The test pyramid: how composite structures get certified

Why you cannot just analyse it and fly

For a metal structure, decades of heritage and well-understood, forgiving failure (metals yield and redistribute load) mean analysis carries a lot of the certification case. Composites are different: the failure modes are many, brittle and coupled — fibre failure, matrix cracking, delamination, buckling of damaged sub-laminates — and they are sensitive to temperature, moisture and manufacturing. No model captures all of that reliably enough to bet lives on.

So aviation regulators — the FAA in the United States, EASA in Europe — require that a structure be shown to meet its requirements by a combination of analysis and physical test. For composites, the test half is large, and it is organised as a pyramid.

The building-block pyramid

The base is the widest and cheapest: thousands of small coupons — flat strips a few centimetres long. They establish the basic ply properties, the statistical scatter (for A- and B-basis allowables), and the environmental knockdowns (hot, wet, cold). Thousands of them, because statistics need numbers and each one is inexpensive.

Above the coupons come elements — specimens that add one real feature at a time: an open hole, a filled hole, a bolted joint, a bonded joint, a ply drop. Then subcomponents: a stiffened panel, a section of spar, a wing-to-fuselage fitting — a real piece of structure with real load introduction. And at the apex, one or a few full-scale test articles: an entire wing or fuselage, loaded in a rig.

The higher you go, the larger, costlier and fewer the tests — the opposite of the coupon base. A full-scale article can cost tens of millions and take years to build and instrument, so you only ever test one or two.

Why a pyramid, and not just the full-scale test

Two reasons: cost and risk. You do not want to discover a basic material or ply-level problem for the first time on a hundred-million-dollar full-scale article near the end of a programme. The cheap coupons at the base catch those problems early, when a fix is cheap.

The deeper reason is that each level validates the analysis, so you can lean on analysis more as you climb. This is the 'building-block' idea: coupons anchor the ply model; elements confirm the model handles holes and joints; subcomponents confirm it handles real load paths. By the time you reach full scale, the analysis is trusted enough that the single full-scale test is a confirmation, not a discovery. Certification is 'analysis supported by test', and the pyramid is how the support is built.

Static, fatigue and damage tolerance

The structure must survive limit load — the most severe load expected in service — with no permanent harm, and ultimate load, which for aircraft is 1.5 times limit, without failing. Full-scale articles are often loaded to limit, then to ultimate, and finally deliberately to failure to confirm where and how it breaks.

Then it must survive a lifetime of fatigue cycles, usually several times the design service life, to account for scatter. And it must meet the damage-tolerance case from the allowables story: with barely-visible impact damage present, it still carries load, and that damage must not grow. Each of these is demonstrated up the same pyramid.

Why composites need more of this than metals

A new metal alloy slots into a mature framework of handbooks, allowables and design rules built over a century. A new composite material or process — a new fibre, resin, or layup method — often means a new test campaign, because its behaviour is not transferable from the last one. Change the resin and the hot-wet knockdowns change; change the process and the void content and the allowables change.

That is the real reason composite certification is expensive and slow, and why the industry adopts new materials cautiously despite their performance. It is also why sustainable and recycled fibres, however promising, move into primary structure slowly: each one has to climb the pyramid.

Where this connects to mpolyco

This site sits at the very bottom of the pyramid and one level below it — the physics and the typical properties that a coupon campaign would then pin down for a specific material and process. The laminate and micromechanics calculators give you the analysis that the pyramid exists to validate; the design-allowables article explains the knockdowns that the coupon and element tests measure.

It is also why provenance runs through everything here. A property with a traceable source is a property that could, in principle, feed a certification case; a number with no source cannot. Teaching the pyramid is teaching why 'where did this value come from?' is the first question a real engineer asks.