A shaft that has broken in service is not a mute object. The plane on which it separated, the texture across that plane, and the size of the region that tore last each record how the shaft was loaded and for how long. Before any hardness test or finite-element model, the fracture face narrows the question from why the shaft broke to a specific loading mode at a specific origin. Much of what follows in a competent investigation confirms what a careful first look already established.
Fracture angle before anything else
The first observation is geometric and needs no instrument: the angle of the fracture plane relative to the shaft axis. A fracture perpendicular to the axis indicates a normal-stress-driven failure, bending or tension. One inclined at roughly 45 degrees follows the plane of maximum shear and indicates torsion. Combined loading gives an intermediate or stepped surface, and the departure from either extreme roughly measures how the modes shared the load.
It is also the most fragile observation available. It is best made on the halves as recovered, in place where possible, and it is easily compromised by a saw cut placed for convenience, by secondary damage as the drivetrain came apart, or by rubbing between fragments still turning after separation.
Beach marks are not striations
The two are routinely conflated and belong to different scales. Beach marks are visible to the eye or at low magnification, and each records a change in loading or environment that shifted the crack front: a shutdown, a load change, a period at rest. Striations are microscopic, resolved only by electron microscopy, and in many materials one corresponds to a single load cycle.
Terminology matters once a report reaches opposing counsel. ASTM E1823 fixes the vocabulary for fatigue and fracture, and using its terms consistently keeps an examination from being attacked on definitional grounds rather than substance.
What the final-fracture zone says about stress
The last region to break is the material that could no longer carry the load, and its size relative to the whole cross-section is informative. A small final zone means the crack grew across most of the section before the remaining ligament let go, implying low nominal stress and long life. A large final zone implies the opposite: high nominal stress, comparatively few cycles.
The reading is qualitative rather than numerical, but it speaks to whether the shaft was working near its design load, and to whether a crack that long should have been found earlier.
Origins, ratchet marks and how sharp the notch was
Fatigue cracks start where local stress is highest. A single origin suggests a modest concentration with one dominant initiation site. Multiple origins around the circumference, separated by the small steps known as ratchet marks, indicate a severe stress concentration or a high nominal stress acting over the whole perimeter: a sharp fillet, a rough machined surface, a corroded band, the edge of a press fit.
Locating those origins precisely is not bookkeeping. Their position relative to the keyway, shoulder, hub seat or spline root ties the fracture to a geometric feature, and that connection is what an opposing expert tests hardest.
Rotating, reversed and unidirectional bending
Rotating bending, a constant-direction load on a turning shaft, puts every surface point through a full reversal each revolution and tends to produce origins distributed around the circumference with the final zone displaced away from the initiation side. Reversed bending on a shaft that does not turn tends to produce origins at two opposed positions. Unidirectional bending produces one.
The distinction is not academic. Endurance data for shaft steels is commonly generated under rotating-bending conditions to ISO 1143, so comparing a service fracture against it requires knowing which regime the shaft saw and how many cycles a given running time represents.
Torsional fatigue reads differently
Torsional fatigue is less familiar than bending fatigue and more often misread. Cyclic torque can drive a crack along the maximum-shear planes, giving a surface roughly parallel or perpendicular to the axis, or along the maximum-tensile plane at 45 degrees. Stepped, star-shaped and helical surfaces all occur, and which appears depends on material and stress level as much as on loading.
The single overload and what excludes fatigue
A shaft that failed on one torque spike, a jam, a locked rotor or a dropped load, shows a shear fracture with no crack-growth region: no beach marks, no discrete origin, and usually visible plastic deformation such as twisted machining marks or a deformed keyway. The absence of a progressive zone is a strong discriminator.
Where that torque came from is then the question, and it is almost always answered away from the shaft: at the driven equipment, at the coupling, or in the control scheme that permitted the event.
What handling destroys
The features this reading depends on are thin and easily lost. Fitting the halves back together to see how they mate rubs the faces and destroys detail at exactly the origin region of interest. Brushing or scrubbing strips corrosion products and deposits that carry their own information about how long the crack was open.
Preservation is unglamorous and effective: both halves kept apart and dry, coupled components left attached, keys and keyways retained, photographs taken before anything moves.
What the fracture face cannot settle
It establishes loading mode, origin location and, loosely, stress level. It does not establish applied torque, conformity of the material to specification, or the adequacy of the design allowable. Those need sectioning, hardness and chemical verification, dimensional inspection and a stress reconstruction, each guided by what the surface showed and none a substitute for it.
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.