Effects of metal strength statistics on the representation of fatigue life in structural components

Abstract

Limits are considered to the deterministic approach and the scope for a probabilistic one in calculating the working lives of structural elements, as well as the advantages and shortcomings of these approaches for estimating the safety of objects subject to elevated hazard.In [1], criteria are considered for the limiting states in cyclic strength of a metal, which are now used in standardization documents on strength and safety, and also in determining the cyclic working life of structural elements for the instance of several loads [2].In the first systematic studies on cyclic strength for metals with several loads, it was found that there is an effect from the loading history on the working life. Fairly detailed studies have been made on the simple state consisting of two batches of loading cycles with differing stress amplitudes with 1-2 and 2-1 packet sequences. A paradox was observed: for σ 1 > σ 2 and a 1-2 sequence of packets, some observers found a reduction in working life while others found an increase, and this paradox has as yet not been elucidated.In this connection, emphasis has been given to various approaches to failure and the effects of the loading history on the working life: a search has been made for the physical causes of this in the plastic few-cycle fatigue mechanism in the formation of cracks in stress concentration zones and a search for statistical regularities for states of one-packet and two-packet loading.The existing method of cyclic strength calculation employs safety coefficients on the number of cycles and the stresses [2], and it has been shown that in the linear fatigue damage summation (LFDS), the calculated fatigue curve does not rule out the probability of attaining a limiting state (crack formation) for various stress amplitudes. Then firstly the representation of absolute conservatism in that method of estimating the cyclic strength does not occur, and secondly, when one examines the causes of cracking and the actual crack growth, it is necessary to incorporate the effects from the loading history on the working life and probabilistic formulation of it. In spite of these considerations, calculations on cyclic strength in the formulation of [2] will clearly be performed in the future, since it is almost impossible to abandon at once the deterministic approach, which has been used for quite a long while. It is therefore necessary to consider the limits to that approach and the scope for a probabilistic approach in calculating working life, and also the advantages and disadvantages of those approaches for the corresponding estimates of safety in objects with elevated hazards.Experts should incorporate the terminology and probabilistic specifications introduced by the Federal Law on Technical Regulation, and distinguish between the calculated representation (to some extent sound) and the purely expert [3] representation of the permissible risk for failure in particular constructions, and also the degrees of safety in their use. An expert needs not only personal experience and t...