Development of a numerical tool for random field discretization

The increasing interest, over the last few decades, in the behavior of structures subject to random structural properties and actions has motivated the development of new and innovative tools to conduct reliability computations. The increase in attention towards the stochastic finite element method has pointed out the necessity of dealing with the random variations of the model parameters. The so-called random field discretization methods can currently be used to achieve a satisfactory representation of the point-to-point variability of the properties of a system. A computational tool, developed by coupling the Karhunen–Loeve series expansion with the finite element method, is here presented. The paper has two innovative aspects. The first one concerns the improvement that has been obtained in the random field approximation by means of an adaptive refinement of the finite element mesh. The second one deals with the accurate application of the perturbation method.