PSI - Issue 5

A. Eremin et al. / Procedia Structural Integrity 5 (2017) 889–895

890

A. Eremin et al/ Structural Integrity Procedia 00 (2017) 000 – 000

2

of factors which affect the crack propagation process and act as driven forces was initiated by studies of Wohler (1870) and Basquin (1910), which correlated the amount of fatigue cycles until failure with applied load. Then, in the 1960s P. Paris et al. (1963) proposed a relationship between fatigue crack propagation rate and the stress level magnitude designated as the stress intensity factor ΔK. This approach was the main method for fatigue life estimation of structures until the crack closure p henomenon was introduced by W. Elber (1970). He replaced ΔK by the effective stress intensity factor range ΔK eff , which is affected by crack closure phenomenon under a certain loading conditions. Recent papers enhancing this approach mostly are based on a strip-yield model used by Newman (1981, 1984). Our previous studies raise questions about load interaction models for evaluating fatigue crack growth mechanisms under variable amplitude loading via relationship between sequence-sensitive near-tip residual stress, threshold SIF range (ΔK th ) and crack opening SIF (Kop). The discovery of the correlation between σ* and closure free ΔK th (R. Sunder, 2015), gives a new understanding of the residual stress effect in metal fatigue. It is obvious that the monotonic plastic zone plays an important role, but it is one that will be modulated on a cycle-by-cycle basis by near-tip stress strain response to load history. Previously (R. Sunder, A. Andronik et al. (2016); R. Sunder, A. Biakov et al. (2016)), it was shown that crack closure, near- tip residual stress (through associated ΔK th ) and crack-tip blunting together determine near-threshold fatigue crack growth response to overload-underload sequences. But the recovery of crack closure in case of various load sequence was not clearly assessed experimentally. Moreover cycle-sequence sensitivity of ΔK th is revealed only when the crack is fully open over baseline cycling, so in the absence of visible cycle-sequence sensitivity, the effect of increased ΔK th may be wrongly attributed to closure. This opens a possibility of new experiments specifically targeting the relationship between crack-tip blunting/resharpening and closure transients after applied overload-underload cycles. Additionally, it could throw more light on the individual contribution of both near tip stress and crack closure. In order to clarify these ambiguities the digital image correlation (DIC) technique is applied for in-situ material response onto the applied loads in the plastic and elastic zones in the vicinity of the crack tip. Fatigue crack growth tests were carried out on single edge crack tension specimens made of Al-Cu alloy (AA 2024) under specially designed load sequences which includes basic cycles, the single overload or underload cycles and measurement cycles for DIC acquisition. The schematic in Fig. 1 illustrates the notched specimen cut from commercially available aluminum AA2024 sheet. Such specimens are used instead of conventional ASTM C(T) coupons due to absence of rotation under loading that would hamper DIC and more uniform stress-strain state due to their low thickness. The specimen was precracked till the crack length extends approximately by 1 mm. Cyclic loading was carried out by BiSS Nano 15 kN testing machine equipped with DIC measurement setup. Altami microscope and Basler piA2400-17 5MP videocamera were the parts of the hardware. All registered images have a physical width of 2.5 mm and resolution of 2454×2056 pixels. In so doing, the scale made 1.04 px/µm. Image capturing was conducted during loading cycles with sampling frequency of 0.03 Hz. Thus it takes about 30 seconds to run one full measurement cycle and 100 images make the total series for the subsequent DIC processing (frame rate – 3 fps). Preliminary automation of the system was performed. In doing so all components of the loading and measuring systems were synchronized to operate automatically during all experimental stages. These guarantees reproducible and constant loading with diminished operator induced errors. All the captured images of specimen surface were processed using Vic 2D (Correlated Solutions) software to calculate strain fields. The software makes it possible to use the virtual extensometers, which might be placed in desired locations at the image in order to estimate absolute and relative elongations along the crack propagation direction (y-axis). The extensometer tool allows measuring the crack opening displacement (COD) at the crack tip as well as at some distance behind the tip (i.e. where crack wake has already formed). Manual manner of crack tip positioning, twisted-like propagation path and nonuniform growth rate through the specimen thickness may result in some scatter in the computed COD values. All obtained data were smoothed by adjacent-averaging filter with kernel size equal to 8. The measurement results are presented in the form of hysteresis loops constructed in “load (kN) – strain (relative units)” coordinate s. Since ambiguous conclusions might be drawn from the analysis of these loops the 2. Experimental procedure