How engineers predict where a structure will fail

When a structure fails, the break is almost never at the point carrying the most load. It is at a bolt hole, a weld toe, a sharp internal corner, a scratch — somewhere the geometry quietly multiplied the stress while nobody was watching. Predicting failure is therefore not a matter of adding up forces. It is a matter of finding where those forces get concentrated, what the material will do when they do, and how many times the structure can survive that before something starts to crack.

Stress, not force, is what a material feels

A material does not respond to the total load applied to it; it responds to the load spread over the area carrying it. That is stress — force per unit area — and its companion strain, the resulting proportional deformation. Doubling a bar's cross-section halves the stress under the same load, which is why a design is judged in stress rather than in tonnes.

For most engineering materials, stress and strain rise together in a straight line up to a limit: the elastic region, where removing the load returns the part to its original shape. Beyond the yield point the deformation becomes permanent, and further along lies the ultimate strength at which the material tears apart. Most design keeps working stresses inside the elastic region with a margin to spare.

Geometry decides where the stress collects

Stress does not distribute itself evenly. It flows through a component rather like water through a channel, and anywhere that flow is forced to turn sharply it crowds together. A hole, a notch, a step change in thickness or a sharp re-entrant corner can locally multiply stress by two, three or more — a stress concentration.

This is why engineers round off internal corners, why a crack in a windscreen is drilled at its tip to blunt it, and why surface finish matters so much on parts that carry cyclic loads. The same component with the same load can be safe or unsafe depending entirely on the radius at one corner.

Materials fail in two very different ways

• Ductile failure gives warning. The material yields visibly, deforms, necks down and then tears — mild steel and aluminium behave this way at ordinary temperatures. A structure that yields before it breaks can redistribute load and is far more forgiving.
• Brittle failure gives none. Cast iron, glass, concrete in tension and even normally ductile steel at low temperature can crack suddenly at stresses well below what a static calculation suggested, because a small flaw propagates catastrophically.

Buckling belongs in a category of its own. A slender column can fail long before its material is anywhere near yield, because the straight shape itself becomes unstable and the column snaps sideways. That is a stability problem governed by geometry and stiffness rather than by material strength — which is why a long thin strut fails at a far lower load than a short one of identical material.

Fatigue: failure by repetition

Many real failures happen at loads a structure carried safely thousands of times before. Under repeated cycling, microscopic damage accumulates at a stress concentration, a crack initiates, and it grows a little with each cycle until the remaining section can no longer hold. This is fatigue, and it explains a large share of failures in aircraft, bridges, rails and rotating machinery.

Fracture mechanics turned this into a quantitative discipline. Rather than assuming a component is flawless, it assumes a crack of some size exists, then calculates whether that crack will grow under the expected loading and how quickly. That reframing is what allows aircraft to be inspected on schedule rather than retired on suspicion — the damage-tolerant philosophy.

Structures do not fail where you loaded them. They fail where the geometry, the flaw and the repetition happened to agree — and the engineer's job is to find that agreement before the structure does.

How prediction is actually done

For simple shapes, closed-form equations and handbook stress-concentration factors are enough. For anything real, engineers use finite element analysis (FEA) — dividing the structure into thousands of small elements and solving the resulting equations numerically to map stress everywhere. FEA is powerful and easy to misuse: a poor mesh, wrong boundary conditions or an unrepresentative material model produce confident, colourful, wrong answers, which is why validation against physical testing remains essential.

On top of the analysis sits a factor of safety, absorbing what the model cannot know: material variability, unforeseen loads, manufacturing defects, corrosion and degradation over decades.

Why it matters for students and researchers

Applied mechanics underpins civil, mechanical, aerospace and biomedical engineering alike, and it is where design stops being a drawing and becomes a prediction. Active research includes fatigue behaviour of additively manufactured parts, composite and multi-material failure, structural health monitoring with embedded sensors, digital twins that update a model from live data, and machine-learning surrogates for expensive simulations. Following the peer-reviewed literature is how engineering students and professionals keep pace with a field where a single missed stress concentration remains a public-safety matter.

Frequently asked questions

What is a stress concentration?

A stress concentration is a local increase in stress caused by geometry — a hole, notch, sharp corner or abrupt change in section. The stress there can be several times the average stress in the component, which is why failures usually start at such features rather than at the point of maximum applied load.

What is the difference between ductile and brittle failure?

Ductile failure is preceded by visible permanent deformation, so the structure gives warning and can redistribute load before breaking. Brittle failure occurs suddenly with little or no deformation, often from a small pre-existing flaw, and can happen at stresses well below the material's nominal strength.

What is fatigue failure?

Fatigue failure is cracking caused by repeated loading rather than a single overload. Damage accumulates at a stress concentration over many cycles, a crack forms and grows incrementally, and the part eventually fails at a load it had carried safely many times before.

How is finite element analysis used to predict failure?

FEA divides a structure into many small elements and solves the governing equations numerically to produce a map of stress and deformation across the whole geometry. Engineers use it to locate stress concentrations and check them against material limits, but results must be validated by testing, since poor meshing or wrong boundary conditions give misleading answers.