4. Instrumentation for SAB testing

Side airbag (SAB) testing is fundamentally a measurement problem. The airbag deploys in 10–30 ms, the cushion pressures the occupant's thorax and pelvis, and the whole event is over before a human can react. To correlate a CAE model against a physical test, every sensor in the chain — accelerometer, load cell, pressure transducer, high-speed camera, and the data acquisition system (DAS) that ties them together — must be selected, mounted, and configured so that the recorded signal is a faithful representation of the physical event, not an artifact of the instrumentation. This chapter covers the instrumentation architecture of a typical SAB sled or vehicle crash test and the practical decisions that determine whether the resulting data is usable for CAE correlation.

Measurement Chain Architecture

A SAB test measurement chain has four stages: the transducer converts a physical quantity (acceleration, force, pressure) into an electrical signal; the signal conditioner amplifies, filters, and bridges the transducer to the recorder; the data acquisition system samples and digitizes the conditioned signal; and the recording/analysis layer stores the time history and applies the post-processing filters required by the relevant standard. High-speed cameras run in parallel, capturing the deployment kinematics that no point sensor can resolve.

The critical constraint is time alignment. All channels — analog sensors and camera triggers — must share a common time zero, typically the moment the airbag ECU fires or the sled pulse begins. A 1 ms misalignment between a pressure channel and a camera frame can make a physically correct CAE model appear to be 15% off in peak timing. The DAS must therefore record a common trigger event and, ideally, a synchronization signal (an LED flash or a dedicated sync pulse) visible in the camera field of view.

Accelerometers

SAB tests use accelerometers primarily on the sled or vehicle body (to characterize the crash pulse) and on the occupant's thorax and pelvis (to measure the response). For body-mounted channels, a ±500 g range with a 10 kHz usable bandwidth is typical; for occupant-mounted channels, ±200 g is often sufficient. The dominant practical issue is mounting: a stud-mounted accelerometer with a flat, rigid interface has a resonance well above 20 kHz, but a wax or adhesive-mounted sensor on a soft trim surface can drop the mounted resonance into the measurement band and produce a false peak. Always verify the mounted resonance or use a mechanical low-pass (a thin layer of compliant material) when the surface cannot be made rigid.

For CAE correlation, the accelerometer's axis orientation must be documented in the model coordinate system. A sensor mounted with its positive axis pointing outboard on the left side of the vehicle will read opposite in sign to the same sensor on the right side. Record the orientation vector for every channel in the test log; a sign flip is one of the most common causes of an apparent "CAE mismatch" that is actually a post-processing error.

Load Cells

Load cells in SAB testing measure the force transmitted to the occupant — typically rib deflection force or pelvic force. They are mounted in-line with the loading path, often between the airbag and a rigid or semi-rigid interface. A 0–5 kN range with 1% full-scale accuracy is common. The key practical points are preload and crosstalk: a load cell that is preloaded during installation will read a nonzero baseline, and a multi-axis cell will pick up off-axis components. Zero the cell after installation, and if the cell is multi-axis, record all axes even if only one is used for correlation.

Pressure Sensors

In-cushion pressure is the most direct indicator of airbag deployment performance and is the channel most sensitive to inflator and venting differences between CAE and test. Sensors are typically 0–500 kPa absolute or gauge, with a response time under 1 ms. Mount the sensor so its sensing port is flush with the cushion interior and not blocked by the fabric or the inflator gas jet. A sensor placed too close to the inflator will read a local stagnation pressure that is not representative of the cushion's average pressure — this is a common source of a CAE over-prediction that is actually a sensor placement artifact.

High-Speed Cameras

Cameras resolve what point sensors cannot: cushion unfolding, seam opening, and occupant interaction. For SAB testing, 1000–2000 fps is usually sufficient for the deployment phase, but 5000 fps is needed to resolve seam burst and fabric flutter. Two practical requirements dominate: lighting (high frame rates demand high-intensity, flicker-free illumination, typically LED arrays synchronized to the camera) and calibration (a known grid or target in the field of view, used to convert pixel displacement to millimeters). Without calibration, camera data cannot be overlaid on CAE geometry.

Data Acquisition Systems

The DAS must sample all channels simultaneously (or with a known, fixed inter-channel skew) at a rate at least 10× the highest frequency of interest. For a 10 kHz accelerometer bandwidth, a 100 kS/s per-channel rate is appropriate. Use 16-bit or better resolution, and enable the pre-trigger buffer so that data before the trigger event is retained — this is essential for establishing the true t = 0 and for verifying the baseline. Record both raw and filtered data; the raw data is needed for re-filtering at a different CFC, while the filtered data is used for immediate comparison.

Practical Configuration Example

The following is a representative channel configuration for a SAB sled test. Values are illustrative and should be adapted to the specific test standard and sensor datasheet.

Before the test, perform a shunt calibration on every bridge-type channel (load cells, pressure sensors) to verify the full signal path from transducer to recorder. A shunt resistor of known value produces a known equivalent output; if the recorded value deviates by more than 2%, the channel is suspect. After the test, verify that the pre-trigger baseline is flat and near zero; a drifting or noisy baseline indicates a grounding or shielding problem that will corrupt the correlation.

The instrumentation choices described here determine the ceiling of correlation quality. A perfect CAE model cannot compensate for a mis-mounted accelerometer, an uncalibrated camera, or a DAS with a 2 ms channel skew. The next step in the workflow is to take these recorded channels and perform the actual test-to-CAE comparison, which is the subject of the following chapter.