4. Occupant model selection and positioning

In MADYMO restraint system design, the occupant model is not a passive placeholder — it is the primary measurement instrument of the simulation. The choice of dummy or human body model determines which injury criteria can be evaluated, how the restraint interacts with the body, and how closely the predicted kinematics correspond to physical sled or vehicle crash tests. Equally important is the positioning of that model: a correctly selected dummy placed with an unrealistic posture will produce misleading belt routing, incorrect head-to-airbag timing, and unreliable injury values. This section covers the practical decision logic for selecting Hybrid III, THOR, or a human body model (HBM), and the workflow for defining a reproducible seating posture in MADYMO.

Model Families and Their Design Intent

MADYMO's occupant model library is organized around three families, each with a distinct validation basis and intended application range.

• Hybrid III (50th, 5th, 95th percentile): The regulatory baseline. Its spine is a rigid segmented structure with a curved lumbar region, and its biofidelity is well characterized for frontal, belt-restrained loading. It is the correct choice when the deliverable is a FMVSS 208, ECE R94/R95, or C-NCAP rating value, because the injury criteria (HIC, Nij, chest deflection via the chest potentiometer, femur load) are defined against this dummy's response.
• THOR (50th percentile, and THOR-5F): A more biofidelic frontal dummy with a flexible thoracic spine, four-point chest deflection measurement, and improved face and neck response. THOR is preferred for occupant safety development where chest deflection distribution, out-of-position (OOP) airbag interaction, or oblique loading matters. Its injury risk functions differ from Hybrid III, so results are not interchangeable.
• Human Body Models (e.g., MADYMO's facet-based or FE human models): Used for research-grade questions — strain-based injury prediction, submarining, occupant variation studies, and pre-crash posture effects. HBMs require more careful boundary conditions and longer run times, and their output is typically strain, stress, or organ-level metrics rather than regulatory criteria.

A practical selection rule: if the result must be compared to a regulation or a physical dummy test, use the dummy that the regulation or test specifies. If the question is "why does the chest deflect this way" or "what happens to a 5th percentile female in a reclined seat," move to THOR or an HBM. Mixing families within one correlation study is a common source of unexplainable deltas.

Positioning: From Reference Posture to Seated Posture

MADYMO positions an occupant model through a combination of a global reference point and joint-level posture. The standard workflow is:

1. Place the model's H-point (hip point) at the target seat H-point using the model's H_point reference and a POS or ORIENT definition in the system tree.
2. Apply the seat back angle and cushion angle so the pelvis and torso contact the seat surfaces without initial penetration.
3. Set joint angles — pelvis, lumbar, thoracic, neck, and limb joints — to the target posture.
4. Run a short settling simulation (typically 20–50 ms) to let the model settle onto the seat under gravity, then freeze the settled state as the initial condition for the crash run.

The settling step is not optional. If the model is initialized in a posture that intersects the seat or the belt, the contact solver will push it out during the first few milliseconds, producing a spurious pre-crash motion that contaminates the entire event.

Defining a Reproducible Seating Posture

Posture definition in MADYMO is expressed through joint rotations relative to the model's neutral (reference) posture. The neutral posture is the model's as-delivered configuration; all user posture changes are deltas from it. This convention must be stated explicitly in any study documentation, because a "10° recline" can mean 10° from neutral or 10° from vertical depending on the source.

For a typical frontal driver study, a reproducible posture specification looks like this:

Occupant: Hybrid III 50th percentile, MADYMO facet model
H-point:  seat H-point, x = 0.0 mm, z = 0.0 mm (seat coordinate origin)
Seat back angle:      25° from vertical
Seat cushion angle:    5° from horizontal
Pelvis rotation:       0° (neutral)
Lumbar joint:         -5° (slight recline)
Thoracic joint:        0°
Neck joint:            0°
Head rotation:         0°
Upper arm (L/R):      15° forward, 5° inward
Forearm (L/R):        30° flexion
Knee (L/R):           95° flexion
Ankle (L/R):          90°

These values are illustrative and must be adjusted to the actual vehicle package. The key discipline is to record every joint value, the seat angles, and the settling duration in the model setup file so the posture can be reproduced exactly in a later run or by another engineer.

Posture Sensitivity and Common Failure Modes

Small posture changes produce disproportionately large changes in restraint interaction. A 5° increase in seat back recline moves the chest rearward and downward, which can cause the shoulder belt to migrate toward the neck and the lap belt to load the abdomen rather than the pelvis — the classic submarining precursor. A 10° forward head rotation changes the head-to-airbag contact timing by several milliseconds, which directly shifts HIC.

Typical symptoms and their posture-related causes:

• Belt slip off the shoulder: seat back too reclined, or shoulder belt anchor too far inboard.
• Submarining (lap belt above the ASIS): pelvis rotated rearward, cushion angle too steep, or lap belt anchor too high.
• Head misses the airbag: head rotated away from the airbag center, or torso leaned outboard.
• Spurious chest deflection at t = 0: initial penetration between the belt and the chest, not a real loading event.

Each of these is diagnosed by inspecting the model at t = 0 and during the first 10 ms, not by looking only at the final injury values.

Model Selection and Posture as a Coupled Decision

Model selection and posture definition are not independent. THOR's flexible spine means its settled posture differs from Hybrid III's for the same joint inputs, and an HBM's posture is defined by skeletal joint angles that do not map one-to-one onto dummy joints. When switching models mid-project, re-derive the posture rather than copying joint values. The figure below summarizes the decision path from study objective to model and posture setup.

Practical Checklist for a New Occupant Setup

The most common project-level error is treating the occupant model as a fixed input while iterating on restraint parameters. In reality, posture and model choice interact with every restraint variable: belt geometry, airbag venting, and anchor positions all produce different results depending on which dummy is seated and how. Establish the model and posture first, document them completely, and only then begin restraint optimization. This ordering keeps the optimization loop interpretable and the final design defensible against physical test correlation.