Fatigue cracks initiate at stress concentrations through cyclic slip creating surface roughening and intrusions/extrusions. Persistent slip bands form from cyclic plastic deformation and act as crack nucleation sites. Initiation typically occupies a significant fraction of fatigue life and depends strongly on surface finish, stress concentration, and microstructural features.
Examine micrographs of fatigue-initiated surfaces to observe slip-band patterns and surface roughness. Conduct fatigue tests on notched versus smooth specimens to quantify stress concentration effects on initiation life.
Fatigue is not purely stress-controlled. Fatigue initiation depends on cyclic plastic strain amplitude, not merely stress amplitude, and is controlled by low-cycle fatigue mechanics below 104 cycles.
Fatigue failure is insidious because it occurs under stresses well below the static yield strength — stresses that, applied once, would cause no visible damage at all. The key is *repetition*. Each loading cycle causes a tiny increment of irreversible plastic deformation, even when the globally applied stress is elastic. Over thousands or millions of cycles, that accumulated microscale damage nucleates a crack, which then grows until the remaining section can no longer carry the load and fracture occurs suddenly. Understanding initiation — the first stage of this process — is critical because it typically consumes the majority of a component's fatigue life.
From your study of stress concentrations, you know that geometric features like notches, holes, and fillets amplify the local stress far above the nominal applied value. The same geometry that concentrates stress also concentrates cyclic plastic strain. Even when the bulk of the material deforms elastically, the highly stressed region at a stress concentration may experience small-scale plastic flow on every cycle. This cyclic plasticity is not randomly distributed — it localizes onto specific crystallographic planes called persistent slip bands (PSBs). These bands form because certain slip systems reach a self-organized steady state of repeated, concentrated back-and-forth shear. The material in the PSB deforms far more than the surrounding matrix, and the band "persists" even after annealing attempts, which is why they're called persistent.
The surface where PSBs intersect the free face is where initiation actually happens. Repeated slip along the band pushes material out of the surface on one stroke and pulls adjacent material in on the next. Over many cycles, this creates extrusions (ridges above the surface) and intrusions (grooves below). An intrusion is geometrically equivalent to a crack embryo: it is a sharp re-entrant notch at the surface that concentrates stress on subsequent cycles. Once the intrusion reaches a depth of roughly a grain diameter, it transitions from Stage I crack growth (shearing along the slip band, ~45° to the stress axis) to Stage II growth (tensile crack opening, perpendicular to the maximum principal stress), and the crack propagation phase begins.
Practical design implications follow directly from this mechanism. Because initiation is a surface phenomenon, surface finish matters enormously — a polished surface has dramatically longer initiation life than a rough machined one. Compressive residual stresses at the surface (from shot peening, case hardening, or roller burnishing) oppose the opening of intrusions and suppress initiation. Stress concentration factors directly reduce initiation life, which is why smooth transitions and generous radii in fillet geometry are specified even when static stress calculations show large margins. The distinction in your misconceptions section is worth internalizing: below roughly 10⁴ cycles, stresses are high enough that macroscopic yielding occurs and strain-based design methods apply; above that, the high-cycle regime is governed by stress amplitude, and the initiation mechanism described here dominates the total life.