce a very comprehensive and powerful tool for practical purposes. In spite of this, until now, the nonsymmet...
»