Now we have come to the task of looking for features that might tell us how the crack happened on the shaft. Note that this is different from determining WHY the shaft broke. Since this part is long gone and we don’t have any background, we will never be able to know with any specific detail WHY the shaft broke. A “smart aleck” and technically correct answer to why the shaft broke is that there was some position on the shaft where the stress exceeded the strength at some point in time. This allowed the crack to initiate. Once a part is cracked, if it continues to experience stresses, the stresses at the crack tip will be greater than the anticipated design stresses. So, it is easier to grow a crack than to initiate one.

When we are focused on the physical aspects of “why,” it is always critically important to try and understand what happened at the crack initiation. Sometimes, there will be a weakness of some sort that facilitated the initiation. If there is no sign of a weakness at that location, then that might indicate that the stresses were unusually high. This is a line of reasoning that is a solid foundation of fracture analysis. Similar lines of reasoning can be invoked using corrosion resistance in the case of a corrosion failure.

Now back to how the part broke. The alternating dark and light bands across the surface of the crack are classical for cracks that propagate over a long enough time that the environment has time to change. Humidity variation at the moment that the crack is growing its next increment will influence the color of the oxide that forms on the newly exposed crack surface. This is why fatigue cracks that happen “in the field” tend to have these bands of varying color. Fatigue cracks that happen in an air-conditioned lab only have these alternating bands of color when the loading conditions vary over time. These load-variation bands may be easy or difficult to see.

The crack surface of the broken shaft appears quite flat. While we don’t have any other views showing the curvature, we see what appears to be a groove-like feature that the crack is associated with all around its circumference. We also see that the surface texture is quite uniform, aside from the slight variations in gray scale from band to band. The very bright areas near the 9 o’clock position as shown and along the arc from 2-5 o’clock are most likely scraping damage that happened after the final separation of the two fragments.

The combination of a reasonably flat fracture surface and a very smooth fracture surface is characteristic of a very low nominal stress having facilitated the crack growth (not the only possible explanation, but characteristic of!). Note that I don’t say “caused” the growth. We just don’t have enough information to say what the “cause” was. The smoothness of the crack is due to the crack surfaces rubbing against each other as the crack opens a tiny bit and then closes back on itself.

Further, we can say that whatever combination of stress and environmental conditions (humidity and temperature) were present may have cycled from one set of parameters to another, and perhaps back to the first. This is because we see the darker gray at the top of the image and the bottom, with a lighter-color gray in the middle. The lighter-color area also shows a finer alternating dark and light pattern within it. Sometimes operating personnel or maintenance logs can confirm such a speculation. Maybe some event happened and the shaft was not visibly damaged. But now, 10 months or 18 months later, the shaft breaks completely apart. Perhaps review of maintenance logs would provide evidence for some other scenario for the timing of the crack.

But “playing with” the stated timing, one might speculate that the crack happened over a time period approaching or on the order of a calendar year in a place that had seasonal humidity and temperature variations. The fact that the roughness is so uniform tells us that there really was not much change in the loading. This would be typical for many machinery applications.

Let’s continue our discussion next time.