3. Instrumentation for physical crash testing

Physical crash testing produces its value only through measurement. A barrier impact that deforms a vehicle exactly as predicted is worthless as correlation evidence unless the deformation, deceleration, and occupant loading were captured with known accuracy and traceable calibration. This chapter examines the four instrumentation families that dominate automotive crash testing — accelerometers, load cells, strain gauges, and high-speed cameras — and explains how each is selected, mounted, configured, and validated so that the resulting data can be trusted during CAE correlation.

Accelerometers: The Primary Crash Signal

Accelerometers convert rigid-body acceleration into an electrical signal, most commonly via piezoelectric or piezoresistive sensing elements. In a full frontal rigid-barrier test, the vehicle body is instrumented at the B-pillar base, the rear seat cross-member, the engine cradle, and the sill — locations chosen because they approximate rigid-body motion of the occupant compartment. A typical sensor is a triaxial piezoelectric accelerometer with a ±500 g range for body-mounted channels and ±2000 g for component-level measurements such as the steering column or pedal box.

Mounting is the dominant error source. A sensor bonded to a flexible bracket will measure bracket resonance, not vehicle deceleration. The standard practice is to mount accelerometers on machined aluminum blocks that are bolted or adhesively bonded directly to structural sheet metal, with the sensitive axis aligned to the vehicle coordinate system. The SAE J211 convention defines the sign convention: positive X is forward (vehicle travel direction), positive Y is to the driver's right, positive Z is downward. Every channel must be documented with its axis orientation before the test, because a reversed polarity will invert the entire crash pulse and invalidate the correlation.

Signal conditioning follows SAE J211/1. The raw signal is passed through a low-pass filter, most commonly a fourth-order Butterworth at CFC 60 (cutoff 100 Hz) for occupant response channels and CFC 180 (cutoff 300 Hz) for structural channels. The filter class must be recorded, because comparing a CFC 60 test signal against an unfiltered CAE output will produce an apparent phase and magnitude mismatch that is purely a post-processing artifact.

Load Cells and Strain Gauges

Load cells measure force directly and are used where the load path is well defined: barrier load cells in a fixed rigid wall, seat belt anchor load cells, and steering column force transducers. A typical frontal barrier is instrumented with an array of load cells covering the full face, allowing the total barrier force and its distribution to be reconstructed. Each cell is calibrated against a known deadweight or proving ring, and the calibration factor (mV/V per kN) is entered into the data acquisition system before the test.

Strain gauges are bonded resistive elements whose resistance changes with surface strain. They are applied to structural members where force cannot be measured directly — B-pillar sections, door intrusion beams, suspension links. A quarter-bridge configuration is common for simple uniaxial strain, while a full Wheatstone bridge with temperature compensation is preferred on crash structures because it cancels bending and axial cross-talk. The gauge factor, bridge excitation voltage, and shunt calibration resistor must all be logged. A practical check before the test is a shunt calibration: a known resistor is switched across one bridge arm, producing a predictable output step that verifies the entire signal chain from gauge to recorder.

High-Speed Cameras

High-speed cameras provide the spatial and kinematic context that point sensors cannot. A typical frontal crash setup uses cameras at 1000–2000 frames per second for exterior views (barrier, front structure, occupant compartment) and 500–1000 fps for onboard views. Resolution and frame rate trade off directly: at 1280×800 and 1000 fps, a camera may record only 1–2 seconds, which is sufficient because the event of interest lasts 100–150 ms.

Camera placement must avoid occlusion by deploying airbags, flying glass, and the barrier itself. Lighting is equally critical — crash halls use high-intensity flicker-free LED arrays synchronized to the camera frame rate to prevent banding. For correlation work, cameras are synchronized to the data acquisition system via a common trigger, and a visible time code or LED flash is recorded in-frame so that video frames can be aligned to the accelerometer time base. Without this synchronization, comparing CAE animation to test video is qualitative at best.

Data Acquisition and Channel Budget

All sensor families feed a single data acquisition system, typically a 16-bit or 24-bit recorder sampling at 10–20 kHz per channel. The channel budget for a full frontal test commonly reaches 60–120 channels. Table 1 summarizes the typical allocation.

The following SVG illustrates the complete instrumentation signal chain from sensor to correlation-ready data, including the calibration and synchronization points that must be verified before every test.

Practical Verification Before the Test

A disciplined pre-test procedure prevents the most common correlation failures. First, verify accelerometer polarity by gently tapping the sensor in the positive axis direction and confirming a positive output excursion. Second, run a shunt calibration on every strain bridge and confirm the output step matches the calculated value within 2%. Third, confirm that all cameras are triggered by the same event and that the in-frame time code matches the DAQ trigger time. Fourth, record the exact filter class and cutoff for every channel in the test log.

After the test, the first validation step is to check the integrated velocity from the body accelerometer against the measured impact velocity from the barrier or the vehicle speed trap. A discrepancy greater than 5% usually indicates a mounting resonance, a saturated channel, or an incorrect filter setting. Only after this check passes should the data be released for CAE correlation. When the test and simulation are compared, the same filter class must be applied to the CAE output, and the time-zero alignment must be established from a common physical event such as barrier contact or airbag trigger.

With the instrumentation chain validated and the data synchronized, the correlation engineer has a trustworthy reference. The next chapter examines how that reference is aligned in time and how correlation metrics are computed.