Structural Test as Substantiation Evidence

Technical provenance

Applicable standards / specifications

• EASA CS-25 — Certification Specifications and Acceptable Means of Compliance for Large Aeroplanes — Example airworthiness framework only; use the applicable certification basis for the product.

References

• EASA Easy Access Rules for Large Aeroplanes (CS-25) — Example aerospace certification framework. The actual certification basis must be established for each product and project.
• MMPDS — Metallic Materials Properties Development and Standardization — Common aerospace source for statistically based metallic material allowables where contractually applicable.

What Is Being Demonstrated?

A structural test demonstrates that a physical article — a coupon, a sub-element, a component or a full-scale structure — withstands a defined set of loads under controlled conditions. As substantiation evidence, a test provides something analysis cannot: a direct, empirical observation of how real material, real geometry and real manufacturing behave under load. This directness is the strength of test evidence. It is also the source of its principal limitation: the observation applies strictly to the article that was tested, under the loads and boundary conditions that were applied, in the environment that was present. Extending the conclusion beyond that specific configuration requires engineering justification, and that justification is itself a form of analysis.

A TEST SUBSTANTIATES THE CONFIGURATION THAT WAS ACTUALLY TESTED — EXTENSION BEYOND THAT CONFIGURATION REQUIRES ENGINEERING JUSTIFICATION.

Why Test Evidence Matters

Analysis predicts behaviour; test reveals it. No matter how carefully a model is built, it contains simplifications — idealised geometry, idealised boundary conditions, idealised material behaviour. A test confronts those simplifications with reality. When a test article survives the required load without failure, the evidence is immediate and compelling. When it fails, the evidence is equally valuable: the test has identified a failure mode, a load path or a manufacturing sensitivity that the analysis did not predict. In both cases, the test has produced information that analysis alone cannot provide. The question is not whether test is better than analysis — it is whether the test evidence is sufficient, representative and correctly interpreted for the requirement being substantiated.

• Test provides direct empirical evidence of structural capability — no model assumptions intervene between the load and the response.
• Test reveals failure modes that analysis may not predict — particularly manufacturing defects, secondary load paths and material variability.
• Test validates or challenges analysis predictions — correlation closes the loop between prediction and observation.
• Test is required by many certification specifications for primary structure and for novel configurations — the requirement for test is not optional.

Types of Structural Test

Structural tests span a range from material characterisation to full-scale structural proof. Each type provides evidence at a different level, and each has a different scope of applicability. Understanding what each type demonstrates — and what it does not — is essential to using test evidence correctly in the substantiation argument.

The Test Article: Representativeness

The test article is the physical specimen or structure that is tested. Its representativeness determines how far the test conclusion can be extended. A test article that matches the production configuration in material, geometry, manufacturing process, heat treatment and surface finish provides evidence directly applicable to that configuration. A test article that differs — even in seemingly minor ways — may produce evidence that does not extend to the production article. Representativeness must be assessed and documented, not assumed. The engineer must ask: does the test article represent the production article in every way that matters for the failure mode being assessed?

• Material: same specification, same heat treat condition, same thickness range?
• Geometry: same dimensions, same tolerances, same surface finish, same fillet radii?
• Manufacturing: same process, same tooling, same inspection?
• Assembly: same fasteners, same torque, same bond line, same cure cycle?
• Environment: same temperature, same moisture, same chemical exposure?
• Load introduction: same method, same location, same distribution?
• Boundary conditions: same constraints, same stiffness, same compliance?

A test article that is "close enough" is not representative unless the differences have been analysed and shown not to affect the failure mode being demonstrated.

Load Application and Boundary Conditions

The loads applied in a test must represent the loads the structure experiences in service — or, where they are deliberately different, the difference must be understood and justified. Load introduction is a particular concern: the way a load is applied in a test can produce a different stress distribution than the way the same load is applied in service. Whiffle trees, hydraulic jacks, pressure bags and reaction frames all introduce their own load paths. If the load introduction in the test does not represent the service load path, the test may demonstrate something different from what the structure actually experiences. Boundary conditions are equally critical: a test fixture that is too stiff or too compliant relative to the actual interface can produce stresses that are unrepresentative of the real structure.

A test with unrepresentative load introduction or boundary conditions demonstrates the capability of the test article under the test conditions — not necessarily under service conditions.

WHAT EVIDENCE IS AVAILABLE?

The evidence produced by a structural test includes more than the pass/fail outcome. A well-instrumented test produces a rich data set that can support the substantiation argument in multiple ways. The engineer should plan the test to capture the evidence needed, not merely the evidence that is easy to capture. Strain gauges, displacement transducers, load cells, acoustic emission, digital image correlation and post-test inspection all contribute to the evidence package. The test plan should specify what data is required, where it will be measured and how it will be used in the substantiation argument — before the test is run.

Test Factors and Load Levels

The load level applied in a structural test is governed by the applicable requirement. Many specifications require that a test be carried to a load above the limit load — often to the ultimate load, and sometimes beyond — to demonstrate capability with margin. The specific factor, if any, applied above the limit or ultimate load is programme-specific and must be taken from the governing requirement. The engineer must not assume a factor from a previous programme or from general practice. The test load must be defined by the requirement, agreed with the authority where applicable, and documented in the test plan before the test begins.

The test load level and any factor above limit or ultimate load must be taken from the governing requirement — not from memory, not from a previous programme, not from general practice.

Extending Test Conclusions Beyond the Tested Article

The central challenge in using test evidence is extension. A test demonstrates the capability of one article. The production fleet contains many articles, each with its own material variability, manufacturing tolerance and assembly variation. Extending the test conclusion to the fleet requires an understanding of variability and the factors that account for it. This is where test and analysis intersect: the test provides the observation, and the analysis provides the justification for extension. The extension argument may rest on statistical treatment of multiple coupons, on conservative allowables derived from test data, on analysis that shows the tested configuration is bounding, or on a combination of these. What it must not rest on is the unexamined assumption that "one test is enough."

• Statistical extension: multiple coupons tested and allowables derived with a defined statistical basis — the governing requirement specifies the basis.
• Conservative extension: the test article is shown by analysis to be a bounding case — thinner, weaker or more highly loaded than the production articles.
• Similarity extension: the tested configuration is shown to be structurally equivalent to the production configuration — documented and justified, not assumed.
• Analysis-supported extension: the test validates an analysis method, and the method is then used to assess the production configuration.

Extension from a single test article to a production fleet is an engineering argument, not a default. The argument must be made, documented and reviewed.

When a Test Fails

A test that fails below the required load is not a failure of the test — it is a failure of the design, the analysis or the manufacturing process. The test has done its job: it has revealed a deficiency that would otherwise have remained hidden. The response to a test failure is not to dismiss the test, to question the instrumentation or to rerun the test in the hope of a better result. The response is to investigate the root cause: Was the failure mode the one predicted by analysis? Was the load introduction correct? Was the test article representative? Was the material within specification? Was the manufacturing process within tolerance? Only when the root cause is understood can the correct action be taken — redesign, rework, reanalysis or, where justified, retest. A test failure that is properly investigated and resolved produces stronger evidence than a test that passed without question.

A test failure is evidence, not an embarrassment. The investigation of a failure often produces more valuable evidence than an uneventful pass.

Test Planning Checklist

The test plan is the document that defines what will be demonstrated, how it will be demonstrated and how the results will be used. A well-written test plan prevents the most common test evidence failures: unrepresentative articles, unrepresentative loads, missing instrumentation and results that cannot be used in the substantiation argument.

• Define the requirement being substantiated — The test must address a specific requirement, not just "see if it is strong enough."
• Define the test article and assess its representativeness — Document material, geometry, manufacturing, assembly and any differences from the production configuration.
• Define the load cases and load levels — Take from the governing requirement; document any factors above limit or ultimate load.
• Define the load introduction method — Assess whether the load path in the test represents the service load path.
• Define the boundary conditions — Assess whether the test fixture stiffness represents the interface stiffness in service.
• Define the instrumentation plan — Specify what data is required, where it will be measured and how it will be used.
• Define the acceptance criteria — State what constitutes a pass, a fail and an inconclusive result — before the test.
• Define the correlation plan — Specify which analysis results will be compared with test measurements.
• Define the post-test inspection — Specify what inspection will be performed after the test to reveal damage.

Common Errors in Using Test Evidence

Several errors recur in the use of test evidence for substantiation. Each undermines the value of the test or the validity of the conclusion drawn from it.

• Extending a test conclusion to a different configuration without documented justification — the most common and the most dangerous error.
• Treating a coupon test as structural substantiation — a coupon establishes material properties, not structural capability.
• Ignoring load introduction effects — the test load path differs from the service load path, and the difference is not analysed.
• Treating a proof test as design substantiation — a proof test confirms a specific article; it does not substantiate the design.
• Cherry-picking favourable test results — selecting the tests that pass and ignoring those that fail or are inconclusive.
• Failing to correlate test with analysis — the test data is collected but never compared with the model, wasting the opportunity to validate.
• Using a test factor from a previous programme without verifying it against the current governing requirement.

Key Takeaways

• A structural test provides direct empirical evidence — the strongest form of evidence when the test is representative.
• The test conclusion applies strictly to the article tested, under the loads and conditions applied — extension requires justification.
• Representativeness must be assessed and documented in material, geometry, manufacturing, loads, boundaries and environment.
• Test evidence types span coupon to full-scale — each provides evidence at a different level with a different scope.
• A test failure is evidence to be investigated, not an embarrassment to be dismissed.
• The test plan must define what is being demonstrated, how and how the results will be used — before the test is run.