4. Anthropomorphic Test Devices (ATDs)

An Anthropomorphic Test Device (ATD), commonly called a crash test dummy, is a calibrated mechanical surrogate for a human occupant. Its purpose is not to reproduce human anatomy faithfully in every detail, but to produce repeatable, measurable responses that correlate statistically with human injury risk under defined loading conditions. Every design decision in an ATD — from the vinyl skin formulation to the lumbar spine stiffness — exists to serve that measurement function. Understanding what an ATD measures, how it is calibrated, and where its biofidelity breaks down is essential for interpreting any crash test result correctly.

Design Architecture and the Hybrid III Family

The Hybrid III 50th-percentile male is the most widely used frontal impact ATD and a useful reference architecture. Its structure is layered: a welded steel and aluminium skeleton provides load paths; segmented rubber and urethane components (neck, lumbar spine, abdomen) provide compliance; and a vinyl skin with foam inserts gives the outer contour and distributes contact pressure. The head is an aluminium skull with a vinyl skin and a removable rear skull cap that houses accelerometers. The neck is a set of butyl rubber discs bonded between aluminium plates, tuned to produce a specific flexion–extension moment–angle response. The thorax contains a steel rib cage with a damping material, and the pelvis is a rigid vinyl structure.

Other family members scale this architecture to different occupant sizes: the Hybrid III 5th-percentile female, the 95th-percentile male, and the 6-year-old and 3-year-old child ATDs. Side impact uses the ES-2re (50th male) and SID-IIs (5th female), which replace the rib cage with a deformable side structure and add a floating arm to manage arm-to-torso interaction. The THOR (Test Device for Human Occupant Restraint) family adds a more human-like clavicle, a four-rib thorax with independent deflection measurement, and a more biofidelic pelvis, at the cost of greater complexity and calibration burden.

Instrumentation and Measurement Channels

ATD instrumentation converts mechanical response into electrical signals that the data acquisition system records. The core channels are accelerometers, load cells, potentiometers, and deflection sensors. In a frontal Hybrid III, the standard channel set includes:

• Head: three orthogonal accelerometers at the head CG, from which the resultant head acceleration and the Head Injury Criterion (HIC) are computed.
• Neck: a six-axis upper-neck load cell measuring Fx, Fy, Fz, Mx, My, Mz, used for Nij and neck injury assessment.
• Chest: a three-axis chest accelerometer and a chest deflection potentiometer (or IR-TRACC in THOR) measuring sternal deflection relative to the spine.
• Femur: axial load cells measuring femur compression for knee–thigh–hip injury assessment.
• Pelvis: a three-axis accelerometer for pelvic acceleration.

Each channel has a defined range, sensitivity, and filter class. For example, head accelerometers are typically rated to ±2000 g with a 1000 Hz channel filter class (CFC 1000) per SAE J211, while chest deflection uses CFC 180. Applying the wrong filter class changes the peak value and therefore the injury metric — a common source of inconsistency between labs.

Calibration: Certifying the Measurement Chain

An ATD is only valid within its calibration envelope. Calibration is performed at defined intervals (typically every test or every few tests, depending on the component) and certifies that each subsystem responds within a specified corridor to a standardized input. The neck, for example, is calibrated by mounting it in a pendulum and measuring the moment–angle response during a controlled swing; the thorax is calibrated by a chest impact with a known pendulum mass and velocity, checking deflection and hysteresis. Head drop tests verify the skull's acceleration response.

The practical workflow for a pre-test check is:

1. Verify the ATD serial number and calibration certificate dates against the test plan.
2. Perform a visual inspection for skin damage, loose fasteners, and sensor cable integrity.
3. Run the specified component calibration tests (neck pendulum, chest impact, head drop) and compare results to the certificate corridors.
4. Zero and range-check every instrumentation channel through the data acquisition system.
5. Document any deviation and obtain engineering disposition before proceeding.

A neck calibration that falls outside the moment corridor, for instance, indicates rubber disc degradation or a loose bond; the neck must be replaced or reworked before the test, because an out-of-calibration neck invalidates Nij and neck injury metrics.

Limitations for Injury Assessment

ATDs are biofidelic only within the loading regimes they were designed and validated for. Key limitations include:

• Anthropometry and stiffness mismatch: the Hybrid III is stiff compared with human tissue, particularly in the thorax and abdomen, which can over- or under-predict injury in oblique and far-side loading.
• Limited degrees of freedom: the spine and pelvis cannot reproduce human kinematics in complex multi-axis loading, so out-of-position and rollover scenarios are poorly represented.
• Injury criteria are statistical, not deterministic: HIC, Nij, and chest deflection are correlates of injury risk derived from cadaver and volunteer data, not direct measures of tissue damage.
• Age and population coverage: child ATDs and elderly occupants are underrepresented; scaling from the 50th male introduces error.
• Sensor placement effects: accelerometers measure rigid-body response at a point, not local tissue strain, so diffuse injury may be missed.

Reading an ATD Response: A Worked Example

Consider a frontal sled test at 56 km/h with a Hybrid III 50th male restrained by a three-point belt and a force-limiting retractor. The head acceleration trace shows a peak of 85 g at 62 ms with a CFC 1000 filter. Integrating the resultant acceleration over the HIC time window (typically 36 ms) yields a HIC value; suppose the computed HIC₁₅ is 620. The chest deflection channel peaks at 38 mm, and the upper-neck load cell shows a flexion moment of 95 N·m with an axial tension of 2.1 kN. Cross-referencing these against the relevant injury assessment reference values (IARVs) — for example, HIC₁₅ ≤ 700, chest deflection ≤ 63 mm, Nij ≤ 1.0 — shows the head and chest are within limits, but the neck Nij should be computed explicitly from the tension and moment channels rather than judged from a single peak.

The lesson is procedural: the ATD produces numbers, but the analyst must apply the correct filter class, the correct time window, and the correct injury criterion formulation before drawing a conclusion. A raw peak value without its filter and window context is not an injury metric.

The diagram above makes the chain explicit: mechanical loading excites the ATD structure, transducers convert the response to electrical signals, the DAQ applies the specified filter class, and only then are injury criteria computed and compared to reference values. Calibration is the feedback loop that certifies every stage before the test. When a result looks anomalous, the analyst should walk this chain in order — loading, structure, transducer, filter, criterion — rather than jumping to a conclusion about occupant safety.

Practical Takeaways

Treat the ATD as a calibrated instrument, not a human substitute. Verify calibration certificates and component corridors before every test; record filter classes and time windows with every metric; and interpret injury criteria as statistical risk correlates rather than deterministic outcomes. When kinematics look atypical, investigate the restraint and the ATD response together, because a metric within limits does not by itself confirm a safe design.