Test Article Configuration & Representativeness

The test article is defined by how it was made

The evidence a test produces is only as representative as the test article and configuration used to generate it. Two components that look identical can behave differently because one was machined from billet and the other cast, because one was heat-treated after welding and the other was not, because one has fasteners torqued to the production value and the other hand-tightened, or because one has a surface finish that introduces residual stress and the other does not. The test article is defined by its geometry, its material, its manufacturing process, its joints and fasteners, its preload, its surface condition, its damage state and any modifications made for test. A test report that documents only the geometry does not document the article.

TEST EVIDENCE IS ONLY AS REPRESENTATIVE AS THE TEST ARTICLE AND CONFIGURATION USED TO GENERATE IT.

Configuration factors and what they affect

The table below lists the configuration factors that most often differ between a test article and the service hardware it represents, and what each factor affects. The list is not exhaustive — the factors that matter depend on the component, the material system and the failure mode being investigated — but it captures the factors that most frequently undermine the representativeness of a test without the test team noticing. For each factor, the question is not "is it identical?" but "does the difference change the response the test is measuring?"

The test article hierarchy

Structural tests are conducted at several levels of assembly, and the level chosen depends on the engineering question. A coupon test characterises a material or a lay-up under a simple stress state; it answers questions about the material, not about the component. A subcomponent test investigates a feature — a hole, a joint, a ply drop — under a controlled load; it answers questions about the detail. A component test exercises a full part — a panel, a fitting, a spar — under loads representative of its in-service environment; it answers questions about the part. A subassembly test exercises the interface between components; it answers questions about load transfer and interaction. A full-scale test exercises the complete structure under the integrated load case; it answers questions about the system. Each level has a different cost, a different lead time, and a different relationship to the service hardware. The art is in choosing the lowest level that adequately answers the question, because higher levels cost more but answer questions the lower levels cannot.

TEST ARTICLE HIERARCHY

  COUPON
  (material / lay-up characterisation;
   simple stress state)
         │
         ▼
  SUBCOMPONENT
  (feature: hole, joint, ply drop;
   controlled load)
         │
         ▼
  COMPONENT
  (full part: panel, fitting, spar;
   representative loads)
         │
         ▼
  SUBASSEMBLY
  (interface between components;
   load transfer, interaction)
         │
         ▼
  FULL-SCALE
  (complete structure;
   integrated load case)

  Each level answers a different question.
  The art is choosing the lowest level
  that adequately answers the question.

Prototype versus production

A common representativeness problem arises when the test article is a prototype manufactured by a different route from the eventual production hardware. A prototype composite panel may be hand-laid where production is automated; a prototype metallic fitting may be machined from billet where production is forged; a prototype assembly may be drilled by hand where production is drilled by a multi-axis machine. Each of these differences can change the residual stress, the defect population, the dimensional tolerance and the joint behaviour in ways that affect the measured response. This does not mean the prototype test is useless — it may answer the question at the level of "does the design concept work?" — but it does mean that the evidence cannot be assumed to represent the production hardware without an explicit assessment of what the manufacturing differences change.

TESTING A PROTOTYPE MANUFACTURED DIFFERENTLY FROM THE PRODUCTION COMPONENT WITHOUT DOCUMENTING THE DIFFERENCES CAN PRODUCE RESULTS THAT DO NOT REPRESENT THE EVENTUAL HARDWARE. The test article is defined by how it was made, not just what it looks like.

Joints, fasteners and preload

Joints are where representativeness most often fails silently. A bolted joint's behaviour depends on the fastener type, the hole preparation, the clamp-up force, the surface condition at the faying surfaces and the sequence in which the fasteners were torqued. A bonded joint's behaviour depends on the surface preparation, the adhesive cure, the bondline thickness and the environmental condition. A test that uses the right geometry but the wrong fasteners, or the right fasteners with the wrong torque, or the right torque in the wrong sequence, can measure a joint stiffness or a load distribution that does not represent the service joint. For tests whose objective is joint behaviour, the joint configuration must be documented to the same level of detail as the geometry.

Damage and environmental conditioning

For tests that investigate residual strength, damage tolerance or environmental behaviour, the article condition is part of the test definition, not a background detail. Impact damage must be introduced at a defined energy, location and boundary condition; fatigue damage must be introduced under a defined spectrum; environmental conditioning must be applied at a defined temperature and humidity for a defined duration. The condition of the article at the start of the test is a measured input, and it must be documented with the same rigour as the load and the instrumentation. A residual strength test on an article whose damage state is not characterised produces a number whose meaning is unclear.

Test modifications and their effects

Test articles are sometimes modified for test: holes drilled for wiring, cut-outs for access, additional mass for sensor mounting, loading tabs bonded or bolted to the surface. Each modification changes the local stiffness, the mass distribution or the load path, and each change has the potential to alter the response the test is measuring. The discipline is to minimise modifications, to assess the effect of those that are unavoidable, and to document them so that the result can be interpreted with the modification in mind. A sensor whose mounting changes the local stiffness of a thin panel can alter the very mode shape it is trying to measure.

Configuration traceability

A defensible test report traces the test article to its origin: the drawing revision, the material lot, the manufacturing route, the joint configuration, the preload, the surface condition, the damage state and any modifications. This traceability is what allows the result to be applied to the service hardware — or to be recognised as not applicable. A test on an article whose configuration is not documented is a test whose evidence cannot be transferred. The traceability requirement is not bureaucracy; it is the condition under which the measurement can support an engineering decision about hardware that was not in the test rig.

What to check before the test

Before the test begins, the configuration of the article should be checked against the test definition. Does the as-built geometry match the drawing within tolerance? Is the material traceable to a certified lot? Were the joints assembled with the correct fasteners, torque and sequence? Is the surface condition representative of service? Is the damage state — if any — characterised? Are test modifications documented and their effects assessed? Each of these checks is a link in the chain from article to evidence, and each one that is missed introduces a doubt about whether the measured response represents what the test was intended to represent.