4. Configuration item identification

Configuration freeze only works if every physical and logical element that can influence the test outcome has been identified, named, and versioned before the test is released. In a crash laboratory, the "configuration" is not a single document — it is a set of configuration items (CIs) spanning the test vehicle, the sled or acceleration system, the barrier or impactor, the ATDs (anthropomorphic test devices), and the entire instrumentation chain. This section explains how to decompose a crash test into these CIs, what attributes each CI must carry, and how to record them so that a later test can be proven identical — or deliberately different.

Why CI identification precedes the freeze

A freeze is a statement that "this is the configuration we will test." That statement is only meaningful if the configuration has been enumerated. If the vehicle build level is recorded but the dummy's calibration date is not, the freeze is incomplete and the test is not reproducible. The practical rule is: any item whose state, part number, revision, or adjustment can change the measured response must be a configuration item. CI identification is therefore a completeness exercise, not a documentation formality. It answers three questions for every element: what exactly is it, what version is it, and what evidence proves that version.

Decomposing the test into configuration items

A frontal offset test, for example, decomposes into five CI families. Each family has a distinct owner, a distinct evidence trail, and a distinct failure mode if misidentified.

Each row is a family, but the freeze operates at the level of the individual item. A vehicle with the correct build level but an unrecorded battery state is not frozen; a dummy with a valid calibration but an undocumented jacket revision is not frozen either.

Vehicle configuration: build level is not enough

The vehicle CI must capture the state that the test is intended to represent. In practice this means recording the build level (prototype phase, part revision), the as-tested mass and its distribution, and any deliberate deviations from the nominal build — for example, a removed airbag, a substituted seat, or a ballasted fuel tank. The as-tested mass is particularly important because it directly scales the kinetic energy at impact; a 40 kg difference on a 1,500 kg vehicle changes the energy by roughly 5% at the same velocity. Record corner weights and the longitudinal center of gravity, and note the measurement method (scales vs. calculated). Tire pressure, fluid levels, and battery state should be recorded as tested, not as nominal, because they are frequently adjusted between tests and are easy to overlook.

Sled and barrier: the controlled inputs

On a sled test, the sled and its propulsion system are the source of the acceleration pulse, so the pulse file itself is a configuration item. Record the file name, version, and a hash, together with the target Δv and peak acceleration. The sled mass with payload must be recorded because the same pulse file produces a different velocity change when the mass changes. For barrier tests, the barrier is a CI in its own right: the honeycomb batch, face dimensions, mounting height, and the load cell map all affect the measured forces. A barrier that has been used several times may have deformed honeycomb; record the usage count and the certification date.

ATD configuration: serial number and channel map

Dummies are the most frequently misidentified CI because they look interchangeable. They are not. Record the ATD serial number, model, calibration due date, and the exact instrumentation package installed. The channel map — which sensor is in which location, with which range and filter — must be attached to the ATD CI, not kept separately, because a swapped channel silently corrupts the result. Positioning is part of the configuration: record the H-point coordinates, seat back angle, and steering wheel position as measured, not as specified.

Instrumentation: the measurement chain as a CI

The instrumentation CI covers the full chain from sensor to recorded channel: sensor serial number, mounting location and orientation, range, calibration factor, DAQ channel, sample rate, and filter class. A sensor is only a CI in the context of its mounting; the same accelerometer mounted 20 mm away measures a different signal. Record the mounting location in vehicle coordinates and state the coordinate convention explicitly — for example, origin at the vehicle centerline at the front axle, x forward, y left, z up — and keep that convention consistent across the vehicle, ATD, and sensor records.

Worked example: a frontal offset test CI register

The following excerpt shows how the five families collapse into a single register with a stable ID scheme. The values are illustrative and represent a plausible test setup, not measured results.

CI-ID     Family          Item                        Version / Serial        Evidence
VEH-001   Vehicle         BIW build level             P3 / rev C             build sheet BS-0412
VEH-002   Vehicle         As-tested mass              1,512 kg (LF 412, RF 398, LR 351, RR 351)
VEH-003   Vehicle         Tire pressure (cold)        2.4 bar all corners    gauge log
SLD-001   Sled            Pulse file                  FO-56kmh-v3.sha256      hash recorded
SLD-002   Sled            Sled mass with payload      1,840 kg               scale ticket
BAR-001   Barrier         ODB face, honeycomb batch   batch H-2291, use 3/10  cert BC-2291
ATD-001   ATD             Driver Hybrid III 50th      SN 0417, cal due 2025-09
ATD-002   ATD             Channel map                 CH-0417-rev2           assignment sheet
INS-001   Instrumentation DAQ channel map             DAQ-2, 20 kHz, CFC 1000
INS-002   Instrumentation Camera sync                 cam 1-4 @ 1000 fps     sync pulse log

Note that the register records evidence, not just values. A version number without a traceable source is a claim, not a configuration. The hash on the pulse file and the batch number on the honeycomb are what make the freeze auditable.

Common identification failures

• Family-level freeze: recording "Hybrid III 50th" without the serial number, so a dummy swap goes undetected.
• Nominal instead of as-tested: recording specified tire pressure or fluid levels rather than measured values.
• Orphaned channel map: keeping the sensor list in the DAQ software but not linking it to the ATD and vehicle CIs.
• Unversioned pulse file: editing the pulse file after freeze without a new version and hash.
• Coordinate ambiguity: sensor locations recorded without stating the origin and axis convention.

The register above is the input to the freeze step: once every CI has an ID, a version, and evidence, the freeze can be declared against a specific register revision. The next step — locking that register and controlling changes — depends entirely on the completeness achieved here.