2. Fundamentals of crash test dummies

Crash test dummies — formally called Anthropomorphic Test Devices (ATDs) — are calibrated mechanical surrogates that stand in for a human occupant during a destructive vehicle test. Their purpose is not to survive the crash but to measure it: every accelerometer, load cell, and potentiometer inside an ATD produces a time history that an engineer converts into a number comparable against a regulatory or consumer-information limit. Understanding what an ATD is, how faithfully it mimics a human, and which criteria turn raw signals into pass/fail decisions is the foundation for everything that follows in LS-DYNA dummy modeling.

2.1 What an ATD Actually Is

An ATD is a segmented, instrumented assembly designed to reproduce the mass distribution, joint stiffness, and deformation response of a human body region. The most widely used family in frontal testing is the Hybrid III (50th-percentile male, and its 5th-percentile female and 95th-percentile male variants). Side-impact work typically uses the ES-2re or WorldSID 50th, while the THOR ATD represents a more biofidelic modern frontal dummy with additional chest and abdomen instrumentation.

Structurally, an ATD is built from a steel or aluminum skeleton covered by flesh-like foam and vinyl skin. Key subassemblies include:

• Head — a metal skull with triaxial accelerometers at the center of gravity (CG).
• Neck — deformable rubber or segmented elements with load cells measuring shear, tension, and bending moment.
• Chest — a ribcage with deflection potentiometers or IR-TRACC (Infra-Red Telescoping Rod for Assessment of Chest Compression) sensors.
• Spine and pelvis — accelerometer packages at the lower thoracic and pelvic locations.
• Femur and tibia — load cells measuring axial force and bending moment.

Each channel is sampled at 10 kHz or higher and filtered per SAE J211 before any criterion is computed. This filtering step is not cosmetic: an unfiltered signal contains high-frequency ringing from the metal skeleton that would corrupt the injury metric.

2.2 Biofidelity: How Human Is the Dummy?

Biofidelity is the degree to which an ATD's response matches that of a human cadaver or volunteer under the same loading. It is assessed region by region against corridors derived from post-mortem human subject (PMHS) tests. A dummy is considered biofidelic for a given region when its force-deflection or acceleration response falls within the corridor across the relevant loading range.

Biofidelity is never perfect. The Hybrid III, for example, has a stiff thoracic spine that over-predicts chest acceleration in some oblique load cases, and its neck is calibrated primarily for frontal flexion. This is why the choice of ATD must match the crash mode: using a Hybrid III in a side impact would produce meaningless chest deflection data because the ribcage is loaded in a direction it was not designed to measure.

2.3 Injury Criteria and Their Standards

Raw measurements become engineering decisions only through injury criteria. Each criterion is a formula that reduces a time history to a scalar, compared against a threshold defined by regulation (FMVSS), consumer protocol (Euro NCAP, US NCAP), or industry standard (SAE J211, ISO 9790). The most important criteria are summarized below.

The Head Injury Criterion (HIC) deserves special attention because it is the most frequently cited and most frequently misapplied. HIC integrates resultant head CG acceleration over a sliding time window, raising the average to the 2.5 power to reflect the nonlinear relationship between acceleration and brain injury. The window length matters: HIC₁₅ uses a 15 ms window, HIC₃₆ uses 36 ms, and the two can differ substantially for long-duration events. Always state the window when reporting HIC.

The N_ij criterion combines axial neck force and bending moment, each normalized by an intercept value that depends on the load direction (tension/compression, flexion/extension). The four combinations (N_te, N_tf, N_ce, N_cf) are evaluated separately, and the maximum is reported. A value above 1.0 indicates the neck has exceeded its injury threshold in that mode.

2.4 From Signal to Criterion: A Worked Example

Consider a frontal sled test with a Hybrid III 50th-percentile male. The head CG accelerometer records a resultant acceleration time history. To compute HIC₁₅:

1. Filter the raw acceleration with a CFC 1000 low-pass filter per SAE J211.
2. Convert to units of g (divide by 9.81 m/s²).
3. For every pair of time points (t₁, t₂) with t₂ − t₁ ≤ 15 ms, compute the average acceleration over that window.
4. Evaluate HIC = (t₂ − t₁) · [a_avg]²·⁵.
5. Report the maximum value across all windows.

Suppose the peak occurs between t₁ = 42 ms and t₂ = 55 ms with an average acceleration of 48 g. Then HIC₁₅ = 0.013 · 48²·⁵ ≈ 0.013 · 15,960 ≈ 207. This is well below the 700 limit, indicating a low head-injury risk for this event. If the same average acceleration persisted over a 36 ms window, HIC₃₆ would be roughly 2.4 times larger — a reminder that window length is not a detail.

2.5 Why This Matters for LS-DYNA Modeling

When you build or import an LS-DYNA dummy model, you are not just creating geometry — you are encoding a measurement system. The model's accelerometer output locations must match the physical ATD's CG locations, the joint stiffnesses must reproduce the calibration corridors, and the post-processing must apply the same SAE J211 filters and criterion formulas used on the test data. A model that produces a beautiful animation but reports HIC without specifying the window, or chest deflection without the correct sensor location, cannot be validated against a physical test.