Page 337 - Biaxial Multiaxial Fatigue and Fracture

BiaxiaVMultiaxial Fatigue and Fracture
Andrea Carpinteri, Manuel de Freitas and Andrea Spagnoli (Eds.)
Q Elsevier Science Ltd. and ESIS. All rights reserved. 32 1

VARIABILITY IN FATIGUE LIVES: AN EFFECT OF THE ELASTIC ANISOTROPY
OF GRAINS?

Sylvie POMMIER
MSSMat, Ewle Centrale Paris,
Grande Voie des Vignes. 92299 Chatenay Malabry Cedex

ABSTRACT

Probabilistic approaches are developed to describe scale effects and scatter in fatigue. When
fatigue cracks are nucleated on defects, four main sources of scatter are usually acknowledged
the probability to find a defect per unit volume, the variability of the defects sizes, of their
distance to the surfaces and of their mutual distance. When fatigue cracks are not nucleated on
defects, the mechanisms at the origin of the scatter remain unclear. In a “defect free” metal
under bulk elastic conditions, crack nucleation is attributed to cyclic slip in “weak” grains.
Therefore, the variability of grain’s size and orientation is expected to be firstly responsible for
the scatter in fatigue lives. The main material dimension, to take into account for predicting a
scale effect, is therefore the grain size. However according to the elastic anisotropy of grains, a
material dimension larger than the grain size may prevail. As a matter of fact, using 3D elastic
finite element analyses and experiments, it is shown that the spatial distribution of stresses and
strains is self-organized in a polycrystal, at a scale larger than the grain size. This scale is
approaching for example 15 grains in copper. This effect is analogous to the “arching” effect,
widely studied in granular material.

KEYWORDS
Fatigue, scale effect, scatter, elasticity, anisotropy, percolation, grains.

INTRODUCTION
Life prediction methods for industrial components are usually based on crack nucleation
criteria. A vital point for the design of industrial component is to evaluate both the fatigue life
of the component and the confidence associated with this prediction. For this purpose, the key
parameters driving crack nucleation and the sources of variability have to be clearly identified.
On the one hand, most fatigue criteria rely on the assumption that in both LCF and HCF,
fatigue damage occurs in grains as a consequence of cyclic slip in “weak” grains or aggregates
of grains. The models enabling the transfer of data collected on samples to industrial
components are therefore evaluating the slip ability in “weak” grains. Under proportional