4. Restraint system performance signals

In a crash event, the occupant restraint system is a coordinated chain of energy-absorbing devices, and each device leaves a distinct physical signature in the recorded data. To reconstruct how an occupant was restrained — and where the restraint may have under- or over-performed — an analyst must read four signal families together: seat belt force, airbag pressure, pretensioner activation, and load limiter behavior. None of these signals is meaningful in isolation; the diagnostic value comes from their timing, magnitude, and mutual consistency.

This chapter explains what each signal physically represents, how it is acquired in a crash test or instrumented vehicle, and how to interpret the combined trace to judge restraint performance. The reference convention used throughout is: positive belt force = tension (belt pulling on the occupant), and positive airbag pressure = internal pressure above ambient. Time is measured from the crash pulse trigger (t = 0 at impact), and all event timings are reported in milliseconds after trigger.

What Each Signal Physically Represents

Belt force is measured with an in-line load cell or a load-sensing buckle/retractor. It captures the tensile load transmitted through the webbing into the occupant's pelvis and chest. A typical three-point belt in a frontal impact shows a rapid rise as the occupant loads the belt, a plateau governed by the load limiter, and a decay as the occupant rebounds. Peak values in a 50 km/h frontal sled test commonly fall in the 4–8 kN range for the shoulder belt, but the exact value depends on the load limiter threshold and occupant mass.

Airbag pressure is acquired with a pressure transducer threaded into the inflator or module cavity. It reflects the gas dynamics inside the cushion: a sharp inflation spike, a venting decay as gas escapes through the vents and fabric porosity, and a secondary rise when the occupant compresses the cushion. Pressure is not the same as force on the occupant, but the pressure–time curve is a reliable proxy for deployment timing and cushion stiffness.

Pretensioner activation is usually a discrete or short-duration signal — a pyrotechnic current pulse, a buckle stroke displacement, or a retractor spool rotation. It marks the moment the slack is removed and the occupant is coupled to the belt. Because pretensioning happens in the first 10–20 ms, it strongly shapes the early belt-force rise.

Load limiter behavior is not a separate sensor but a characteristic visible in the belt-force trace: once force reaches the limiter threshold, the belt pays out webbing (via a torsion bar, stitch tear, or friction element) and the force plateaus rather than continuing to climb. The plateau level, its duration, and the payout distance are the diagnostic features.

Reading the Combined Trace

The SVG below shows a schematic timing relationship between the four signals for a representative frontal impact. It is a teaching schematic, not measured data: the shapes and relative timings are illustrative and intended to show how the signals interlock.

Three coupling rules govern interpretation. First, the pretensioner pulse should precede the belt-force rise by only a few milliseconds; a long gap suggests slack that the pretensioner failed to remove, or a delayed firing command. Second, the belt-force plateau should coincide with the load limiter payout window; if force continues to climb while payout is active, the limiter may be binding or the payout path may be obstructed. Third, the airbag pressure peak should arrive after belt engagement but before the occupant's forward excursion reaches the cushion's maximum compression — a pressure peak that arrives too early means the occupant is out of position, and one that arrives too late means the cushion is already venting when the chest arrives.

Acquisition and Signal Conditioning

Belt-force load cells are typically full-bridge strain-gauge devices sampled at 10 kHz or higher, with a low-pass filter at 1–2 kHz to remove structural ringing while preserving the plateau. Airbag pressure transducers are sampled at the same rate; because the inflation spike is steep, undersampling will clip the peak and bias the timing estimate. Pretensioner activation is often captured as a digital channel or a current shunt trace; if only a discrete flag is available, the analyst loses the pulse duration, which matters for diagnosing partial activation.

A practical conditioning step is to align all channels to a common trigger and to verify sensor polarity before analysis. A reversed belt-force channel will show a negative-going trace that looks like belt slack; a reversed pressure channel will show a vacuum spike. Both are common wiring errors and should be ruled out before any performance judgment is made.

Worked Example: Diagnosing a Late Belt Engagement

Consider a frontal sled test at 50 km/h with a 50th-percentile male Hybrid III. The recorded traces show a pretensioner pulse at 8 ms, a belt-force rise beginning at 22 ms, a plateau at 5.5 kN from 35 to 70 ms, and an airbag pressure peak at 28 ms. The analyst notes that the belt-force rise lags the pretensioner by 14 ms — longer than the expected 3–6 ms coupling window.

The diagnostic sequence is: (1) confirm the pretensioner pulse duration and stroke; a short pulse with no buckle stroke indicates a firing circuit fault rather than a mechanical delay. (2) Check the belt-force channel for a flat segment before the rise; a flat segment at zero force means the belt was still slack when the pretensioner fired. (3) Compare the airbag pressure peak timing with the belt rise; here the pressure peaks before the belt engages, which means the occupant's chest is being decelerated by the airbag before the belt is loaded — a classic signature of excessive initial slack or a retractor that did not lock promptly.

The corrective action in this example is to inspect the retractor locking mechanism and the pretensioner stroke, and to re-run the sled test with a buckle-stroke displacement channel added. The expected verification is a belt-force rise within 3–6 ms of the pretensioner pulse and a pressure peak that follows belt engagement.

Interpreting Load Limiter Behavior

The load limiter is the component most often misread, because its effect appears as a change in slope rather than as a discrete event. A well-behaved limiter produces a flat plateau at its design threshold, with payout occurring smoothly over the plateau window. Three failure signatures are worth memorizing:

• No plateau: force rises monotonically through the expected limiter window. This indicates the limiter did not activate — a seized torsion bar, a stitch pattern that did not tear, or a limiter threshold set above the achieved force.
• Premature plateau: force flattens well below the design threshold. This suggests the limiter activated too early, possibly due to a manufacturing variance or a mis-set threshold, and the occupant may be under-restrained.
• Stepped plateau: force shows two or more distinct levels. This can indicate a multi-stage limiter operating as designed, or a binding payout path that releases in increments. The distinction requires the payout displacement channel.

Injury analytics, the load limiter plateau directly sets the chest deflection and the belt-induced rib loading. A plateau that is too high raises the risk of rib fracture and thoracic injury; a plateau that is too low allows greater forward excursion and increases the risk of contact with the interior. The analyst's job is to confirm that the observed plateau matches the design intent for the occupant size and impact severity, and to flag any deviation for further investigation.

Putting the Signals Together

The four signal families form a coherent story only when read as a sequence: pretensioner activation removes slack, belt force rises and then plateaus at the limiter threshold, and airbag pressure peaks in the window between belt engagement and maximum chest compression. Any deviation from this sequence — a late belt rise, a missing plateau, a pressure peak that precedes belt engagement — is a diagnostic lead, not a conclusion. The analyst should confirm each lead against an independent channel, check sensor polarity and scaling, and only then attribute the deviation to a restraint performance issue. This disciplined reading is what separates a defensible injury-analytics finding from a plausible-looking but unsupported one.