A shaft rarely breaks in a plain length of uniform section. It breaks at a keyway, a shoulder, a fillet, a snap-ring groove, a cross-hole or the edge of a press fit, anywhere the section changes and stress crowds into a small volume of material. That much is expected and is designed for. The forensic question is narrower and harder: whether the concentration at the origin was the one the designer allowed for, one introduced in manufacture, or one created by a repair that nobody recorded.
Concentration is designed for, not designed out
Every practical shaft carries discontinuities because it has to transmit torque through something. The design response is not to eliminate them but to quantify them, applying a stress-concentration factor appropriate to the geometry against the fatigue allowable rather than the static one. A shaft can be entirely adequate in static strength and still be deficient in fatigue at the same feature, which is why static rating calculations rarely resolve a fatigue failure.
Theoretical versus effective concentration
Handbook concentration factors are elastic and purely geometric. What a real shaft experiences is a lower effective value, reduced by the material's notch sensitivity, and a tougher lower-strength steel is more forgiving of a sharp notch than a hard high-strength one. The practical consequence is counterintuitive and frequently disputed: substituting a stronger material into an existing sharply notched design can shorten fatigue life rather than extend it.
Fillet radius: the drawing versus the part
The most common finding at a shoulder-origin fracture is a fillet radius smaller than the drawing called for, or a fillet reworked, ground flat or undercut during machining or a later repair. The relationship is steep, since the concentration factor climbs sharply as the ratio of radius to diameter falls, so a departure that looks trivial on a caliper can be significant in stress.
This is measurable evidence rather than opinion. Radius gauges, optical comparators and profile scans on the as-found part, compared against the released drawing and its revision history, either establish the departure or exclude it.
Keyway geometry and the end of the keyseat
A keyway concentrates stress at its corner fillets and, more severely, where the keyseat ends. A profiled end-milled keyseat terminates abruptly and is the harsher of the two common forms; a sled-runner keyseat runs out gradually and is milder. Dimensional conventions for keys and keyseats are set out in ASME B17.1, and departures such as an unradiused corner or a keyway extended in the field to suit a replacement hub are recoverable by measurement.
Where the origin sits relative to the end of the keyseat is therefore worth locating precisely, because it distinguishes a keyway that was merely present from a keyway that was governing.
Surface finish and the marks machining left
Fatigue cracks start at surfaces, and roughness shifts the endurance limit measurably. Circumferential turning marks in a fillet, chatter, a grinding burn or a stray tool witness line all act as small notches superimposed on the designed geometry. Surface texture is specified and measured under ASME B46.1 and its ISO equivalents, so the as-found finish can be compared directly with what the drawing required.
Grinding burn deserves separate mention because it does two things at once. It alters the local microstructure and it can leave tensile residual stress in precisely the layer where a fatigue crack would otherwise struggle to start.
Residual stress cuts both ways
Compressive residual stress at the surface, from shot peening, fillet rolling, nitriding or induction hardening, suppresses crack initiation and is often what makes an aggressive shaft design viable in the first place. Its absence is then a finding in itself. Where a drawing specifies peening or rolling and the part shows none, or where a repair grind removed the treated layer, the shaft in service was not the shaft that was designed.
Material and heat treatment are separate questions
Verification is routine and its methods are settled: composition by chemical analysis, tensile properties under ASTM E8/E8M, toughness by Charpy under ASTM E23, and hardness by Rockwell under ASTM E18 or microindentation under ASTM E384 where a case-to-core traverse is needed. Sections prepared under ASTM E3 and etched under ASTM E407 reveal decarburization, case depth, inclusion content and quench cracking.
These tests answer whether the material met its specification. They do not answer whether that specification was adequate for the duty, which is a design question argued on separate evidence.
Press fits and the concentration nobody inspects
At a shrink fit or a spline the concentration is not purely geometric. Contact pressure at the edge of the fit, combined with micro-slip under cyclic load, produces fretting damage and initiates cracks well below the smooth-bar endurance limit of the same steel. Because that damage sits under the hub, no in-service inspection short of disassembly will find it, and a shaft can be within tolerance on every dimension and still fail there.
Design, manufacture, or neither
The distinction the parties care about is seldom settled by origin location alone. It turns on whether the as-built geometry matched the design, whether the allowable already contained the concentration factor for that feature, and whether the duty the shaft actually saw was the duty the design assumed. Those are three separate records, drawings and revisions, design calculations, and operating history, and a finding that reaches only one of them is incomplete.
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.