2. NCAP rating protocols and pretest relevance

A pretest NCAP readiness review only makes sense if the team agrees on which protocol version governs the rating, how the star score is actually computed, and which test configurations the vehicle will be subjected to. A vehicle that scores well under one program's frontal offset test can lose stars under another program's mobile progressive deformable barrier test, not because the structure is fundamentally worse, but because the loading path, barrier stiffness, and occupant criterion differ. This chapter establishes the protocol landscape, the mechanics of star scoring, and the configuration dependencies that must be locked before any pretest evidence is judged credible.

Key global programs and what they actually measure

Four programs dominate the global rating landscape and each has a distinct philosophy. Euro NCAP uses a weighted four-pillar structure — Adult Occupant Protection (AOP), Child Occupant Protection (COP), Vulnerable Road User (VRU) protection, and Safety Assist — with each pillar scored as a percentage and then combined. US NCAP (NHTSA) reports separate star ratings for frontal crash, side crash, and rollover resistance, using instrumented crash tests and the Static Stability Factor for rollover. IIHS does not issue a single star score; it awards Good/Acceptable/Marginal/Poor per test and grants TOP SAFETY PICK / TOP SAFETY PICK+ based on a threshold set of tests, including the driver-side and passenger-side small overlap frontal tests. C-NCAP and other regional programs (Latin NCAP, ASEAN NCAP, ANCAP, JNCAP, KNCAP, Bharat NCAP) follow similar pillar logic but with region-specific weightings, speed offsets, and dummies.

The practical consequence for pretest planning is that "five stars" is not a portable claim. A pretest matrix must be built per target market, and the same physical vehicle may need different restraint tuning, different bumper beam geometry, and different pedestrian protection countermeasures to satisfy two programs simultaneously. Where a single global platform is intended, the pretest review should identify the binding constraint — the program and test whose margin is thinnest — and treat it as the gating item.

Star rating mechanics: from raw signals to a published score

Star ratings are not derived from a single number. The general pipeline is: instrumented test → channel filtering and normalization → injury criterion computation → test-level score → pillar percentage → weighted overall score → star band. Each stage introduces dependencies that a pretest review must trace.

Injury criteria are computed from filtered accelerometer and load-cell channels. For the frontal 50th-percentile male driver, the head injury criterion HIC₁₅ is computed over the 15 ms window that maximizes the integral of resultant head acceleration; the chest deflection is read from the chest potentiometer; the femur loads from the load cells in the femur. Each criterion is mapped to a score via a piecewise-linear or exponential transfer function defined in the protocol. For example, a simplified AOP head score might award full points below HIC = 650, zero points above HIC = 1000, and interpolate linearly between. The exact breakpoints are protocol-specific and version-specific, which is why the pretest review must cite the protocol revision, not just the program name.

Pillar percentages are then combined with weights. Euro NCAP's overall score is a weighted sum of the four pillar percentages, and the star band is assigned from the overall percentage with a minimum-pillar constraint: a vehicle cannot achieve five stars if any single pillar falls below a floor. This constraint is frequently the source of late-stage rating risk, because a strong AOP result can be nullified by a weak VRU or Safety Assist pillar. The pretest review should therefore track pillar-level margins, not just the aggregate.

Test configuration dependencies

The same vehicle model produces different results depending on configuration. The dependencies that most often change a star outcome are:

• Seating position and dummy size: driver vs. front passenger, 5th-percentile female vs. 50th-percentile male, and rear-occupant configurations each have their own criteria and score allocations.
• Barrier and impactor specification: full-width rigid barrier, offset deformable barrier (ODB, 40% overlap), mobile progressive deformable barrier (MPDB), side barrier (AE-MDB), side pole, and pedestrian leg/head impactors each load the structure differently.
• Impact speed and angle: a 64 km/h ODB test and a 50 km/h MPDB test are not interchangeable; the MPDB adds a compatibility assessment (ODB-based) that penalizes overly stiff front structures.
• Restraint and airbag state: pretensioner firing, airbag venting, and seat-belt load limiter settings must match the production calibration; a pretest built on a development calibration is not evidence of production rating.
• Vehicle mass and ride height: optional equipment, battery pack variants, and suspension variants change the center of gravity and the barrier engagement height.

These dependencies mean the pretest matrix is combinatorial. A disciplined approach is to fix the highest-risk configuration first — typically the heaviest variant with the smallest dummy in the most demanding barrier test — and to record the exact configuration identifier (variant code, calibration part number, dummy serial number, barrier batch) alongside every result. Without that traceability, a pretest result cannot be defended in a rating review.

Worked example: tracing a pretest result to a rating risk

Consider a mid-size battery-electric SUV targeting Euro NCAP five stars. The pretest team runs a 50 km/h MPDB test with the 50th-percentile male driver and records HIC₁₅ = 720, chest deflection = 28 mm, and a femur load of 3.1 kN. Under the protocol revision in force, the head score maps to approximately 0.78 of full marks, the chest to 0.85, and the femur to 0.95. The AOP pillar for this test configuration is therefore around 0.84, which is above the five-star floor but with only a modest margin. The pretest review should flag this as a watch item, not a pass, because the same vehicle in the 64 km/h ODB configuration — a different barrier and a different speed — may produce a different chest deflection and a different pillar percentage. The correct pretest action is to run the ODB configuration with the same calibration and compare pillar-level margins, then decide whether restraint tuning is required before the official test.

The example also illustrates why the configuration identifier matters. If the MPDB test used a development airbag vent and the ODB test uses the production vent, the two results are not comparable, and the pretest review cannot conclude anything about the production rating. The traceability fields in the dependency map above — variant code, calibration part number, dummy serial number, barrier batch — are the minimum evidence set for a defensible pretest conclusion.

What the pretest review must lock before testing

Before any physical test is scheduled, the review should confirm four things. First, the protocol revision for each target market, including the injury-criterion breakpoints and pillar weights, because these change between revisions and invalidate prior evidence. Second, the binding configuration — the variant, dummy, barrier, and speed combination with the thinnest expected margin — and the rationale for selecting it. Third, the calibration state of every restraint component, confirmed against the production part number. Fourth, the traceability record that will accompany each result, so that a later rating review can reconstruct exactly what was tested.

With these locked, the star rating mechanics become a predictable calculation rather than a post-hoc surprise, and the pretest evidence can be used to populate the rating risk register with quantified margins rather than qualitative concerns.