Three of the mechanisms that end a turbine blade's life leave damage in much the same place and, to the unaided eye, in much the same form: a crack in the airfoil. Creep, high-cycle fatigue and thermal-mechanical fatigue are nonetheless different failures pointing at different questions — material and temperature exposure, vibration and resonance, and operating cycles and cooling design respectively. Separating them is the forensic work, and the separation is made on microstructure, fracture-surface morphology and dimensional evidence rather than on where the crack happens to sit.
Crack location narrows very little
Cooling-hole edges, leading and trailing edges and the airfoil root concentrate stress for every one of these mechanisms. A crack at a cooling hole is consistent with thermal-mechanical fatigue, with high-cycle fatigue and with creep alike, so location alone rarely decides anything.
What does separate them is time dependence. Creep accumulates with hours at temperature and stress. High-cycle fatigue accumulates with vibratory cycles at frequency. Thermal-mechanical fatigue accumulates with the number and severity of thermal transients. Each mechanism runs on its own clock, and each leaves a record of that clock.
Creep is written in the microstructure
Creep damage is distributed rather than local. A metallurgical cross-section shows grain-boundary cavitation, elongated grains and, in nickel-base superalloys, coarsening or rafting of the strengthening precipitates. Field replication under ASTM E1351 captures that without destroying the component; sectioning, preparation and etching under ASTM E3 and E407 settles it.
Because the damage is distributed, adjacent unfailed blades from the same row carry the same exposure. Comparing the failed blade against its neighbours is what distinguishes a row-wide thermal exposure from a single anomalous blade.
Dimensional evidence and the creep clock
Creep produces permanent strain. Airfoil length, twist and profile measured against the original design geometry quantify deformation that fatigue does not produce at all, which makes coordinate measurement one of the few genuinely discriminating tests.
Remaining rupture life is estimated from creep and stress-rupture testing under ASTM E139 on material taken from the failed part or a sister blade, then projected to service conditions through a time-temperature parameter. That projection is where the opinion is usually contested: how far the extrapolation reaches, and whether the tested material represents the service microstructure.
High-cycle fatigue and the excitation behind it
High-cycle fatigue initiates at a discrete surface origin and records itself as fine striations, beach marks and ratchet marks. Spacing is very fine because the stress amplitude is small and the cycle count enormous — a blade accumulates millions of cycles in hours of running.
The engineering question is what supplied the excitation: nozzle passing frequency, partial admission, a stall cell, a rub, or a fouled or damaged upstream row. Comparing blade natural frequencies against operating speed and excitation orders on a Campbell diagram shows whether a resonance sat inside the operating range. Vibration records kept to ISO 20816 and trip history show whether the machine actually dwelt there.
Thermal-mechanical fatigue counts starts, not hours
Thermal-mechanical fatigue is driven by a small number of large-amplitude strain cycles from startup, shutdown, trip and load transient. Origins cluster where thermal gradients are sharpest — cooling-hole edges, trailing edges and thin sections — and the striations are coarser and far fewer.
The oxide on the fracture faces is itself evidence. A crack that opened and then dwelt at temperature between cycles oxidises internally; a fracture produced by ambient-temperature vibratory loading does not. Laboratory characterisation under ASTM E2368 for strain-controlled thermomechanical fatigue, and ASTM E606 for strain-controlled low-cycle fatigue, supplies the material behaviour against which the service damage is read.
Pure cases are rare
Real blades frequently show more than one mechanism. Creep-weakened material fails at lower fatigue stress. A thermal-mechanical crack initiates and vibratory loading propagates it. An oxidising crack tip advances by a combination of the two.
The useful question is therefore not which single mechanism was present but which one initiated the crack and which one controlled the final instability. An opinion asserting a single pure mechanism where the evidence shows interaction is the one most likely to come apart.
The operating record has to agree
Fired hours, starts, trips, load profile and hot-section inspection intervals are checked against the manufacturer's published life limits. A creep opinion needs hours at temperature. A high-cycle fatigue opinion needs an excitation source and evidence of dwell. A thermal-mechanical opinion needs a transient count.
Where the metallurgy says one thing and the operating record says another, one of them is wrong. Resolving that conflict is part of the analysis, not a footnote to it.
What earlier inspection did and did not see
Fluorescent penetrant examination under ASTM E1417 covers nickel-alloy hot-section blading; magnetic particle examination under ASTM E1444 applies where the blading is a ferromagnetic martensitic stainless, as much steam-path blading is. Borescope reports fill the gaps between outages.
Detection limits matter as much as the results. An indication recorded and dispositioned, an interval extended, or a repair performed are all findings. So is the absence of any recorded indication — but that is not proof no crack existed, since a subsurface or coating-covered crack can sit below the method's threshold.
Where these opinions are challenged
Predictably: that the replica came from a location unrepresentative of the hottest section; that the rupture-life extrapolation reached beyond a defensible range; that the frequency analysis used design values for a blade whose stiffness had changed; and that a striation count assumed a uniform propagation rate.
Work that states which piece of evidence carried which conclusion, and what observation would have falsified it, survives that scrutiny.
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.