🔍 Failure Analysis: Why Things Break
Investigate how and why structures and parts fail — and why the famous story of a disaster is usually the wrong one. You'll read fatigue, corrosion, and overload signatures off a fracture surface, the
What you’ll learn
- The Broken Piece Is a WitnessExplain what failure analysis is and why it reads evidence rather than narrative.A broken part is a witness: the fracture surface records where a crack started, how it travelled, and how fast. The analyst works backward from the origin to the mechanism, and treats 'whose fault' as a separate, later question. The course's argument: the popular story of a famous failure is usually wrong, and the wrong story blocks the fix.
- "Strong" Is the Wrong QuestionDistinguish stiffness, yield strength, ultimate strength, and toughness — and explain why toughness is what gives warning.'Strong' hides at least four different properties, readable from a stress–strain curve: slope (stiffness), where it bends (yield), the peak (ultimate), and the area beneath (toughness). Glass is strong and stiff but absorbs almost no energy; mild steel yields sooner and is far tougher. Nearly every disaster in this course involves a structure that was strong enough and not tough enough.
- Reading a Fracture SurfaceRead a fracture surface for the ductile-versus-brittle signature and use chevron marks to locate an origin.Ductile fractures show visible deformation, a dull fibrous surface, cup-and-cone geometry in round bars, and microscopic dimples. Brittle fractures show no deformation, a bright faceted cleavage surface, and chevron marks that point back to the origin. Crucially, ductile and brittle are conditions, not labels — temperature, loading rate, and notches flip the same steel between them.
- Fatigue: Failure on the Installment PlanExplain fatigue: why parts fail below yield stress, the three stages, and how to read beach marks and striations.Fatigue is the most common cause of in-service mechanical failure: cyclic loading grows a crack at a stress the material survives statically. It runs in three acts — initiation (often most of the life), slow propagation, then final overload in milliseconds, which is the only part witnesses see. Beach marks curve around the origin and striations record one ripple per cycle, so the surface reveals both where it started and how heavily the part was loaded.
- The Stress ConcentratorExplain stress concentration, why corner radius governs it, and why a crack is the limiting case.Stress crowds around geometric interruptions. For a small circular hole in a wide plate under tension, classical elasticity (the Kirsch solution, 1898) gives a stress concentration factor of exactly 3. Sharpening the corner radius drives the factor up without limit, which is why a crack — a notch of near-zero radius — is such an efficient machine for extending itself, and why drilling a round hole at a crack tip is a real repair.
- The Comet: What the Story Gets WrongCorrect the popular account of the Comet failures and state what the investigation actually changed.The 1954 Comet losses were traced by full-scale water-tank pressurisation testing to fatigue cracking from a nearly-square ADF antenna cutout in the cabin roof — not the passenger windows of the popular story. The distinction matters: the real lesson was not 'round your windows' but that structures must be fatigue-tested against their true service load spectrum, which drove modern damage-tolerant design.
- The Liberty Ships and the ColdExplain the Liberty ship fractures via ductile-to-brittle transition, notches, and welded continuity.Hundreds of the ~2,700 welded Liberty ships cracked and several broke in two — some in calm water — because their steel's ductile-to-brittle transition temperature lay above North Atlantic winter conditions. Constance Tipper's work identified the mechanism. Three conditions had to coincide: cold, a notch, and an all-welded hull that gave a crack an uninterrupted path. The fix was not abandoning welding but specifying notch toughness at service temperature and adding crack arrestors.
- Corrosion: The Failure That Has TimeDistinguish uniform, galvanic, pitting, crevice, and stress corrosion cracking — and explain why the hidden ones are the dangerous ones.Corrosion needs only time. Uniform rust is visible and can be allowed for; the killers are localised. Galvanic attack builds a battery from dissimilar metals; pitting drills hidden holes; crevice corrosion works where you cannot look. Stress corrosion cracking needs a susceptible alloy, a specific environment, and tensile stress simultaneously — so individual tests pass while fine branching cracks grow beneath a clean-looking surface.
- Tacoma Narrows: Flutter, Not ResonanceExplain why the Tacoma Narrows collapse was aeroelastic flutter rather than resonance, and why the distinction changes the fix.On 7 November 1940 the bridge failed in a torsional mode under a roughly steady ~42 mph wind. Resonance requires a periodic driving force at a matching frequency, and a steady wind has no frequency — the energy came from the wind but the timing came from the deck's own motion. This is flutter: aerodynamic feedback pumping energy in faster than damping removes it, driving total damping negative. The fix is wind-tunnel testing of deck shape, not a frequency calculation.
- Hyatt Regency: The Load Path That DoubledTrace how a fabrication change doubled a connection load at the Hyatt Regency, and identify the organisational failure.On 17 July 1981 two walkways fell, killing 114 people. Splitting a continuous hanger rod into two shorter rods meant the lower walkway hung from the upper walkway's box beam rather than running independently to the roof — doubling that connection's load, on a detail already short of code. No material or weld failed. The change passed through review because it looked like a detailing question, and the profession's response was procedural.
- Why Things Actually BreakSynthesise the four cases into transferable habits: trace the load path, distrust tidy stories, assume the crack.In all four cases the famous story blamed someone's stupidity while the evidence showed a gap between model and world. The unifying technical idea is the load path — force flows where the structure actually routes it, not where the drawing says. The unifying organisational idea is normalisation of deviance. The durable response is damage tolerance: assume flaws exist and design to survive them until inspection finds them.
Questions this course answers
Why does a failure analyst insist on finding the crack's origin before asking who was at fault?
The origin tells you WHERE and therefore WHAT mechanism started it — the physical question. Fault is a legal and organisational question that comes later, if at all. Starting with blame tends to produce a satisfying story rather than a correct diagnosis.
Glass resists a very high stress before yielding, yet shatters when tapped. Which property is it missing?
Toughness is the AREA under the stress–strain curve — energy absorbed. Glass's curve is a steep line that stops dead, so there's almost no area beneath it. It's strong and stiff but has no way to soak up energy except by making cracks.
You find a fracture surface with nested chevron marks. What do they tell you?
Chevrons are a brittle-fracture signature, and they point back upstream toward the origin. Following them to where they converge takes you to the exact starting point — a weld flaw, a sharp corner, a tool mark.
A steel passes its tests at the mill as ductile, then fails in a brittle manner in service. What is the best explanation?
The same metal behaves ductile or brittle depending on conditions. Cold temperature, fast loading, and the presence of a notch all push toward brittle behaviour — which is exactly how steel that tested fine warm could shatter in the North Atlantic.
A shaft fails after eleven years at a stress well below the material's yield strength. Witnesses say it snapped suddenly. What most likely happened?
This is the classic fatigue signature. Cyclic loading well below yield grows a crack over years — the majority of the life is often just initiation. The remaining ligament finally tears in a millisecond, which is the only part anyone sees. The metal never 'crystallised'; the model was just wrong.
Two identical fatigue failures differ in one way: part A has a large rough final-fracture zone, part B a tiny one. What does that tell you?
The size of the final-overload zone reveals how highly loaded the part was. A high working load means the remaining metal fails while the crack is still small — a big rough zone. A lightly loaded part forces the crack to consume nearly the whole section first, leaving only a small final zone.
Grounded in trusted sources
- ASM International — ASM Handbook Vol. 11 (Failure Analysis and Prevention), Vol. 12 (Fractography), Vol. 13 (Corrosion), Vol. 19 (Fatigue and Fracture)
- K. Y. Billah & R. H. Scanlan (1991), 'Resonance, Tacoma Narrows bridge failure, and undergraduate physics textbooks', American Journal of Physics 59(2), 118–124
- NIST/NBS — Investigation of the Kansas City Hyatt Regency Walkways Collapse (NBS BSS 143): https://www.nist.gov/publications/investigation-kansas-city-hyatt-regency-walkways-collapse-nbs-bss-143
- Wikipedia — Hyatt Regency walkway collapse: https://en.wikipedia.org/wiki/Hyatt_Regency_walkway_collapse
- Wikipedia — de Havilland Comet: https://en.wikipedia.org/wiki/De_Havilland_Comet
- Wikipedia — Tacoma Narrows Bridge (1940): https://en.wikipedia.org/wiki/Tacoma_Narrows_Bridge_(1940)
- Constance Tipper — The Brittle Fracture Story (Cambridge University Press, 1962); Imperial College London profile: https://www.imperial.ac.uk/Stories/constance-tipper/
- DoITPoMS, University of Cambridge — Mechanical properties and the ductile-to-brittle transition: https://www.doitpoms.ac.uk/tlplib/mechanical_properties/ductile.php
Every Wunder lesson is built from real, reputable sources — never invented.
Related Science courses
Wunder is a personalized learn-anything platform — tell it any topic and it builds a beautiful, fact-checked course in minutes, with narration, a knowledge check, and a college-style University track.
Browse more Science courses · All topics · Home
© 2026 Wunder Learning LLC · Terms & Privacy