2. Crash test facility setup

A full-scale vehicle crash test is only as credible as the facility that produces it. Before any correlation with CAE results can be trusted, the physical test environment must be controlled to a level where the measured response is dominated by the vehicle structure, not by the apparatus. This section examines the three pillars of facility setup: sled test systems for component and subsystem work, the barrier types used for full-scale impacts, and the data acquisition chain that converts mechanical events into defensible signals.

Sled Test Systems

A sled system accelerates or decelerates a test buck along a guided track to reproduce a target pulse without destroying an entire vehicle. The two dominant architectures are acceleration sleds and deceleration sleds. An acceleration sled uses a propulsion system — typically a pneumatic or hydraulic ram, or a bungee/catapult arrangement — to drive the sled from rest up to a target velocity, then releases it. A deceleration sled is brought to speed first and then stopped by a programmable energy-absorbing device, so the specimen experiences the crash pulse as a deceleration.

The pulse-shaping element is the heart of the system. In a deceleration sled, the sled is stopped by a set of metal or composite crush tubes, hydraulic dampers, or a combination of both. By selecting tube wall thickness, material grade, and the number of active tubes, the engineer tunes the shape of the deceleration pulse. A typical frontal pulse might be specified as a target corridor: for example, a peak of approximately 35–40 g reached within 30–50 ms, with a total pulse duration of about 100–120 ms. The measured sled acceleration must fall inside the corridor at every time step, not merely match the peak.

Two parameters deserve particular attention during setup. First, sled velocity is usually measured by a light gate or a magnetic pickup and must be verified before the run; a 1% velocity error can shift the entire pulse. Second, pulse repeatability is checked by running the same tube configuration several times and comparing the acceleration traces. If the peak varies by more than a few percent, the crush tubes or the damper settings are suspect.

Barrier Types for Full-Scale Impacts

Full-scale barrier tests fall into two broad families: fixed rigid barriers and deformable or moving barriers. The choice of barrier defines the loading condition and therefore the CAE load case that must be correlated.

For a rigid barrier, the critical setup checks are foundation stiffness and surface flatness. A barrier that deflects or rotates under load will absorb energy that should have gone into the vehicle, producing a pulse that is too soft and a CAE correlation that appears worse than it is. For deformable barriers, the honeycomb or aluminum block must be inspected for moisture, prior damage, and correct density; a barrier that has been left in a humid environment can crush at a noticeably different force.

Alignment is equally important. The vehicle is positioned relative to the barrier using a laser or plumb-line reference, and the overlap, yaw angle, and vertical offset are recorded. A yaw error of even one or two degrees can change the deformation mode in an offset test, so the setup sheet should capture these values before every run.

Data Acquisition for Full-Scale Impacts

The data acquisition chain begins at the transducer and ends at a filtered, time-aligned signal. A typical full-scale frontal test may carry 40–80 channels, including accelerometers on the vehicle body and occupant compartment, load cells in the barrier or on the sled, displacement sensors, and high-speed video synchronization signals.

Accelerometers are the primary structural channels. They are mounted on rigid locations — the rocker, tunnel, B-pillar base, and rear seat crossmember — using studs or adhesive pads, with the sensitive axis aligned to the vehicle coordinate system. The standard convention is +X forward, +Y to the vehicle's left, +Z upward, and this must be stated explicitly in the test report so that CAE and test signals can be compared without sign errors.

Signal conditioning follows a defined chain: anti-alias filtering before digitization, then a channel filter class specified by the test protocol. For occupant safety signals, the commonly used filter is a CFC 60 (Channel Frequency Class) low-pass filter, while structural accelerometers are often filtered at CFC 180 or CFC 600 depending on the frequency content of interest. The sampling rate is typically 10 kHz or higher to preserve the high-frequency content that survives filtering.

Example channel setup (illustrative):
Channel 12: B-pillar base accelerometer, X axis
  Sensitivity: 1.0 mV/g
  Filter class: CFC 60
  Sample rate: 20 kHz
  Zero offset: corrected pre-trigger
Channel 27: Barrier load cell, left block
  Range: 0–200 kN
  Filter class: CFC 600
  Sample rate: 20 kHz

Time zero is defined by the trigger — usually a contact switch on the barrier or a light gate at the barrier face. Every channel must be time-aligned to this reference. A common failure mode is a small time offset between the accelerometer and load cell channels, which produces an apparent phase error in the force–displacement curve and makes a correct CAE model look wrong.

Integrating the Facility Setup

The three elements are not independent. The sled pulse defines the loading history used to calibrate the model; the barrier defines the boundary condition for the full-scale validation; and the data acquisition chain determines whether the comparison is meaningful. A well-run facility documents all three in a single setup sheet: sled tube configuration and target pulse, barrier type and alignment values, and the channel list with filter classes and sampling rates.

The diagram above shows how the three facility elements feed into a single correlation workflow. The sled system provides a controlled pulse for model calibration; the barrier defines the full-scale boundary condition; and the data acquisition chain determines the fidelity of the measured response. When a CAE model fails to match a test, the first question should always be whether the facility setup was correctly documented and repeatable — because a correlation error is just as likely to originate in the test as in the model.