4. Dummy models in MADYMO

MADYMO does not ship a single "dummy." It ships a family of validated finite-element and multi-body occupant models, each tied to a specific physical anthropomorphic test device (ATD), a specific regulation or consumer-test protocol, and a specific range of loading conditions. Choosing the wrong model is one of the most common and most expensive mistakes in an occupant safety study: the simulation may converge, the animations may look plausible, and the injury numbers may still be meaningless because the model was never validated for that impact direction or that body region. This chapter explains how the four main model families differ, what their validation status actually covers, and how to scale them when the target population does not match the 50th-percentile baseline.

4.1 The Four Model Families and What They Represent

MADYMO's dummy library is organized around the physical ATDs used in crash test laboratories. The Hybrid III family (50th, 5th, and 95th percentile male/female) is the workhorse of frontal regulation: FMVSS 208, ECE R94, and most NCAP frontal programs are written around it. Its MADYMO implementation is available both as a multi-body model (fast, robust, ideal for parameter sweeps and restraint optimization) and as a facet/FE model (needed when belt interaction, chest deflection distribution, or local contact pressure matters).

THOR (Test Device for Human Occupant Restraint) is the modern frontal ATD, designed to fix well-known Hybrid III weaknesses: a more biofidelic thorax with four-point deflection measurement, a flexible lumbar spine, improved pelvis and femur load paths, and a face/head design that responds more realistically to airbag loading. In MADYMO, THOR is predominantly an FE model. It is the reference dummy for many advanced frontal protocols and for out-of-position (OOP) airbag studies, but it is also stiffer computationally and more sensitive to contact definitions than Hybrid III.

WorldSID is the side-impact counterpart: a 50th-percentile male side-impact ATD with a segmented ribcage, abdominal inserts, and instrumentation for chest, abdomen, and pelvis. It is the reference for UN R95, Euro NCAP side impact, and pole tests. MADYMO provides WorldSID as a detailed FE model; the multi-body variant is used mainly for early restraint-layout studies where rib-level fidelity is not yet required.

Q-series dummies (Q0, Q1, Q1.5, Q3, Q6, Q10) cover children from newborn to roughly ten years old. They are the reference for ECE R44/R129 (i-Size) and for child-restraint system (CRS) development. Q-series models in MADYMO are FE models with scaled geometry and mass distribution; they are not simply "shrunk adults," and treating them as such is a frequent source of error in CRS studies.

4.2 Validation Status: What the Model Is Actually Good For

Every MADYMO dummy model ships with a validation report. Reading that report before running a study is not optional. The report tells you which body regions were correlated against cadaver or PMHS data, which impact directions were tested, which loading corridors were matched, and — critically — where the model is known to be weak.

A practical way to think about validation status is to separate three questions:

• Regulatory validity: does the model reproduce the certification corridors of the physical ATD (e.g., Hybrid III head drop, chest calibration, knee slider)? This is necessary but not sufficient for injury prediction.
• Biofidelity: does the model respond like a human under the loading you are simulating? This is what matters for injury risk curves.
• Application validity: has the model been used and correlated for your specific scenario — belt-only, belt-plus-bag, OOP, pole, far-side, rear-seat, etc.?

Hybrid III has the broadest regulatory validity and the longest application history, which is why it remains the default for frontal restraint optimization. THOR has better biofidelity in the thorax and pelvis but a shorter application history; its validation is strongest for frontal and oblique frontal loading. WorldSID is well validated for pure lateral and oblique lateral loading, but its response under combined frontal-lateral loading is less mature. Q-series biofidelity is strongest for frontal CRS loading; lateral and rear-facing cases rely more on scaling assumptions.

4.3 Scaling: From 50th Percentile to the Target Occupant

Scaling is the process of adapting a baseline dummy model to represent a different occupant size, mass, or posture. MADYMO supports two conceptually different approaches, and confusing them is a common error.

Geometric scaling (also called "morphing" or "node scaling") transforms the mesh: node coordinates are scaled by a factor derived from anthropometric ratios, and shell thicknesses and solid elements are adjusted to preserve mass distribution. This is the correct approach when you need a different body size but want to keep the same ATD family and its validation basis. MADYMO's SCALE utility and the dummy_scaling tools in the MADYMO Dummy Models library implement this.

Model substitution means switching to a different, independently validated model (e.g., from Hybrid III 50th to Hybrid III 5th). This is preferred when the target population is already covered by an existing validated model, because it avoids introducing scaling error.

The scaling factors themselves are not arbitrary. A standard approach uses mass scaling and length scaling derived from anthropometric data:

lambda_m = m_target / m_baseline
lambda_l = (m_target / m_baseline)^(1/3)   # assuming constant density
lambda_t = lambda_l                         # time scales with length for
                                            # dynamically similar systems

For a Hybrid III 50th (78 kg, 1.77 m) scaled to a 5th-percentile female (49 kg, 1.51 m), the mass ratio is 0.628, giving a length scale of about 0.857 and a time scale of the same value. In practice, MADYMO's scaling utility applies these factors per body segment rather than globally, because segment mass proportions differ between percentiles. The result is a model that preserves the baseline's validation corridors in a scaled sense, but the injury criteria must be re-evaluated against the scaled occupant's own risk curves — not the 50th-percentile curves.

4.4 A Worked Selection and Scaling Example

Consider a frontal OOP study: a 5th-percentile female driver, seat in the rearmost position, airbag deploying while the occupant is close to the module. The question is which dummy model to use and how to set it up.

The load case is frontal, the occupant is small, and the interaction is dominated by the airbag-to-chest and airbag-to-face contact. Hybrid III 5th percentile is the natural starting point because it is a validated model for that percentile and is widely accepted in OOP protocols. THOR 5th percentile would offer better thorax biofidelity, but its OOP validation is less mature, so it is better used as a cross-check than as the primary model.

Setup steps in MADYMO:

1. Load the Hybrid III 5th percentile FE model from the dummy library and verify the model version against the validation report.
2. Position the dummy using the INITIAL.JOINT_POS and INITIAL.FE_MODEL_POS cards to match the OOP posture defined by the protocol (e.g., head 50 mm from the module).
3. Define the airbag with the correct FOLDER and GASFLOW properties; OOP results are extremely sensitive to the initial bag shape and the inflator mass-flow curve.
4. Run a baseline at the protocol severity and inspect chest deflection, head acceleration, and neck forces against the OOP injury criteria.
5. If the target occupant is a 5th-percentile female but the available model is 50th, apply segment-wise scaling and re-run; compare the scaled result against the 5th-percentile model result to quantify scaling error.

Typical failure symptoms and their causes: if the head penetrates the airbag mesh, the contact thickness or the bag's CONTACT definition is wrong; if chest deflection is unrealistically low, the belt is slipping or the rib material model is too stiff; if the simulation terminates with negative volume errors, the scaling factors have distorted solid elements beyond their stable range.

4.5 Model Selection Decision Logic

The following diagram summarizes the selection logic across the four families, including the validation-status caveats that should drive the final choice.

4.6 Practical Checklist Before Running a Dummy Model

Before committing to a full simulation matrix, verify the following for the chosen model:

The key takeaway is that dummy selection in MADYMO is a validation-scope decision, not a preference. Hybrid III, THOR, WorldSID, and Q-series each carry a specific validation envelope; staying inside that envelope — or explicitly quantifying the error when you leave it — is what separates a defensible occupant safety study from an animation that merely looks convincing.