Every composite pressure vessel standard sets a minimum burst pressure. Meeting it with one prototype is easy. Meeting it with every vessel the factory will ever build, without wasting expensive carbon fiber, is the real design problem. Statistics is how good designers solve it.
Why a single burst test is not enough
Composite strength scatters. Two vessels wound on the same machine from the same materials will not burst at exactly the same pressure, because fiber tension, resin content, voids and small wrinkles all vary. A design that passes qualification with a handful of prototypes can still produce the occasional weak vessel in production. And because regulators approve designs on a limited number of tests, the burden is on the manufacturer to show that the whole population, not just the test samples, will be safe for the vessel's whole service life.
Fiber strength comes from your own burst tests
Handbook values and fiber supplier strand data are a starting point, not a design value. What a fiber delivers inside a wound pressure vessel depends on the winding process, the resin, the cure and the laminate. The reliable approach is to burst-test a series of vessels, often subscale test bottles, back-calculate the fiber stress at burst from each test using stress analysis or strain gauge data, and then work out the average strength and its scatter. That data set belongs to one factory and one process. Change the fiber, the resin or the machine, and it needs to be checked again.
Setting the target: enough strength, but not too much
With the average and standard deviation in hand, the designer chooses a reliability level and sets the target average burst pressure so that the low tail of the distribution still clears the minimum. For a normal distribution, a design whose average sits three standard deviations above the minimum puts about 99.87% of production above it; four standard deviations gives about 99.997%.
Here is a simple illustration. Suppose the standard requires a minimum burst of 600 bar and your burst tests show a coefficient of variation (standard deviation divided by average) of 4%. The target average works out as the minimum divided by one minus the number of standard deviations times 4%:
| Design rule | Share of production above the minimum | Target average burst (4% scatter) |
|---|---|---|
| Average minus 3 standard deviations | 99.87% | about 682 bar |
| Average minus 4 standard deviations | 99.997% | about 714 bar |
The difference between the two rows is about 5% more burst strength, which for a carbon fiber vessel means roughly 5% more of its most expensive material. A process with less scatter needs less margin, so every improvement in consistency on the factory floor comes straight back as lighter, cheaper vessels.
Few tests means more caution
The average and standard deviation from a small set of burst tests are themselves estimates. With five or ten results, the true scatter may be larger than it looks. Statisticians handle this with tolerance limits, which widen the margin to reflect the sample size and the confidence wanted. The fewer the tests, the larger the factor applied to the standard deviation. The NIST/SEMATECH e-Handbook of Statistical Methods sets out the method.
Keep measuring in production
Statistics is not only for qualification. Cylinder rules typically require burst tests from production: under DOT-CFFC, FRP-1 and ISO 11119-2, for example, a cylinder from each lot of up to 200 is burst-tested. Tracking those results, their average and their scatter, over time shows whether the process is drifting long before a lot fails. A rising scatter is an early warning that something on the line has changed.
Statistics covers strength. FMEA covers everything else.
Burst statistics answer one question: is the vessel strong enough when new? Long-term safety also depends on impact, cuts and abrasion, fire, UV, low temperatures, chemical attack, permeation, contaminated gas, stress rupture and accidents. A failure modes and effects analysis (FMEA) works through each of these and shows how the design, the protective materials, the testing or the inspection program deals with it. Regulators, including the US DOT for special permit applications, often expect to see one. Standards set the minimum. A good FMEA finds the gaps the standards do not cover.
What this means for buyers and investors
When you assess a manufacturer, ask how they set their design burst targets, how many burst tests their fiber strength values are based on, and whether they track production burst results over time. The answers say a lot about how consistent the product will be, and about how much carbon fiber, and cost, is in each vessel.
This article explains a general engineering method with illustrative numbers. Design values must come from your own materials, process and test data, and meet the standard that applies.