Issue 35

Chahardehi et al, Frattura ed Integrità Strutturale, 35 (2016) 41-49; DOI: 10.3221/IGF-ESIS.35.05 42 Therefore it is understandable that the apparent bias in design codes and standards, when treating the fatigue crack growth phenomenon, has been towards conservatism in the prediction of growth rates. As such, emphasis has usually not been placed on treatment of some of the beneficiary features of fatigue - such as closure or the treatment of compressive stress cycles. In some engineering components, where a crack grows in the parent material, away from any residual stresses due to welds (and the larger inherent scatter in the fatigue behaviour in this region), the mean stress could be low or even compressive (e.g. self weight) and therefore benefit may be taken from the effect of fully compressive or part-compressive stress cycles. Although fatigue crack initiation has been observed under purely compressive remote stress cycles, current code recommendation could be misinterpreted, assuming no damage accumulation during a fully compressive stress cycle. Conversely, most national standards do not allow accounting for the beneficial effect of closure, and some even ignore the partial benefit of the compressive stress cycles in fatigue cracks growth. In BS7910 : 2013 [1], which more closely reflects UK experience and practice, the recommended FCG curves are presented for the cases of R  0.5 and R<0.5. The R  0.5 is recommended for welded joints, and data for this family of curves is gathered from results of tests on welded samples. The data for R<0.5 are gathered from a pulsating tension (minimum stress = 0 ~ 0.1) [2] taking into account the full stress intensity range [3]. No data has been used with inclusion of compressive stresses in the loading. The ESDU Data Sheet Item 80036 [4] notes two options for calculation of  K when the stress cycle is partly compressive: calculating the range from the tensile part only, and calculating it from full range. The key is that the same method should be used both for predicting growth and in the presentation of the basic data. This is also in accordance with the guidance of ASTM E647 [5]. The DNV Classification Note 30.2 (1984) [6] stated that compressive stresses do not contribute to crack propagation, but the recommendation was that the whole of the stress cycle should be considered. The current paper provides a critical state-of-the-art review of current understanding, with a view to highlighting current issues and challenges which may complicate more accurate modelling of the fatigue crack growth phenomenon in codes and standards. Note – In keeping with the accepted nomenclature, R-ratio is defined as the ratio of the minimum to maximum stress in a give load cycle, and U is defined as the ratio of the effective stress intensity factor and the total stress intensity factor ranges. E FFECTS OF C OMPRESSIVE L OADING – TIMELINE OF OBSERVATIONS atigue crack initiation and propagation under cyclic compression has been reported and investigated as early as the 60’s. Almen [7] shows examples of cracks generated under compressive loading such as compressive cracking of coil springs and ‘shelling’ in rails. Gerber [8] and Hubbard [9] and a number of other authors observed fatigue crack initiation under compression in laboratory and components, and this is a well-documented phenomenon. See for example the works of Suresh [10] and Fleck [11] for crack initiation in steels, and Solis et al [12] for initiation in ceramics, to give a few examples. These tests are generally conducted on notched samples, under remote compressive stress cycles. The observations from the tests were cracks that initiated at the notch, and grew to a small length and would then arrest – the final length before arrest was 0.6mm from Suresh’s work [13] and between 0.68mm and 2.48mm in Fleck’s tests [11]. The experiments performed by Fleck [11], Suresh [10], Pippan [12], Hermann [14] and Kasaba [15], all resulting in nucleation and growth of cracks, were all performed under fully compressive loading. In all of these experiments, the cracks grew with decreasing growth rate until complete crack arrest at a certain crack length. Yu et al [16] demonstrated the significance of the compressive stress on fatigue crack propagation rate of aluminium alloy 2024-T351. Tests performed by Tack and Beevers [17] showed that the fatigue crack propagation rate under a negative stress ratio R is greater than that for R=0.1. Pommier [18] showed that the addition of a compressive part to the loading could increase the crack growth rate by a factor of five. The observed behaviour was attributed to plastic properties of the material and its kinematic hardening. The general outcome of these observations is that compressive loading in notched samples can lead to fatigue crack initiation, and in the presence of purely compressive loading the crack would grow to a finite albeit small size before arrest. The fact that crack initiation and growth occurs under compressive loading is significant because in some engineering application, not foreseeing this crack could cause unexpected loss of the component. In the following sections, the factors affecting growth and subsequent arrest of these cracks are reviewed. F