reference 1 that shear lips grow in width, even in a constant ∆K test, until an equilibrium width has been reached. A linear relation was found between equilibrium width and ∆Keff for AA 2024. An equivalent behaviour for AA 5083 was assumed. In figure 2 a fatigue fracture surface is shown. The shear lip is suppressed on one side of the specimen by making a scratch with a depth of 0.1 mm along the crack growth direction. Figure 2. Shear lips suppressed on one side in AA 5083. Hardly any difference was found in crack growth rate between specimens with scratches on both sides, where shear lips were fully suppressed, or specimens with shear lips on one side or specimens with full shear lips. Also the change in slope occurred equally, for da/dN=0.1 µm/cycle, in all three different kind of specimens. Based on a preliminarily study of the fracture surfaces the equilibrium shear lip width is estimated (for ts,eq >0): ts,eq = 0.90∆Keff–5 (mm) (4) If the actual ts,eq and ∆Keff do not fulfill this equation, the shear lip width will increase or decrease until the equilibrium situation is reached, according to the equation given above. The rate of change in shear lip width is taken proportional with the difference in actual width and equilibrium width, in formula form: dts/da = C(ts,eq-ts) (5) For AA 5083 the C-value is estimated as: C= 0.40/ts,eq (mm-1) (6) In a constant amplitude test the ∆Keff changes continuously and hence also ts,eq and ts; ts lags behind ts,eq for the growing crack. It is assumed that the crack driving force ∆K is influenced by the slope changing (micro)mechanism and that the shear lip width is a measure for it. The ∆K that is responsible for the lower da/dN increase in the shear lip area and beyond is called ∆K(shear). It is defined as: ∆K(shear) = ∆K/(1+2(√2)–1)ts/t) (7) It means that the product of ∆K and the transverse crack length is assumed to be constant. If a shear lip angle of 45° is taken ∆K(shear) =∆K/√ 2 for a complete slanted surface with ts=t/2, and for ts=0 it gives ∆K(shear) =∆K. A computer program was written to calculate ∆K(shear) as a function of the crack length for constant amplitude tests. Also da/dN versus ∆K is calculated, using the da/dN-∆K equation in the crack growth area without shear lips. Result of the prediction The prediction is tested on the crack growth rate result for R=0.1 as shown in figure 3. The slope of the results is about 5.5 for da/dN<0.1 µm/cycle, above which value it transforms to about 2.5 until da/dN =1 µm/cycle. From 1 to 10 µm/cycle the slope is again about 5. Above 10 µm/cycle static effects are probably superposed on the fatigue crack growth rate. For the calculation of ∆K the ASTM formula for a centre-cracked tension specimen (width w) is used: w a a K π π σ sec = (8)
RkJQdWJsaXNoZXIy MjM0NDE=