Abstract
Work published for the first time at the ASME PVP 2017 conference showed that when on the upper-shelf, the toughness measured directly from surface-cracked pipe tests decreased as the flaw depth increased. A similar trend existed in SENT tests. Initially it was found that this flaw depth sensitivity of the toughness occurred for a very tough material like TP304 stainless steel. The significance of that result was that even for a material where limit-load was thought to exist, as the flaw depth increased the toughness dropped appreciably, and the failure analysis mode changed from limit-load to elastic-plastic fracture. Experimentally, this made sense because it explained the observed phenomena of load-controlled leak-versus-break behavior for circumferential surface-cracked pipes (as will be shown for several pipe tests), but that LBB behavior is not predictable from circumferential flaw limit-load analysis.
Furthermore, the flaw depth effect on toughness also exists for axial surface cracks and even in flat plates with surface cracks. For axial surface cracks the implication was that the long-used empirical surface-crack bulging factor from Maxey/Kiefner (incorporated in many international codes and standards) actually incorporated both the bulging factor and the toughness changes with flaw depth. Because of the change in toughness with flaw depth, when using detailed finite-element fracture analyses for the crack-driving force it is possible to have more error in the failure stress predictions if a constant toughness is assumed for all surface-flaw depths. In fact, in another paper in the ASME 2019 PVP conference it will be shown that the toughness in a wrought TP304 elbow at crack initiation of a circumferential surface crack that was 68% of the thickness was about 1/3rd of the toughness from a standard 1T CT specimen made from the same material. Those results will also be reviewed.
Similar results of toughness decreasing with flaw depth in surface-cracked pipes and SENT specimens for various materials over a large range of strain-hardening behavior will show the toughness decrease trend with flaw depth is consistent.
To understand these trends more theoretically, 3D FE analyses were also conducted for one initial set of TP304 SENT specimens with a wide range of a/w values (0.3 < a/w < 0.9). The initiation toughness decreased by a factor of 5 to 6 as the crack depth increased; however, the Q value coinciding to the load at the start of ductile tearing was constant for the wide range of a/W values. Q at the start of ductile tearing in the SENT (Qi) was more consistent at normalized distances from the crack tip, rσo/J that were in the range from 0.25 to 1.5 rather than just the popularly considered rσo/J = 2. Hence, by having one SENT test result with a single a/W value, the Ji value for any other a/W can then be calculated. This is consistent with the experimental trends to date, but unfortunately Ji was found to be not proportional to the Q values as is conventionally assumed by many researchers at this time.