Multiaxial fatigue of welded structures - Problems and present solutions

The assessment of multiaxial fatigue in design codes for welded steel structures is based on the use of nominal- or- hot-spot stresses, or by conventional hypotheses like von Mises, Tresca, principal stress or critical plane approaches. However, these fail, for non proportional loading which, compared with proportional loading, reduces fatigue strength and increases the scatter in fatigue lives. The use of lower design curves may cover current results, but could still be unsafe for cases not yet investigated. In contrast, the plane oriented hypothesis of equivalent effective stress (EESH) for ductile structural steels, which considers the interaction of damaging shear stresses, successfully correlates fatigue data obtained under uniaxial and multiaxial loading, including nonproportional and variable amplitude loading. However, in the case of combined spectrum loading, it proves necessary to assume a damage sum of D(real)= 0.35 instead of unity, as proposed in design codes. The present situation should be improved by further development of hypotheses based on local parameters, including fracture mechanics. However, the assessment of multiaxial fatigue cannot be achieved using a general hypotheses, only by different ones taking into account particular ductility-dependent damaging mechanisms