3. Regulatory and consumer test protocols

In occupant restraint CAE correlation, the test protocol is not background information — it is the specification that defines what "correct" means. A restraint model can be numerically stable, mesh-converged, and physically plausible, yet still be judged a failure because it was correlated against the wrong loading condition, the wrong impact angle, or the wrong injury criterion. FMVSS 208, Euro NCAP, and C-NCAP each define distinct impact configurations, dummy types, instrumentation, and injury assessment functions. Before any correlation target is set, the analyst must extract from the applicable protocol the exact boundary conditions and the exact pass/fail metric that the physical test will report.

What Each Protocol Actually Specifies

FMVSS 208 is a regulation, not a rating. Compliance is binary: the vehicle either meets the injury criteria limits or it does not. The core occupant protection requirements include a 56 km/h (35 mph) full frontal rigid barrier test at 0° with the Hybrid III 50th percentile male in the driver position, a 40 km/h (25 mph) 30° angled right-side rigid barrier test, and a 32 km/h (20 mph) perpendicular rigid pole test for side impact (the pole test is now largely superseded by FMVSS 214 for side, but 208 retains frontal and offset elements). Injury limits are explicit: HIC15 ≤ 700, chest deflection ≤ 63 mm, and femur axial force ≤ 10 kN for the 50th percentile male. Because these are hard limits, the CAE correlation target is typically the measured physical response itself — the model must reproduce the sled or vehicle pulse and the resulting dummy readings within an agreed tolerance.

Euro NCAP is a consumer rating program. It uses a frontal offset deformable barrier (ODB) test at 64 km/h with 40% overlap, a full-width rigid barrier (FWRB) test at 50 km/h, and a side mobile deformable barrier (MDB) test at 60 km/h with a 1,400 kg trolley. The dummy set includes the Hybrid III 50th male, the Hybrid III 5th percentile female, and the Q-series child dummies. Euro NCAP scores are continuous, not binary, and the frontal assessment combines ODB and FWRB results into a single score. Critically, Euro NCAP applies a "modifier" for the FWRB test based on the restraint system's ability to manage the occupant in a full-width engagement — this changes the correlation target because the model must match not just peak values but the shape of the force–displacement and the timing of restraint engagement.

C-NCAP (China New Car Assessment Program) has evolved through multiple versions. The 2021 version introduced a 50 km/h full frontal rigid barrier test with the Hybrid III 50th male, a 64 km/h ODB test with 40% overlap, and a 60 km/h side MDB test. C-NCAP also includes a 32 km/h side pole test and, in recent versions, a far-side occupant protection assessment. The scoring weights differ from Euro NCAP, and C-NCAP uses its own injury criteria thresholds — for example, a HIC15 limit of 650 for the 50th male in the full frontal test, which is stricter than FMVSS 208. This means a model correlated to FMVSS 208 limits may not automatically satisfy C-NCAP correlation targets.

How Protocol Differences Propagate into Correlation Targets

The most common correlation error is treating all frontal tests as equivalent. They are not. The ODB test at 64 km/h with 40% overlap produces a different vehicle deceleration pulse than the FWRB test at 50 km/h. The ODB pulse has a longer duration and lower peak because the deformable barrier absorbs energy; the FWRB pulse is shorter and sharper. If the CAE model is correlated only to the ODB pulse, the restraint system's timing — airbag deployment, seatbelt load limiter force, pretensioner stroke — may be tuned to a pulse that does not represent the FWRB condition. The result is a model that matches one test but fails the other.

Similarly, the dummy type matters. The Hybrid III 50th male has a different chest deflection response than the 5th percentile female, and the Q-series child dummies have different mass and geometry. A correlation target set for the 50th male cannot be directly applied to the 5th female without re-validating the chest and pelvis response. Euro NCAP's use of multiple dummies in the same test matrix means the CAE model must be correlated across a dummy family, not a single dummy.

Injury criteria also differ. FMVSS 208 uses HIC15, chest deflection, and femur force. Euro NCAP uses HIC15, chest deflection, and also considers the viscous criterion (VC) for chest injury in some configurations. C-NCAP uses HIC15, chest deflection, and femur force, but with different thresholds. The correlation target must specify which criterion is being matched and at what tolerance. A common practice is to set a target of ±10% on peak HIC15 and ±5 mm on peak chest deflection, but these tolerances must be justified by the test-to-test variability of the physical test itself, not chosen arbitrarily.

Practical Correlation Target Definition

When defining correlation targets for a specific program, the following inputs must be extracted from the protocol before any simulation is run:

• Impact configuration: barrier type, overlap percentage, impact speed, impact angle, and vehicle attitude (e.g., 0° or 30°).
• Dummy specification: percentile, certification status, and instrumentation channels (head accelerometers, chest potentiometer, femur load cells, etc.).
• Injury criteria and thresholds: which criteria are reported, and what the pass/fail or scoring thresholds are.
• Vehicle boundary conditions: sled pulse or vehicle deceleration pulse, seat position, seatbelt anchorage locations, and airbag deployment timing.
• Test-to-test variability: the standard deviation of the physical test results across repeated runs, which sets the minimum achievable correlation tolerance.

For example, in a Euro NCAP ODB correlation, the target might be defined as: peak HIC15 within ±8% of the mean of three physical tests, peak chest deflection within ±4 mm, and belt force time history within ±10% of the mean curve. The ±8% and ±4 mm are not arbitrary — they are derived from the observed repeatability of the test facility. If the facility's own repeatability is ±12%, then a ±8% target is unachievable and the correlation program will chase noise.

Protocol Influence on Model Fidelity Requirements

The protocol also dictates the required model fidelity. FMVSS 208 compliance testing is a single-point assessment — the model must be accurate at the specific speed and angle. Euro NCAP, with its scoring system, requires the model to be accurate across a range of speeds and overlap conditions because the score is a function of performance across the test matrix. C-NCAP's inclusion of far-side occupant protection requires the model to capture occupant kinematics in a side impact with no direct loading on the occupant's side — a condition that demands accurate seatbelt geometry and seat friction modeling, not just airbag deployment.

In practice, this means the correlation target for a Euro NCAP program is not a single number but a response surface. The model must reproduce the physical test at the ODB condition, the FWRB condition, and the side MDB condition, and the correlation is judged on the combined score, not on any single test. This is a higher bar than FMVSS 208 correlation, and it requires a more detailed model — finer mesh in the chest and pelvis, more accurate material models for the seat and belt, and a validated airbag deployment model.

From Protocol to Correlation Plan

The practical workflow is to build a correlation plan that maps each protocol requirement to a specific CAE output channel and a specific tolerance. For a program targeting both Euro NCAP and C-NCAP, the plan must account for the fact that the same vehicle may be tested under both protocols, and the correlation targets may differ. A model correlated to Euro NCAP ODB at 64 km/h may need re-validation for C-NCAP ODB at the same speed but with different barrier stiffness and dummy thresholds. The correlation plan should therefore list each test configuration separately, with its own target values and tolerances, rather than assuming that one validated model covers all protocols.

Finally, the protocol defines the reporting format. FMVSS 208 correlation is reported as a pass/fail against the regulatory limits. Euro NCAP and C-NCAP correlation is reported as a score or a percentage of the maximum score. The CAE correlation report should mirror this format so that the correlation status can be directly compared to the physical test result. If the physical test scores 85% and the CAE model scores 82%, the correlation gap is 3 percentage points — a meaningful metric that is not visible if the report only lists peak HIC15 values.