Abstract
Traditionally, the most commonly used mesh adaptive strategies for linear elastic problems are based on the use of an energy norm for the measurement of the error, and a mesh refinement strategy based on the equal distribution of the error between all the elements. However, little attention has been paid to the study of alternative error norms and alternative refinement strategies. This paper studies the feasibility of using alternative mesh refinement strategies based on
— the use of the classical error energy norm and an optimality criterion based on the equal distribution of the density of error,
— the use of alternative error norms based on measurements of the point wise error contained in the main magnitudes that control the equilibrium problem and/or the material constitutive equations such as the stresses (e.g. the Von Mises stress).
The feasibility of using all the described strategies is demonstrated through the solution of a benchmark example. This example is also used for comparison between the described refinement criteria.
