Figure 3 has a more classically shaped thumbnail at the initiation, but the thumbnail area is not as distinctive-looking compared to the rest of the crack surface. This part, on the other hand, has a distinct and significant stress concentration due to the change of diameter. According to the theory stated in part 1, we’d expect the curvature of the crack to reverse. But, sadly, the crack changed to overload when the crack was still small (small shear lips along edge of circular cross section to the upper left of the marked “fatigue area”), so we never had a chance to see if the crack would have changed direction.

Figure 4 shows a bolt that broke at the first thread due to (probably) combined effects of fatigue and corrosion (perhaps hydrogen embrittlement). To the lower right of the pink dashed line, the crack is smooth, and this is the “thumbnail.” Again, the stresses were high relative to the strength, so the part broke in sudden fashion after it passed the pink dashed line. Bolts inherently have stress concentrations in the thread roots.

Figure 5 shows another bolt that has a faintly visible thumbnail (highlighted with the black down-pointing arrow and the thin grey line). The gray upward-pointing arrows show the multiple ratchet marks that tell us that the crack started along this edge. The beach marks (or crack-arrest lines) are invisible until just below the small crescent area marked final fracture. There we see some very faint light and dark bands.

The crack-arrest marks are essentially invisible because the fatigue load was constant and the machine was operating in an air-conditioned lab. One other thing to note is that the final separation area is extremely small. This tells us that there was a large safety factor relative to the operational stresses. My student questioner asked about this too. Why would a crack start at all if only 2% of the cross section can keep the part nominally (if temporarily) intact?

This might indicate a higher load early on that damaged the part without creating an actual crack. Fatigue-crack “precursors” are microscopic.

We get back to the curvature issue next time in part 3.