3. Nij Normalization Constants Selection and Sensitivity

Why the Normalization Constants Matter

The Nij metric is a ratio, and like every ratio it is only as meaningful as its denominator. In the standard formulation, the numerator is the instantaneous combination of axial neck force and bending moment, while the denominator is a pair of critical values — a critical force F_crit and a critical moment M_crit — that represent the load level at which a reference population is expected to reach a defined injury risk. Because the same physical load history can produce very different Nij values depending on which critical values are substituted, constant selection is not a cosmetic bookkeeping step; it directly determines whether a test is judged a pass or a fail.

The general form used throughout this course is:

Nij(t) = Fz(t) / F_crit + My(t) / M_crit

where Fz is the upper-neck axial force (tension positive) and My is the sagittal bending moment (flexion positive in the convention adopted here). The four load-direction combinations — tension-extension (NTE), tension-flexion (NTF), compression-extension (NCE) and compression-flexion (NCF) — each carry their own F_crit and M_crit, because the neck tolerates tension and compression, and flexion and extension, very differently. Selecting constants therefore means selecting the correct quadrant as well as the correct magnitude.

Choosing Constants for a Given Occupant and Direction

Critical values are population- and direction-specific. A practical selection workflow for a 50th-percentile adult male in a frontal impact looks like this:

1. Fix the occupant reference. Identify the anthropometry the constants were derived for (e.g., 50th-percentile male, Hybrid III midsize). Using 5th-percentile female constants against a 95th-percentile male dummy is a common and serious mismatch.
2. Determine the governing quadrant from the load history. Inspect the sign of Fz and My at the candidate peak. A frontal sled pulse typically drives the head into flexion with the neck in tension or compression depending on kinematics; the quadrant must be read from the data, not assumed.
3. Pull the matching constant pair. For the NCF quadrant, a representative teaching set is F_crit = 4.0 kN (compression) and M_crit = 190 N·m (flexion). For NTE, a representative set is F_crit = 3.3 kN (tension) and M_crit = 57 N·m (extension). These are illustrative values for the exercise, not a substitute for the governing regulation or the source you are working to.
4. Record the provenance. Note the standard, dummy, and direction each constant came from so the result is auditable.

Worked Selection Example

Suppose a frontal pulse yields a candidate peak at t = 78 ms with Fz = -3.1 kN (compression) and My = +142 N·m (flexion). The signs place this in the NCF quadrant, so the compression and flexion constants apply:

Fz = -3.1 kN  -> |Fz| = 3.1 kN, F_crit = 4.0 kN
My = +142 N·m -> My = 142 N·m, M_crit = 190 N·m

Nij = 3.1 / 4.0 + 142 / 190
    = 0.775 + 0.747
    = 1.522

The result exceeds 1.0, so this load case would be flagged as injurious under the selected constants. The next question is how robust that conclusion is to the constant choice itself.

Sensitivity Sweep

A sensitivity sweep perturbs F_crit and M_crit around their nominal values and re-evaluates the peak Nij, revealing how much the diagnosis depends on the constants. A one-at-a-time sweep of ±10% and ±20% on each constant, holding the other fixed, is a compact and interpretable design. The table below shows the expected Nij for the example peak above.

Two observations follow. First, the Nij value stays above 1.0 across the entire ±20% band, so the injurious classification is robust to constant choice for this peak. Second, the force term and moment term contribute almost equally (0.775 vs 0.747), so neither constant dominates — a balanced sensitivity, unlike cases where one term is near zero and the other carries the ratio.

Reading the Sweep and Acting on It

The near-overlapping curves in the figure are themselves the finding: for this peak, force and moment constants have almost identical leverage on the result, and the whole ±20% band sits comfortably above the 1.0 threshold. That is a robust fail. The opposite pattern — a nominal Nij of, say, 1.03 that drops below 1.0 under a +10% constant perturbation — is a marginal result and should be reported with the sensitivity band attached, not as a bare pass/fail.

When a sweep shows high sensitivity, check three things before trusting the number: whether the correct quadrant constants were used, whether the occupant reference matches the dummy, and whether the peak was taken from a filtered or unfiltered signal (filtering is the subject of a later case and can move the peak materially). Document the nominal constants, their source, the perturbation range, and the resulting Nij band alongside the single peak value so any reviewer can reproduce the diagnosis.