4. Test article selection and build documentation

In environmental qualification testing of restraint systems, the test article is not a single part — it is a controlled assembly of representative hardware, sensors, and inflator variants whose configuration must be traceable from the moment it is selected to the moment the qualification report is signed. A qualification unit that cannot be reconstructed, or whose build differs from the production intent in an undocumented way, invalidates the test evidence regardless of how well the chamber performed. This section covers how to select representative hardware, how to manage sensor and inflator variants, and how to maintain configuration control across the qualification build.

What "Representative" Actually Means

Representative hardware is hardware that matches the production design in every characteristic that influences the environmental response being evaluated. For a seatbelt retractor, that includes the housing material and wall thickness, the locking mechanism geometry, the pyrotechnic charge mass, the gas generant formulation, the connector interface, and the mounting bracket stiffness. It does not necessarily include cosmetic finishes, label artwork, or non-structural trim, provided those items are documented as non-influential in the qualification test plan.

The practical rule is: if a characteristic can change the thermal mass, the mechanical resonance, the seal integrity, or the pyrotechnic output, it must be built to production intent. If it cannot, it may be substituted, but the substitution must be recorded with a rationale. This distinction is what separates a defensible qualification unit from a convenient one.

Selecting the Qualification Build

Selection begins with the production part numbers and the engineering release level that the qualification is intended to cover. From that baseline, the test engineer identifies the worst-case configuration for each environmental stress. Worst-case is not always the largest or heaviest part; it is the configuration that produces the most severe response at the critical location.

For a restraint system subjected to thermal cycling, the worst case is often the configuration with the highest thermal mass and the slowest time constant, because it lags the chamber air temperature and experiences the largest internal gradients. For vibration, the worst case is typically the configuration with the lowest structural damping or the resonance closest to the excitation band. For pyrotechnic shock, the worst case is the configuration with the stiffest mounting and the shortest gas path.

These worst cases may not coincide. When they do not, the qualification program either builds multiple dedicated units — one per worst-case configuration — or selects a single unit that envelopes all of them and justifies the envelope analytically. The envelope approach is cheaper but requires a documented argument that the chosen configuration is at least as severe as each individual worst case. In practice, most programs build two or three units: a thermal/vibration unit, a pyrotechnic unit, and a control unit that remains unexposed for reference measurements.

Sensor and Inflator Variants

Sensors and inflators are the components most likely to vary within a qualification build, because they are often sourced from multiple suppliers or produced in multiple revisions during the program. Each variant must be treated as a distinct configuration, not as an interchangeable part.

For sensors, the relevant variant parameters are the measurement range, the sensitivity, the frequency response, the mounting thread and torque, and the connector keying. A sensor with a different range may saturate during the test; a sensor with a different frequency response may distort the measured shock spectrum. Both conditions produce data that cannot be compared to the qualification limits.

For inflators, the relevant variant parameters are the gas generant formulation, the charge mass, the initiator type, the burst disc rating, and the outlet geometry. These parameters directly control the pressure-time curve and therefore the structural response of the restraint. An inflator variant that produces a 10% higher peak pressure can shift the measured strain beyond the qualification margin even though the hardware is otherwise identical.

The table below summarizes the variant parameters that must be captured for each unit.

Configuration Control for Qualification Units

Configuration control means that every qualification unit has a documented build record that ties each installed component to a specific part number, serial number, and revision level. The record must be created before the unit enters the chamber and must be updated if any component is replaced during the test campaign.

A practical build record contains the following fields: unit serial number, build date, assembly drawing revision, bill of materials revision, and for each installed component the part number, supplier, lot or serial number, and revision. The record also captures the as-built measurements that matter for the test — for example, the measured mounting torque, the sensor calibration date, and the inflator lot. These measurements are the evidence that the unit is representative.

When a component is replaced mid-campaign, the replacement is treated as a configuration change. The build record is updated, the reason for replacement is documented, and the test engineer assesses whether the change invalidates any previously completed test. A sensor replaced because of a failed calibration, for example, does not invalidate the thermal test already completed, but it does require that the vibration test be re-baselined against the new sensor response.

Worked Example: Build Record for a Retractor Qualification Unit

Consider a qualification unit designated QR-04, intended for thermal cycling and random vibration. The production baseline is retractor P/N 4100-01, revision C. The test engineer selects the following configuration:

• Housing: P/N 4100-01-01, revision C, lot H-2291, wall thickness 2.4 mm (production nominal 2.4 mm ±0.1 mm)
• Accelerometer: P/N 352C33, serial 88214, range ±500 g, sensitivity 10.2 mV/g, calibrated 2024-03-11
• Inflator: P/N 4100-01-07, revision B, lot I-4470, charge mass 1.85 g, generant formulation G-12
• Mounting bracket: P/N 4100-01-12, revision A, fastener torque 9.0 N·m ±0.5 N·m

The as-built measurements are recorded on the build record: mounting torque measured at 9.2 N·m, sensor mounting torque at 2.8 N·m, and inflator installation torque at 12.0 N·m. The unit is photographed before and after assembly, and the photographs are attached to the record.

During the vibration test, the accelerometer at the retractor base shows a resonance at 187 Hz with an amplification factor of 4.2. This is within the expected band for the production configuration. If the measured resonance had been at 240 Hz, the test engineer would need to investigate whether the bracket stiffness or the mounting torque differed from production intent — a classic symptom of a configuration that is not representative.

Common Failure Modes in Test Article Selection

Three failure modes account for most configuration problems in restraint qualification. First, using a pre-production inflator because the production lot is not yet available; this is sometimes unavoidable, but it must be documented and the qualification must be repeated with production hardware before design release. Second, mixing sensor variants within a single test campaign without re-baselining; this produces data that appears valid but is not comparable across units. Third, omitting the as-built measurements from the build record; without them, there is no evidence that the unit matched the intended configuration.

The corrective action for all three is the same: treat the build record as a controlled document, not as a formality. It is the single artifact that allows a reviewer to reconstruct the qualification unit years later and confirm that the test evidence applies to the production design.

The diagram above shows the relationship between the production baseline, the worst-case selection, the component variants, and the controlled build record. The dashed feedback path is the key control: any component replacement or configuration change during the test campaign must flow back into the build record before the next test is run. Without that loop, the qualification unit loses traceability and the test evidence becomes unusable for design release.