ABSTRACT

The present paper suggests a probabilistic damage tolerance supplement to the design SoN curves for welded joints. The reliability levels are derived from extensive testing with fillet welded joints (Fclass detail) for which the entire crack growth history is measured, not only the final fatigue life. The statistics in time to reach given crack depths are determined. A Markov model is fitted to the data base and scaled to in-service conditions. Results for some frequent cases occurring in practice for offshore structures are readily deri ved and presented. The scope is to provide the practising engineer with simple graphical aids that predict the reliability against fatigue fracture during service life. The impact of chosen fatigue design factors and uncertainty in applied stresses is revealed. The effect of an in-service inspection program is also predicted. The graphical tools are suggested for use in support of decision-making at the design stage, without any advanced fracture mechanics modelling and stochastic simulation.

1 INTRODUCTION

Fatigue durability and inspection planning have for a long time been important issues in the design and scheduled inspection of welded components and structures. For welded structures subjected to cyclic repetitive loading, the admissible stresses in the vicinity of welded joints are relatively low due to the joints vulnerability to fatigue damage. The susceptibility to fatigue damage is due to stress concentrations, microscopic initial flaws and residual welding stresses. Even large complex structures may often be divided into a relatively low number of elementary joints. To verify the fatigue life and control the fatigue damage during service of such structures, it is necessary to study in detail the fatigue behaviour of these simple joints. It has been recognised that the fatigue process is a random phenomenon. Fatigue laboratory tests exhibit considerable scatter.

This content is only available via PDF.
You can access this article if you purchase or spend a download.