In a post two weeks ago (here) I asked the question: “Are dual gearbox failures, or similar failures, more likely now that one has occurred?” And then I went on to say: “What is in the category of “similar failures”?”
This is a key principle in sustainment, especially for your assessment experts. Artificial intelligence has not yet progressed to the point where this kind of abstract pattern can be detected by big data algorithms. It requires human expertise in assessment analysis and system knowledge.
If your gearboxes are failing because specific gear axles are failing, for instance. What other parts in your system were manufactured with that kind of material manufactured in that particular way? If your gearboxes were failing because of insufficient repairs at the depot then what other repairs were being performed in an insufficient manner? If your gearboxes were failing because they were lot 6 gearboxes and all the other earlier gearboxes are OK, then what changed in lot 6? Materials? Processes? Inspections?
This kind of analysis is always complicated by lack of data and information. Some gets lost over the decades. Some never existed to begin with. But since the goal is to detect and mitigate threats to the mission lead time ahead, any little clues help. And sometimes the clues consist of emerging failure modes that point to other potential failures lurking in the shadows.
The moral of the story is: encourage and grow your assessment and system experts.