Fixtures for testing pin-end columns : A. W. Huber. ASTM Bulletin, no. 234, Dec. 1958, p. 41–45

Abstract

A model, proposed earlier, is modified in an attempt to explain a number of curious behaviors of corrosion-fatigue crack propagation (CFCP). The behaviors include effects of load ratio R in air and salt water vs vacuum, and effects of loading frequency at fixed R in these environments. Assumptions of the modeling are reviewed in detail in view of earlier objections to them. The ingredients of CFCP per this model: Poisson contraction, strain hardening, ligament surface attack/annihilation, and stress relaxation are developed and related to conditions of the crack tip locale. In the modeling, a parameter G, for growth rate factor, is developed solely as a function of the form of the ordinary or of the cyclic stress-strain curve. Previous work had developed a G1 for the ordinary curve, to be associated with the surface attack effect as in stress-corrosion cracking, and one G2 for the cyclic curve, to be associated with the stress relaxation effect as in fatigue crack propagation (FCP). A hybrid G21 is developed, combining attributes of both, which seems to successfully describe the corrosion induced augmentation of G2. Parametric curves of G21(+G2) correspond well with stage II frequency-dependent growth in CFCP. However, alone they do not explain the frequency-wise stage II threshold shift nor the frequency-independent air-environment FCP rate. It is found that these trends can be represented by loci of constant plastic strain rate, due to crack loading and propagation, relative to the surface annihilation rate. Such loci are determined by comparing growth rate factor maps with strain rate maps, using parametric curves of equal geometric-series spacing. Maps of this sort are used to analyze about a dozen cases of CFCP including two titanium alloys and three steels, with one of the steels of four different tempers. Stress-strain curves of the low strength steels are processed to remove the Lüder band effect to facilitate the modeling. The scheme for data organization involves a representation of indexes of the two kinds of parametric curves fitting the data, and the process zone size implied by the fitting. Model predication of load-ratio effects on the fatigue crack growth threshold is in good correspondence with literature data. Comparison of estimated process zone sizes with literature data of microstructural and fractographic size measurements is encouraging.