6423 Fatigue Damage Unnotched Specimens

W. S. Johnson is considered a pioneer in the field of fatigue damage in uMMCs. From the early days of his Ph.D. thesis in 1979 [61] and up to 1988 [62], he suggested that the main cause of fatigue failure in any unidirectional composite system is the damage accommodated by the fiber (failure) and the corresponding loss of stiffness. Experimental observations in alumina-fiber-reinforced aluminum composites, conducted by the U.S. Air Force Materials Laboratory [63], convinced researchers that the most significant damage mechanism is extensive fiber damage, including multiple fractures of individual fibers. They found that even at the late stages of the fatigue life of a specimen, the fatigue resistance of a composite was still superior to that of an unreinforced matrix, since a sufficient number of broken fibers were still able to carry load quite effectively. The so-called matrix-dominated damage is based on the belief that the matrix material requires less strain than the fibers to initiate damage. With the introduction of SiC-reinforced Ti-matrix composites, this failure mechanism was more than ever verified. It was suggested in [64] that the superior endurance of the SCS-6/Ti-15-3 system is mainly controlled by the high fatigue limit of the SCS-6 fiber (approximately 1300 MPa [65]), as compared to an average strength of 3800 [66] or 4500 MPa [27]).

However, based on further observations, Johnson [57, 67] suggested that the fatigue failure of a SCS-6/Ti-15-3 and SCS-6/Ti-6-4 uMMCs occurs in a self-similar damage manner. This was supported by the similar endurance limit strains of the fiber and the matrix, and scanning electron microscopy (SEM) examination of fatigue fracture surfaces, which revealed low-level or negligible fiber pull-out (Fig. 6.11). Minimum fiber pull-out signifies fiber failure close to the crack plane. Furthermore, since the fatigue limit (approximately 600 MPa for the SCS-6/Ti-6-4 [68] and 650-700 MPa for the SCS-6/Ti-15-3 [69]) is significantly lower than the yield stress, the matrix may nucleate fatigue cracks without global yielding [70]. Johnson argued that since the strain at matrix fatigue limit is close to the fiber failure strain, matrix and fibers should fail simultaneously. Self-similar damage was also reported for boron/titanium uMMCs [71].

In 1991, subsequent studies of fatigue crack growth in SCS-6/Ti-15-3 and SCS-6/Ti-6-4 composites [72] indicated that fatigue failure does not occur in a self-similar damage manner because cracks were found to be bridged by intact fibers. In the same work, cracks were found to initiate from several different fabrication and manufacturing defects (broken fibers at edges, touching fibers, voids at the interface, etc.).

Even though extensive research has been conducted on crack initiation and growth of unnotched Ti-alloy-based MMCs [64, 68, 73, 74], most of the workers have agreed that the issue is quite confused since it involves the understanding of three basic parameters that could act individually as well as simultaneously. The first observation concerns the ratio of the applied strain to the time-dependent

FIGURE 6.11 SEM micrograph showing minimum fiber pull-out due to fiber failure close to crack plane for 32% SCS-6/Ti-15-3 [0]8 tested at 0max = 600 MPa, R = 0.1. The light gray area represents plasticity passage. (Photo taken from [69]).

fracture strain of the interface especially at medium stress levels. The second parameter is the tendency of the matrix material to initiate secondary cracks, particularly at the center of the specimen. The last parameter is the ability of the MMC to arrest these secondary cracks by constraining the crack micro-plasticity, FCE, or by producing adequate FB, and stress relaxation during bridging and debonding, respectively.

Furthermore, from experimental observations conducted on SCS-6/Ti-15-3 and SCS-6/Ti-6-4 smooth specimens (40% SCS-6/Ti-15-3 [0]6, 35% SCS-6/Ti-6-4 [0]6, 32% SCS-6/Ti-15-3 [0]8, and 32% SCS-6/Ti-6-4 [0]8 [52, 68, 72]), the fatigue damage behavior of both materials was classified into three distinct regimes depending on the maximum applied stress level.

At high peak applied stresses (about 80% of the quoted tensile strength), the fracture surface of the two types of SCS-6/Ti-15-3 materials were found to exhibit a flat morphology, that is, most of the fiber breakages were found close to the crack plane (no fiber pull-out). Fatigue damage was observed close to the fiber-matrix interfaces with a random distribution throughout the specimen (changes in the reaction layer thickness could develop stress concentrations [68]). Small number of secondary cracks were observed, which suggests that fatigue failure was mainly controlled by fiber breakage accumulation. Similar observations have been quoted for the two types of SCS-6/Ti-6-4 composites.

At medium stresses (40-80% of the tensile strength), the fracture surface of the SCS-6/Ti-15-3 was reported as irregular (similar to a tensile fracture surface [72]) and composed of several relatively flat fatigue cracking regions that extend from the specimen surface [72]. Significant fiber pull-out, with random distribution was also detected (Fig. 6.12). Cracks were found to initiate from broken fibers and

FIGURE 6.12 SEM micrograph showing fiber pull-out for 32% SCS-6/Ti-15-3 [0]8 tested at omax = 960 MPa, R = 0.1. (Photo taken from [69]).

interfaces, especially at the machined edges, while a large number of secondary cracks was detected at the specimen center [73, 68]. For the SCS-6/Ti-6-4 both reference sources confirmed a similar fracture surface to that of the SCS-6/Ti-15-3. However, a small number of secondary cracks at the specimen center were detected for the 32% SCS-6/Ti-6-4 [68] while no secondary cracking was found for the 35% SCS-6/Ti-6-4 [72]. This disagreement was attributed to differences on the reaction zone thickness (thicknesses of 1.7 and 2.43 ^m were quoted for the SCS-6/Ti-6-4 and SCS-6/Ti-15-3, respectively, in the as-fabricated condition [72]) and the interfacial shear strength [68]. Clearly, lower interfacial shear strength increases the number of broken interfaces and thus the probability of secondary cracking. Values of 124 MPa for the SCS-6/Ti-15-3 and 156 MPa for the SCS-6/Ti-6-4 were quoted, respectively [44].

At lower stresses, when the applied strain level to the composite is lower than the fracture strain of the interface, the fatigue damage pattern of the 35% SCS-6/Ti-15-3 was reported to be limited by matrix crack initiation at the specimen edges as a result of broken fibers due to machining while no secondary matrix cracking from broken interfaces was observed [72]. Also, metallographic inspection of the specimens revealed that after 106 cycles the fibers were still intact and bridged the cracked matrix. The same fatigue damage pattern was reported for the 40% SCS-6/Ti-6-4 [72]. However, in tests conducted on 32% SCS-6/Ti-15-3 at 600 MPa, cracks were found to grow from the reaction layer in the same manner as at higher stresses [68]. In the same work, cracks in the 32% SCS-6/Ti-6-4 were observed to grow not only at edges but also from "warts" on the fibers (see Fig. 6.13). Fiber warts were not observed in the SCS-6/Ti-15-3.

Cycling was found to degrade the tensile properties of both composites. After 106 cycles fatigue testing, the elastic modulus and tensile strength of the

FIGURE 6.13 Warts on fiber in 32% SCS-6/Ti-6-4 [0]8. (Photo taken from [68]).

35% SCS-6/Ti-15-3 were measured as 130 GPa and 1103 MPa, respectively, compared to initial values of 210 GPa and 1572 MPa [72]. For the 40% SCS-6/Ti-6-4 a similar degradation of the tensile strength was reported (postfatigue value of 1034 MPa as compared to an initial value of 1572 MPa [72]). However, cycling was found to produce lower degradation of the elastic modulus for the SCS-6/Ti-6-4 composite (initial value of 213 GPa; fatigued value 193 GPa [72]). Considering that for the same load history both materials have approximately accumulated similar crack length, differences on the degradation of the elastic modulus can only be explained by differences in the number of bridged fibers that failed during cycling.

It should be noted that the above findings do not represent a universal picture of the fatigue behavior of uMMCs for a number of reasons. The most critical are:

1. Most of the tests were conducted on strip or dogbone specimens. These specimens, even after careful polishing at the edges to minimize the effect of coarse finish, are vulnerable to additional crack initiation and therefore may underestimate the true fatigue life of the material when compared to traditional circular section specimens [68].

2. There is a minimum amount of data about the effect of load ratio [75, 76].

3. Control mode (strain-controlled or load-controlled), especially at different stress ratios, shows in most of the cases an unpredictable behavior [75].

4. The material's behavior under tension-compression loading cannot be fully appreciated since their typical thickness is about 2.5 mm and therefore are unable to withstand significant compression loads [75, 77].

0 0

Post a comment