Surface finish decides fatigue life
Where cracks start, and why polishing is a structural operation.
A fatigue crack almost always begins at a part's surface rather than somewhere inside its bulk, because ordinary machining marks, tool lines and scratches left on that surface act as tiny, microscopic stress concentrators, and it is at exactly those tiny concentrations that the cumulative damage described in the previous article in this set actually gets started, which means the finish left on a surface is not a cosmetic detail, it is a genuine structural property of the part.
Why the surface matters more than the bulk
The interior of a well-made part is comparatively smooth and uniform at the microscopic scale, but its surface, wherever a cutting tool has passed, wherever a part has been ground, drilled or simply handled, carries a landscape of tiny peaks, valleys and marks invisible to the naked eye and yet large enough, relative to the microscopic scale fatigue actually operates at, to concentrate stress locally far above the average level the rest of the part experiences. A stress concentration does not need to be a dramatic, visible notch to matter for fatigue, since the same local stress-raising effect that a sharp corner produces on a large, visible scale is produced in miniature by an ordinary rough machined surface, and fatigue damage accumulates fastest wherever local stress happens to be highest, which is almost always at one of these small surface irregularities rather than anywhere in the smoother material beneath them.
The chipped-windscreen comparison
A small stone chip on a car windscreen looks harmless the day it happens, a tiny mark easy to ignore on an otherwise intact sheet of glass, but weeks of ordinary temperature swings and road vibration afterward, a crack very often begins growing outward from that exact chip and nowhere else on the windscreen, because the chip's rough, damaged edge concentrates stress in a way the smooth glass surrounding it never does. The glass elsewhere on the windscreen is seeing the identical daily temperature swings and the identical vibration, and it does not crack, because it has no equivalent flaw to concentrate that ordinary stress into something locally severe enough to start a crack growing. The chip did not weaken the whole windscreen, it simply gave the same ordinary stress everywhere else on the glass one specific place to finally take hold. A machined part's rough surface finish is covered in thousands of miniature versions of that same stone chip, and fatigue cracking, when it eventually happens, begins at whichever one of those tiny surface flaws happened to be the most severely stress-concentrating rather than at any average location across the part as a whole. Predicting exactly which one will win that quiet competition is rarely possible in advance, which is precisely why the finish is controlled everywhere rather than only at whichever single spot looks the most suspect. A windscreen repair kit works on exactly this principle rather than by making the glass stronger overall, injecting clear resin into the chip to fill its sharp, damaged edges and restore a smoother transition back to the surrounding glass, which is why a chip repaired promptly, while it is still small, stops a crack from ever finding the concentrated stress it needed to start growing in the first place.
Why polishing changes the outcome without changing the part's size
Polishing a surface after machining removes or smooths over the sharpest of those microscopic peaks and valleys, lowering the local stress concentration at the surface without changing the part's overall dimensions, material, or bulk strength in any way a strength calculation would notice. Because fatigue cracks overwhelmingly start at the surface rather than in the bulk, this purely cosmetic-looking change can measurably extend how many cycles a part survives before a crack finally initiates, sometimes dramatically, which is why aerospace and other fatigue-critical components frequently specify a controlled surface finish, or a deliberate surface treatment such as shot peening that leaves the surface under a beneficial compressive stress, as a genuine structural requirement on the drawing rather than as an optional finishing touch left to whichever process happened to be convenient. Shot peening achieves its benefit by bombarding the surface with a stream of small, hard particles, each impact leaving a tiny dimple and, more importantly, leaving the material immediately beneath that dimple in a state of residual compression, and because a fatigue crack can only grow by being pulled open under tension, a surface held in compression has that compression to overcome first before a crack even has the chance to begin opening at all.
One figure worth keeping in mind
A rough, as-machined surface can reduce a part's fatigue life by a large fraction compared with the identical part properly polished, sometimes cutting survivable cycles by more than half, a difference achieved with no change whatsoever to the part's material, cross-section or nominal strength, purely by controlling how sharply the surface itself concentrates stress at the microscopic scale where every fatigue crack actually begins.
Where the effect matters more, and where it matters less
A part loaded well below any of its material's fatigue-relevant thresholds barely notices its own surface finish, since there is so little cumulative stress being placed on it cycle after cycle that even a locally concentrated version of that small stress rarely amounts to much across a realistic service life. The closer a part's ordinary working stress sits to the edge of what its material and cycle count can actually tolerate, the more that surface finish stops being a minor refinement and starts being one of the most consequential decisions on the entire drawing, since it is operating exactly where the previous article in this set showed damage actually accumulates fastest, at the highest local stress a part experiences anywhere on its surface.
Why this matters in practice
A part redesigned to remove material, exactly the kind of change the rest of this set has spent real effort defending, deserves its surface finish reconsidered alongside its geometry, since a thinner, more efficient part carrying higher stress for its size is also a part where any given surface flaw now concentrates that already-higher stress even further, and the margin that finish alone provides can be the difference between a lean, successful redesign and one that fails in fatigue at a fraction of its intended service life. A specification calling out surface finish explicitly, rather than leaving it to whatever a given machine happens to produce by default, costs comparatively little against the material savings a leaner design has already earned, and it is very often the cheapest structural improvement available on the whole drawing.