When a blade liberates in a running machine, the damage that follows is far more visible than the damage that started it. Downstream rows are battered, the casing is marked, and the originating airfoil may survive only as fragments. Yet almost every question that matters afterwards — defect, ingestion, resonance, missed inspection or exceeded life limit — depends on establishing which blade failed first and by what mechanism. The reconstruction is a sequencing problem, and it is won or lost in the hours immediately after the machine is opened.
The loudest damage is not the origin
A single liberated airfoil delivers a great deal of energy into the rows behind it, and those secondary failures are usually more dramatic than the first. Investigations that start from the most destroyed component routinely start from a victim.
The origin is found by fracture morphology, not by severity. It is the one blade whose fracture surface shows a crack that grew before the event, rather than one that separated instantaneously.
Map it before anything moves
Every fragment should be recovered and tagged with its as-found location by stage and by circumferential position, alongside photographs of the section as opened and any borescope imagery taken before disassembly. Witness marks on casings, shrouds and diaphragms record trajectory and are destroyed by cleaning.
Fragments discarded as scrap cannot be recovered later, and the missing one is frequently the one carrying the origin. Preservation here costs very little and is the difference between a sequence that is established and one that is argued.
Origin blade or impact victim
A progressive fracture — a distinct initiation site, beach marks, arrest lines, a flat thumbnail region followed by a rougher final zone — indicates a crack that existed before the event. Impact-driven separations show overload features instead: shear lips, gross plastic deformation, torn edges and no progressive region at all.
Ratchet marks and the direction of the fracture markings point back toward the initiation site, which is what allows an origin to be located even on a fragment recovered in pieces.
Foreign object or domestic object
Foreign object damage arrives from outside the machine: ingested hardware, tooling left in the gas path, birds or ice in aero service, debris from upstream ducting. Domestic object damage comes from a component the machine shed itself, and it is a consequence of an earlier failure rather than a cause.
The two are separated by the chemistry of embedded material under energy-dispersive analysis, by the condition of upstream rows and inlet hardware, and by where in the flow path the first impacts appear. Calling domestic debris foreign reverses the whole causal chain.
Casing, containment and energy
Whether the casing retained the fragments, and where it was penetrated or deformed, records both trajectory and the energy released. In industrial machines the applicable framework is the relevant equipment standard — API 616 for gas turbines, API 612 for special-purpose steam turbines, API 617 for axial and centrifugal compressors — together with the manufacturer's own casing design basis.
Aero engines are different in kind, because blade containment is demonstrated by test as part of engine certification under the FAA and EASA engine airworthiness requirements. A release that escaped containment in that context raises a certification and continued-airworthiness question rather than only a mechanical one.
The disc and the root attachment
Blades do not always break in the airfoil. Fir-tree and dovetail attachments fail by fretting fatigue and by root cracking, and a blade that came out of its slot presents a different problem from one that fractured mid-span.
The disc also determines what happens next commercially. A rotor with cracked or deformed slots is not returned to service on a blade replacement, so establishing whether this was a blade event or a rotor event drives the repair scope, the outage length and the size of the business-interruption claim.
Recorded data narrows the sequence
Vibration monitoring maintained to ISO 20816 often shows the moment of liberation as a step change in amplitude and phase, and may show a rising trend before it. Control system logs, sequence-of-events records, exhaust temperature spread and performance data bound the timing independently.
Sampling rate governs how much of this is usable. A one-minute average shows that something happened; a high-resolution record can distinguish a progressive degradation from an instantaneous release, and that distinction often decides whether an earlier intervention was available.
Intervals, life limits and attribution
Manufacturers publish hot-section inspection intervals and part life limits in fired hours, starts or equivalent operating hours, and service bulletins amend them. Aero practice is stricter again, with life-limited parts managed under published mandatory replacement intervals and continued-airworthiness obligations.
Attribution turns on whether the interval was met, whether it was extended and on what basis, whether a bulletin was actioned, and whether the counting method used matched the manufacturer's definition. Calendar hours substituted for equivalent operating hours is a common and consequential error.
Where these opinions are challenged
Predictably: that the origin was assigned to the most damaged blade rather than established fractographically; that not all fragments were recovered, so an alternative origin cannot be excluded; that the sequence rests on data too coarse to resolve it; and that life-limit compliance was assessed against the wrong counting convention.
An opinion that identifies the origin, states the evidence that fixes it, and says candidly which fragments were never found is far more durable than one that presents a sequence as though nothing were missing.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.