Fatigue Crack Growth and Closure at High Stress Ratios

Abstract

Fatigue crack growth tests have been carried out on a medium carbon structural steel over a wide range of stress ratios, i.e. from 0 to 0.7. All tests were conducted under constant amplitude loading conditions corresponding to growth rates in the Paris regime. Crack closure behaviour was observed experimentally by a surface strain gauge technique, and numerically by a finite element analysis under plane stress condition. While the crack closure could not be detected by experimental measurements at stress ratios equal to or greater than 0.5, the numerical results showed that closure occurred even at high stress ratios up to 0.7. The differences between experimentally and numerically determined crack opening levels were found for each stress ratio. A cause for these differences is discussed. In addition, two new types of crack tip parameters which have been proposed recently are evaluated by finite element analysis and their relevance to fatigue crack growth are discussed. It is concluded that fatigue crack growth rates are substantially determined by the effective stress intensity factor range which is based on the crack closure concept. NOMENCLATURE a = crack length d, = length of closed crack during loading da/dN = crack growth rate F = geometry factor AK = stress intensity factor range AK,= effective stress intensity factor range AKRP = stress intensity factor range which corresponds to the applied stress range (0--uRpG) AKtt = stress intensity factor range which corresponds to the applied stress range (urn -utt) L, = crack length that is close at umin P-, Pmax =minimum and maximum applied load R = stress ratio (umin/umu) ~y = strain in y-direction u,,, = crack opening stress uII =applied stress at which stress at crack tip becomes tensile oy =stress in y-direction ays = yield stress u-, urn= =minimum and maximum applied stress uRpG = applied stress just when plastic deformation occurs in the vicinity of crack tip

An algorithm for evaluating crack closure from local compliance measurements

“…In measurements referred to earlier, Tsukuda et al 7 . failed to detect crack closure at R ≥ 0.5, although their FEM analyses did indicate the occurrence of this mechanism for R ‐ratios up to 0.7.…”

“…have observed that the d a /d N versus Δ K eff data are still considerably separate for each positive R ‐ratio, the shift of a given plot in the positive direction of the Δ K eff axis increasing with the R ‐value. Results like those presented in Refs [6–8] are likely to imply that the CC concept cannot sufficiently account for the R ‐ratio influence on crack growth. However, Toyosada and coworkers (Toyosada et al unpublished observations) did obtain a good correlation of the d a /d N versus Δ K eff data from different positive R ‐ratio tests on specimens of the same material and geometry as those used by Skorupa et al 8…”

“…The CC measurements in the studies reported in Refs [7,8] were carried out using the so‐called local compliance method. A set of small surface strain gauges were bonded prior to the test on central crack tension specimens just above the crack line 7,8 or straddling that line (Toyosada et al unpublished data). CC readings were taken from the nearest strain guage behind the crack tip, 7 ahead of the crack tip (Toyosada et al .…”

“…A set of small surface strain gauges were bonded prior to the test on central crack tension specimens just above the crack line 7,8 or straddling that line (Toyosada et al unpublished data). CC readings were taken from the nearest strain guage behind the crack tip, 7 ahead of the crack tip (Toyosada et al . 2000, unpublished observations) or at either of these locations 8…”

“…In the studies reported in Refs [7,8] the crack opening levels ( P op ) were estimated from the measured load–deformation data using well‐known procedures, namely the offset displacement method 9 and the compliance offset method 10 . In both approaches it is assumed that a plot of load against any specimen displacement or strain is linear when the crack is fully open and becomes non‐linear when the crack begins to close.…”