3. Occupant injury criteria and regulatory requirements

In a restraint-system DOE, the objective functions are not "how much the dummy moved" but whether specific body regions exceed regulated or consumer-rating injury thresholds. Before any sampling plan or surrogate model can be built, the analyst must map each regulation's acceptance criteria onto concrete LS-DYNA output quantities, at the correct location, in the correct coordinate convention, and with the correct filtering. This section establishes that mapping for FMVSS 208, Euro NCAP, and C-NCAP, and shows how to request the corresponding data in a keyword deck.

What each program actually measures

FMVSS 208 (49 CFR 571.208) governs frontal occupant protection in the United States. For the belted driver and right-front passenger in a 56 km/h (35 mph) rigid barrier test, the compliance quantities are the Head Injury Criterion (HIC15, limit 700), the 3 ms chest acceleration clip (60 g), and the chest deflection criterion (63 mm for the Hybrid III 50th male). Femoral axial force (10 kN) and the Nij neck injury criterion are also assessed. FMVSS 208 additionally contains unbelted and out-of-position requirements, but for restraint DOE the belted frontal case is the usual anchor.

Euro NCAP frontal assessment (the Mobile Progressive Deformable Barrier, MPDB, and frontal full-width tests) uses HIC15 (lower assessment threshold 500, upper 700), chest deflection, and adds the head injury criterion accumulated over 3 ms plus neck tension/compression and shear. Euro NCAP also scores femur and knee loads and, importantly, applies a modifier for bottoming-out of the restraint and for excessive occupant excursion. The rating is banded, not pass/fail, so the DOE objective is usually to minimize the normalized score rather than to clear a single limit.

C-NCAP (China) frontal 50 km/h full-width rigid barrier and 64 km/h ODB tests use HIC15 (limit 700 for full marks, with a high-performance threshold of 650), 3 ms chest acceleration (38 g high-performance, 60 g limit), chest deflection, and neck criteria. C-NCAP's scoring is also banded and includes a penalty for restraint-system bottoming-out and for occupant excursion beyond a defined envelope.

The three programs share the Hybrid III 50th percentile male as the primary frontal dummy, so a single LS-DYNA model can serve all three — but the thresholds, banding, and modifier rules differ. Never reuse a Euro NCAP score function for a C-NCAP objective without re-deriving the band edges.

From regulation to LS-DYNA output: the request chain

Injury criteria are computed from time histories, not from peak scalars. The standard LS-DYNA workflow is: (1) request the relevant nodal/element/rigid-body time-history data via *DATABASE_ASCII_option or *DATABASE_BINARY_D3PLOT; (2) post-process with a script (LS-PrePost, Python, or a DOE driver) that applies the regulation's filter and window; (3) write the scalar criterion into the DOE response table. The figure below shows this chain for the head and chest channels, which dominate frontal restraint DOE.

Injury-criterion request chain: model output → filtered history → DOE response 1 · LS-DYNA output requests *DATABASE_ASCII NODOUT (head CG accel) ELOUT (chest deflection) *DATABASE_HISTORY_NODE head CG node set chest accelerometer nodes *DATABASE_BINARY_D3PLOT dt = 0.5 ms (animation) for excursion / bottoming check *DATABASE_HISTORY_RBODY femur / knee rigid bodies axial force, moment 2 · Filter & compute Head a_res = √(ax²+ay²+az²) CFC 1000 filter HIC15 = max over 15 ms window Chest a_res, CFC 180 filter 3 ms clip: max of 3 ms avg deflection from ELOUT / spring Neck / Femur Nij = Fz/Fzc + My/Myc femur axial force (kN) knee slider displacement Excursion / bottoming from D3PLOT geometry 3 · DOE response table (one row per run) run HIC15 a3ms(g) defl(mm) Nij femur(kN) 01 612 48.2 41.5 0.71 4.8 02 688 55.7 52.1 0.83 5.6 03 574 44.0 38.9 0.66 4.2 04 731 61.3 58.4 0.91 6.1 … … … … … … Thresholds applied downstream FMVSS 208: HIC15 ≤ 700, a3ms ≤ 60 g, defl ≤ 63 mm Euro NCAP: HIC15 500/700 band, defl band C-NCAP: HIC15 650/700, a3ms 38/60 g Normalize each to [0,1] before MOO (see Section 4 for normalization)

Concrete keyword requests

The following deck fragment requests the channels needed for HIC15, the 3 ms chest clip, and chest deflection on a Hybrid III 50th male. Node and element IDs are illustrative; in practice they come from the dummy's *SET_NODE / *SET_SOLID definitions.

*DATABASE_ASCII
$   DT     LCDT    BINARY   ASCII    ...
   1.0e-4   1.0e-4   0        1
$   NODOUT   ELOUT   ...
    1        1

*DATABASE_HISTORY_NODE
$   NSID
    1001      $ head CG node
    1002      $ chest accelerometer node (upper)
    1003      $ chest accelerometer node (lower)

*DATABASE_HISTORY_SOLID
$   ESID
    2001      $ chest deflection element set

*DATABASE_BINARY_D3PLOT
$   DT
    5.0e-4

Two details routinely cause wrong HIC values. First, the head CG acceleration must be the resultant of the three translational components, filtered at CFC 1000 (SAE J211), before the 15 ms window search. If you compute HIC on an unfiltered or single-axis signal, the value will be biased high or low by several percent. Second, the 3 ms chest clip is the maximum of the running 3 ms average of the resultant chest acceleration filtered at CFC 180 — not the peak instantaneous acceleration. A common error is to report the raw peak, which overstates the clip by 10–20 g.

Banded scoring and the modifier problem

Euro NCAP and C-NCAP do not simply pass or fail; they assign points within a band and then apply modifiers. For a DOE, the cleanest approach is to convert each criterion to a normalized penalty in [0, 1] using the program's own band edges, then sum with the program's weights. For example, a Euro NCAP HIC15 of 600 sits at (600−500)/(700−500) = 0.50 of the penalty range. The modifier for restraint bottoming-out is binary and must be evaluated from the D3PLOT animation or from a contact-force threshold on the belt/airbag interface — it cannot be read from a scalar history alone. This is why the D3PLOT request in the figure is not optional: the excursion and bottoming checks are geometric.

Do not mix criteria from different programs in a single objective without documenting the normalization. A DOE that minimizes "HIC15" against the FMVSS 208 limit of 700 will systematically favor stiffer restraints than one scored against the Euro NCAP 500-point band edge. State the target program explicitly in the DOE setup and keep the band edges in a versioned configuration file.

Practical checklist for the output-request stage

  • Confirm the dummy variant (Hybrid III 50th vs. THOR) — the chest deflection measurement location and scaling differ, and THOR adds abdominal and iliac channels.
  • Verify the accelerometer node locations against the dummy manual; a 10 mm offset at the head CG shifts HIC by roughly 2–4%.
  • Set DT in *DATABASE_ASCII to at least 0.1 ms so the CFC 1000 filter is not undersampled; 0.05 ms is safer for HIC.
  • Request D3PLOT at ≤ 0.5 ms for excursion and bottoming review.
  • Log the LS-DYNA version and the dummy model revision alongside each run — injury criteria are sensitive to model updates.

With the request chain fixed and the band edges documented, the response table becomes a clean, program-specific set of objectives. The next step — normalizing these objectives and choosing the multi-objective optimization formulation — is where the DOE design and the injury criteria meet.

← Previous 📖 Contents Next →