2. Crash dummy fundamentals

An Anthropomorphic Test Device (ATD) is a calibrated mechanical surrogate for a human occupant, built to reproduce selected biomechanical responses under crash loading while surviving repeated, highly repeatable use. In seating-position and belt-fit verification, the ATD is not a passive mannequin: it is the reference body that defines where the belt should lie, how the pelvis should be oriented, and which loads and accelerations must be recorded. Understanding what each dummy family measures — and what it deliberately does not measure — determines whether a belt-fit check is meaningful or merely cosmetic.

What Makes an ATD an Instrument

Every ATD is a chain of four subsystems: a skeletal structure (metal or composite spine, ribs, pelvis, and limbs), deformable soft tissue (foam or vinyl flesh and damping elements), joints with defined stiffness and friction, and instrumentation with a defined calibration corridor. The dummy's response is only valid inside that corridor. A Hybrid III 50th percentile male, for example, is calibrated for frontal impact at approximately 48 km/h sled pulses; using it to judge a low-speed static belt-fit geometry is acceptable, but using it to infer lateral occupant response is not.

Two conventions must be fixed before any measurement is interpreted. First, the SAE J211 sign convention: for a vehicle coordinate system with +X forward, +Y to the occupant's left, and +Z upward, accelerations and forces are reported with those signs, and the dummy's local axes are aligned to the vehicle at the initial seated position. Second, the H-point reference: the ATD's hip point is the origin for belt-fit geometry, and all belt coordinates (e.g., belt centerline offset from the H-point, lap-belt angle relative to horizontal) are expressed relative to it. Mixing a dummy-local frame with a vehicle frame without an explicit transform is one of the most common sources of belt-fit disagreement between labs.

The Three Dummy Families in This Course

Hybrid III is the workhorse frontal ATD, available in 50th percentile male, 5th percentile female, and 95th percentile male sizes, plus child variants (CRABI). Its spine is a segmented rubber-and-metal column, its neck is a deformable rubber element, and its pelvis is a rigid metal casting with a molded flesh form. Instrumentation typically includes head accelerometers (three linear, three angular), upper and lower neck load cells, chest accelerometers and a chest deflection potentiometer, lumbar spine load cells, and femur loads. For belt-fit work, the Hybrid III's rigid pelvis and well-defined iliac crest geometry make it the standard reference for lap-belt placement: the lap belt should cross the pelvis between the anterior superior iliac spine (ASIS) and the femur, not the abdomen.

THOR (Test Device for Human Occupant Restraint) is the modern frontal ATD, with a more human-like spine, four-point chest deflection measurement, a flexible pelvis, and a face/head design that supports facial and brain-injury metrics. THOR's advantage for belt-fit verification is its more realistic pelvis and abdomen: it can distinguish a belt that has slipped onto the abdomen from one properly seated on the pelvis, because its abdominal pressure sensors and pelvic rotation instrumentation respond differently. THOR is also more sensitive to pretensioner and load-limiter strategies, which matters when belt geometry is evaluated under dynamic rather than static conditions.

WorldSID is the side-impact ATD, with a biofidelic shoulder, thorax with rib deflection measurement, abdomen, and pelvis, plus a head designed for side-impact loading. It is not a frontal belt-fit device, but it is essential when the seating position is evaluated for lateral restraint, far-side occupant protection, or when a belt routing crosses the shoulder in a way that could interact with side airbags. WorldSID's shoulder geometry is deliberately different from Hybrid III's, so a shoulder-belt fit that looks acceptable on Hybrid III may not transfer directly.

Measurement Capabilities and What They Mean for Belt Fit

The instrumentation list matters because belt-fit verification is ultimately a question of load path. A lap belt that sits on the abdomen rather than the pelvis will, under frontal loading, produce a different force distribution: the Hybrid III will show elevated lumbar spine load and reduced femur load, while THOR will additionally show abdominal pressure and altered pelvic rotation. A shoulder belt routed too far outboard will show asymmetric chest deflection on THOR's four-point measurement, and a shoulder belt that slips off the shoulder will show a sudden drop in chest deflection and a spike in head excursion. These are the observable signatures that let an engineer judge whether a static belt-fit measurement is likely to hold up dynamically.

For static belt-fit work, the practical measurement set is smaller but must be precise. The standard procedure places the ATD in the seat using the H-point machine or a defined settling procedure, applies the belt with a specified tension (commonly 10–20 N for static fit checks, or the pretensioner's locked position), and then measures belt centerline position relative to the ATD's skeletal landmarks. Key measurements include: lap-belt centerline height above the H-point, lap-belt angle in side view, shoulder-belt centerline offset from the neck, and shoulder-belt angle in front view. Each is read against the dummy's known geometry, which is why the ATD's skeletal landmarks — not its flesh surface — are the reference.

Practical Verification Example

Consider a static belt-fit check on a driver seat with a Hybrid III 50th percentile male. The seat is set to the design H-point, the dummy is settled per the lab procedure, and the belt is routed with 15 N tension. The measured lap-belt centerline sits 42 mm above the H-point and at 52° to horizontal in side view; the shoulder-belt centerline is 95 mm outboard of the neck centerline. The engineer compares these against the program's belt-fit target window (for example, lap belt 30–50 mm above H-point, 50–60° in side view, shoulder belt 80–110 mm outboard). All three fall inside the window, so the static fit is accepted. The same seat is then checked with THOR: the lap-belt height reads 46 mm and the abdominal pressure sensor shows a small preload, indicating the belt is close to the abdominal boundary. The engineer flags this for dynamic confirmation rather than rejecting the static fit, because THOR's flexible pelvis is more sensitive to the boundary condition. This is the correct use of the two dummies: Hybrid III for the stable geometric reference, THOR for the boundary-sensitive check.

Typical failure symptoms are equally instructive. If the lap belt reads low and the femur load cell shows preload during settling, the belt is likely routed under the ASIS and onto the thigh — a geometry error, not a dummy error. If the shoulder belt reads inboard and the chest deflection baseline is asymmetric, the belt is likely pinched by the seat back or the D-ring height is wrong. If the H-point measurement itself drifts between setups, the settling procedure or the seat cushion compression is the suspect, and no belt-fit number should be trusted until the H-point is stable. In each case, the ATD's instrumentation tells you where to look; the skeletal landmarks tell you what the correct geometry should be.

Selection Guidance

Choose the ATD by the question being asked, not by availability. Use Hybrid III when the task is a stable, repeatable geometric reference for frontal belt routing and H-point-based measurements. Use THOR when the belt path is near the abdominal boundary, when pretensioner and load-limiter interaction matters, or when the program requires abdominal and multi-point chest data. Use WorldSID when the seating position involves lateral restraint, far-side protection, or shoulder-belt interaction with side structures. In all cases, record the dummy type, calibration date, settling procedure, and coordinate convention with the measurement — because a belt-fit result is only reproducible when its reference body is fully specified.