Results The results from 11 constant load amplitude tests on material AA 5083 in air are shown in figure 1. It can be noticed that at about da/dN = 0.1 µm/cycle, the crack growth rate curves change in slope. Observations of the different fracture surfaces revealed that at the corresponding ∆K-values the crack appearance started changing from flat to slant. At first the change in slope was attributed to a possible different crack closure situation due to the growing shear lips. But it was shown that suppression of shear lips leads to the same change of slope, thus extra closure due to shear lips probably doesn’t exist in AA 5083. As the start of shear lips corresponds with different Kmax-values (e.g. Kmax = 8.8 and 13 MPaŒm respectively for R = 0.1 and 0.5), it can be concluded that the start of the shear lips is also probably not a matter of plane stress/plane strain. The only obvious correlation is with the crack growth rate or, which is equivalent, with the ∆Keff. This is in agreement with observations on AA 2024 [1]. In seawater the change in slope starts at about 0.2 µm/cycle, at 10 Hz [2]. Crack closure relations For AA 2024 the well-known relation of Elber will be used, with U=0.5+0.4R for – 0.1 < R < 0.7. This relation fits the results for AA 5083 not very well. Crack closure for this alloy is not measured directly, but it is found by correlating 7 of the 11 constant amplitude da/dN versus ∆K results shown in figure 1. One test at R=0.7, one test at R=0.5, two tests at R=0.1, one test at R=-0.25 and two tests at R=-1. The best quadratic crack closure function U= a+bR+cR2 is found by taking all combinations of a, b and c and calculating ∆Keff=U∆K for all measurement points of the 7 tests. A power formula of type da/dN=C ∆Keff m is fitted through the da/dN-∆Keff points. The combination of a, b and c with a maximum value of the correlation coefficient, based on log(da/dN) and log(∆Keff), is considered the best combination. The combination is normalized, i.e. the normalized coefficients are found by dividing the best calculated coefficients by their sum, meaning that U=1 if R=1. The best crack closure formula found in this way for this material is U=0.66+0.32R+ 0.02R2. The formula for the best power fit is given in equation 1, where da/dN in (µm/cycle) and ∆Keff in (MPaŒm): da/dN=0.70*10-3 ∆Keff 2.8 (1) However the results in figure 1 show that the curves for R=0.5 and 0.7 are about the same, meaning that crack closure is not important above R=0.5. If it is assumed that U=1 already for R>=0.5, then this constraint can be added to the calculation procedure. The result of the calculation is now U=0.80+0.39R+ 0.03R2 for R<=0.5 and U=1 for R>0.5. With this constraint da/dN can be calculated as: da/dN=0.41*10-3 ∆Keff 2.8 (2) The same crack closure calculation was also applied using only datapoints without shear lips, i.e. for da/dN<0.1 µm/cycle. For the constraint situation with U=1 for R=1, the result is U=0.65+0.32R+0.03R2 and da/dN=0.80*10-4 ∆Keff 4.16 (3) The da/dN-∆Keff fit result for the constraint situation with U=1 for R=1, eq.1, is also shown in figure 1 for all data points at different R. A parallel shift is applied to better show the result, i.e. the corresponding ∆Keff results have to be multiplied by a factor of 2. The geometric crack growth prediction model Although the start of shear lips is not the cause of the change in slope of the log(da/dN)- log(∆K) line, it seems to be a consequence of another, not yet known (micro)mechanism, that results in both shear lip start and growth and also in a lower slope of the log(da/dN)-log(∆K) relation. The shear lip width is chosen to represent this unknown mechanism. It was shown in
RkJQdWJsaXNoZXIy MjM0NDE=