2. Injury channel definitions and measurement chains

In a crash test, the dummy is not merely a passenger; it is an instrumented transducer platform. Every injury criterion used downstream—HIC, chest deflection, femur load, neck moment—is computed from a small set of physical quantities measured by specific sensors at specific locations. Before any filtering, integration, or criterion calculation can be trusted, the measurement chain from physical stimulus to recorded channel must be understood end to end. This section defines the principal injury channels, the transducers that produce them, and the calibration discipline that keeps them defensible.

From Physical Quantity to Recorded Channel

An injury channel is a named, calibrated data stream whose physical meaning is fixed by convention: for example, HEAD_ACC_RES is the resultant acceleration at the head CG, expressed in g, in the dummy head coordinate system. The chain that produces it has four links: the transducer converts a mechanical quantity into an electrical signal; the conditioning and acquisition system amplifies, filters, and digitizes it; the channel definition assigns sign convention, coordinate system, units, and zero offset; and the calibration record ties the digital counts back to a traceable physical reference. A break or ambiguity anywhere in this chain silently corrupts every criterion computed from the channel.

The Four Transducer Families

Accelerometers. Piezoresistive (MEMS) accelerometers dominate crash dummy applications because they respond to DC, which is essential when the signal must be integrated to velocity and displacement. They are mounted at the head CG (three orthogonal axes or a triaxial block), the upper and lower spine, the pelvis, and the tibia. A uniaxial accelerometer is a single channel; a triaxial block yields three channels sharing one mechanical mount. Because integration amplifies low-frequency error, accelerometer zero offset and drift are the dominant error sources for HIC and for chest a(t) traces.

Load cells. Strain-gauge load cells measure force where the dummy structure transmits it: femur axial force, neck upper/lower load cells (three forces and three moments), lumbar spine, and steering-column or knee-bolster contact. A six-axis load cell produces six coupled channels, and its calibration matrix is not diagonal—cross-axis sensitivity must be characterized, not assumed away. Load cells are stiff by design; mounting torque and surface flatness directly affect the measured force path.

Potentiometers and displacement sensors. Chest deflection in the Hybrid III is measured by a potentiometer-based chest band or an IR-TRACC (infra-red telescoping rod and string potentiometer) in the THOR dummy. These are ratiometric devices: output is proportional to excitation voltage, so a drifting supply appears as a drifting deflection. String potentiometers also require a defined zero position and must be checked for free travel before every test.

Other channels. Rotary potentiometers measure neck and knee rotation; pressure sensors appear in some thoracic and abdominal applications. Each follows the same chain logic, but the calibration reference differs—angle fixtures, pressure deadweights, or displacement micrometers.

Calibration: What It Actually Establishes

Calibration is not a single number. For each channel it establishes at least four things: sensitivity (physical units per volt or per count), zero offset (the reading at the defined reference state), linearity and hysteresis over the working range, and cross-axis or cross-channel sensitivity where the transducer is multi-axis. A typical accelerometer certificate might state sensitivity as 10.02 mV/g with 1% transverse sensitivity; a six-axis load cell certificate provides a full 6×6 sensitivity matrix. The acquisition system applies these coefficients to convert raw counts into engineering units, and the channel definition file records the result.

Two calibration events matter operationally. The laboratory calibration is periodic (commonly annual) and traceable to a national standard. The pre-test and post-test checks are performed on the assembled dummy: a shunt calibration or known-load check confirms the electronics path, and a physical zero check confirms the transducer has not been preloaded by mounting. A post-test check that drifts beyond a defined tolerance (for example, more than 2% of full scale on a load cell, or more than 0.5 g offset on an accelerometer) is grounds for flagging the test data, even if the trace looks plausible.

Worked Example: Verifying a Femur Load Channel

Consider a left femur load cell with a certificate sensitivity of 0.500 mV/V at 10 kN full scale and 5 V excitation. The expected full-scale output is 0.500 × 5 V = 2.5 V, so the sensitivity is 2.5 V / 10 kN = 0.25 mV/N. If the acquisition system has a 16-bit ADC over ±5 V, one count is 10 V / 65536 ≈ 0.153 mV, corresponding to about 0.61 N per count. A pre-test shunt calibration that injects a known equivalent load should reproduce the certificate value within tolerance; a deviation of several percent points to a bridge or cabling fault rather than to the dummy.

During the test, the channel is filtered to CFC 600 for femur force per SAE J211. If the post-test zero has shifted by 0.3 kN against a 10 kN full scale (3%), the peak force reading is suspect and the channel should be flagged. The correct response is not to subtract the drift and proceed silently, but to document the shift, inspect the mount for loosening or preload, and decide with the test engineer whether the data are usable.

Common Failure Modes and How They Present

The discipline that ties this together is simple: every injury channel must be traceable from the physical quantity at the dummy to the engineering-unit value in the data file, with a documented calibration and a pre/post-test check. When that chain is intact, the criteria computed downstream—HIC, chest deflection, femur force, Nij—inherit its credibility. When it is not, no amount of sophisticated post-processing can recover the truth.