Page 18 - Biaxial Multiaxial Fatigue and Fracture
P. 18
BiaxiallMultiaxial Fatigue and Fracture
Andrea Carpinteri, Manuel de Freitas and Andrea Spagnoli (Eds.)
Q Elsevier Science Ltd. and ESIS. All rights reserved. 3
ASSESSMENT OF WELDED STRUCTURES BY
A STRUCTURAL MULTIAXIAL FATIGUE APPROACH
Ky DANG VAN', Andr6 BIGNONNET2 and Jean-Luc FAYARD2
'taboratoire de Mkcanique des Solides, Ecole Polytechnique, 91 128, Palaiseau, France
PSA Peugeot Citroen, Route de Gizy, 78943, Vdizy- Villacoublay Cedex, France
ABSTRACT
A structural multiaxial computing method for the fatigue assessment of welded structures is
presented. This approach is based on the use of a local equivalent stress, or design stress,
derived from the shear stress and the concomitant hydrostatic pressure previously proposed by
Dang Van. Associated with a specific shell finite element meshing methodology, the method is
successfully used to assess the fatigue resistance of welded automotive structures. The
approach also allows the role of the welding process upon the fatigue behaviour to be
addressed by a better description of the influence of the local residual stress state.
KEYWORDS
Welded structures, multiaxial fatigue, residual stresses, computing methods, design stress
INTRODUCTION
Although extensive work has already been done, the prediction of the fatigue strength of
welded structures is still a widely open subject. Engineers in design offices do not dispose of
reliable and accurate methods to evaluate the fatigue life of such structures, with regard to the
results provided by the modern structural calculation methods (Finite Element Method). If
some propositions exist, they are most of the time inapplicable so that, in practice, engineers
use simplified methods of poor accuracy. For example, for the calculation of metallic bridges, a
design stress S is evaluated from the nominal stress derived from a beam calculation; then the
fatigue life is estimated from S-N curves given by the EUROCODE ITI established
experimentally by class of structural details. One can imagine that if there are a few changes in
the geometry or the loading mode of these details, one can be out of the limits covered by the
fatigue tests and therefore have erroneous predictions.
Beside the global approaches as the EUROCODE III, several proposals called local
approaches )> exist [I]. Among them, one can distinguish those which study the crack initiation
and those which consider that microcracks are already formed and only take into account their
propagation. The latter uses the Paris law, or derivative laws, which appears to be well founded
on the recognised concepts of the Fracture Mechanics, but are nevertheless not so easy to apply
on actual structures. As noted recently by D.L. Mc Dowel1 [2], the first cracks initiated in