4. Test data acquisition for reconstruction

Reconstruction of a linear impact test with airbag CAE is only as credible as the measurement chain that feeds it. Before any LS-DYNA deck is tuned, the analyst must answer a deceptively simple question: what did the physical event actually do, and when did it do it? This chapter covers the four pillars of test data acquisition for reconstruction — accelerometers, load cells, high-speed video, and the synchronization infrastructure that binds them into a single coherent time base.

Accelerometers: Capturing the Rigid-Body and Structural Response

Accelerometers are the primary transducers for characterizing the sled or vehicle deceleration pulse and the occupant/airbag interaction. In a linear impact configuration, the most common placements are:

• Sled/barrier accelerometers — mounted on the moving mass or fixed barrier to capture the primary impact pulse. These are typically piezoresistive or capacitive MEMS sensors with a range of ±500 g to ±2000 g and a frequency response from DC to 5–10 kHz.
• Occupant/ATD accelerometers — head, chest, and pelvis accelerometers on the Anthropomorphic Test Device (ATD), typically ±250 g to ±500 g range, used to validate the airbag's interaction with the occupant.
• Structural accelerometers — mounted on the seat frame, steering column, or instrument panel to capture local vibration modes that may affect airbag deployment timing.

Sensor selection is governed by the expected peak acceleration and the frequency content of interest. For airbag reconstruction, the critical frequency band is typically 0–2 kHz; higher-frequency content above 5 kHz is usually filtered out during post-processing to avoid aliasing and to match the CAE model's mesh resolution. A common mistake is to select a sensor with insufficient range, causing clipping, or with excessive range, reducing resolution. As a rule of thumb, the sensor range should be 1.5–2× the expected peak.

Mounting is equally critical. A poorly mounted accelerometer can introduce resonance artifacts that masquerade as real structural response. Stud mounting with a thin layer of cyanoacrylate or a mechanical clamp is preferred over adhesive tape. The mounting resonance should be at least 5× the highest frequency of interest.

Load Cells: Measuring Force at the Interface

While accelerometers capture motion, load cells capture the forces that cause it. In linear impact testing, load cells are used in several key locations:

• Barrier load cells — instrumented barriers or load walls measure the total force transmitted during impact. These are typically piezoelectric or strain-gauge-based, with capacities ranging from 50 kN to 500 kN depending on the test severity.
• Seat belt load cells — inline load cells on the belt webbing measure restraint forces, which are essential for validating the airbag-belt interaction in the CAE model.
• Steering column load cell — measures the force transmitted through the column during driver airbag deployment and occupant contact.

Load cell data is particularly valuable for reconstruction because it provides a direct boundary condition for the CAE model. Unlike acceleration, which must be integrated to obtain velocity and displacement, force data can be applied directly to the model as a time-dependent load. However, load cells are sensitive to off-axis loading and temperature drift; proper calibration and zeroing before each test are mandatory.

High-Speed Video: The Visual Ground Truth

High-speed video serves as the visual reference that validates the timing and sequence of events. For airbag reconstruction, typical frame rates range from 1,000 to 10,000 fps, with resolution sufficient to resolve the airbag fabric and the ATD's head/chest motion. A 1,000 fps camera provides 1 ms temporal resolution, which is adequate for capturing the overall deployment sequence; 5,000–10,000 fps is used when detailed fabric folding or venting behavior must be observed.

Camera placement follows the reconstruction objectives:

• Frontal view — captures the airbag deployment toward the occupant and the ATD's forward excursion.
• Side view — captures the airbag's lateral expansion and the interaction with the seat structure.
• Overhead view — captures the out-of-position (OOP) geometry and the airbag's interaction with the steering wheel or instrument panel.

Lighting is often the limiting factor. High frame rates demand intense, flicker-free illumination. LED arrays synchronized to the camera shutter are standard. Without adequate lighting, motion blur renders the video useless for quantitative analysis.

Data Synchronization: The Backbone of Reconstruction

The most overlooked yet most critical aspect of test data acquisition is synchronization. Accelerometers, load cells, and cameras operate on different clocks, and even a 1 ms misalignment can cause significant errors in the reconstruction. A 1 ms timing error at 50 km/h corresponds to approximately 14 mm of displacement — enough to change the airbag's interaction with the occupant.

Synchronization is achieved through a common trigger signal, typically a TTL pulse generated by the data acquisition system (DAQ) at the moment of impact. This pulse is routed to:

• All transducer signal conditioners, to start or mark the acquisition window.
• The high-speed camera controller, to trigger recording or to mark a specific frame.
• A time-code generator, which embeds a synchronized timestamp into the video stream.

In practice, the DAQ samples all analog channels at a common rate (typically 10–100 kHz) and records the trigger pulse as a dedicated channel. The camera system records its own frame timestamps, which are then aligned to the DAQ time base using the trigger pulse. Post-test, the analyst verifies synchronization by checking that the trigger pulse appears at the same relative time in both the analog and video data.

Putting It Together: A Typical Acquisition Chain

The following SVG illustrates the signal and data flow from the physical test to the reconstruction-ready dataset. It shows how accelerometers, load cells, and cameras are triggered from a common source and how their data streams are aligned in the time domain.

Practical Considerations and Common Pitfalls

Even with a well-designed acquisition chain, several issues routinely degrade data quality:

• Ground loops and noise — accelerometer and load cell signals are low-level and susceptible to electromagnetic interference. Use shielded twisted-pair cables, isolate sensor grounds, and verify noise floor before the test.
• Camera frame drops — high-speed cameras can drop frames if the internal buffer overflows. Always verify the recorded frame count against the expected count for the recording duration.
• Trigger latency — different devices may have different trigger latencies. Measure and compensate for these latencies during post-processing.
• Time base drift — if the DAQ and camera clocks drift relative to each other, long recordings will show increasing misalignment. Use a common time-code generator for recordings longer than 500 ms.

From Acquisition to Reconstruction

The output of a successful acquisition is a synchronized dataset containing:

• Time-history curves for all accelerometers and load cells, with a common time base.
• High-speed video files with frame timestamps aligned to the same time base.
• A test log documenting sensor locations, ranges, sampling rates, filter settings, and any anomalies observed during the test.

This dataset becomes the input for the CAE reconstruction. The accelerometer data defines the sled pulse; the load cell data provides boundary conditions; the video provides the visual sequence for validating the airbag deployment timing and occupant kinematics. Without rigorous synchronization, the reconstruction becomes a curve-fitting exercise rather than a physics-based validation. With it, the CAE model can be tuned with confidence, and the resulting airbag performance predictions carry real engineering weight.