3. Severity Ranking for Airbag Deployment Failure Modes Using AIAG-VDA Criteria
Why Severity Is Rated Before Occurrence and Detection
In an AIAG-VDA DFMEA, severity (S) describes the seriousness of the effect of a failure mode on the end user, the vehicle, or regulatory compliance. It is deliberately independent of how often the failure occurs or how easily it is detected. This ordering matters for airbag work: an inadvertent deployment in a parked vehicle and a failure to deploy in a 50 km/h frontal crash can both be catastrophic, yet their occurrence rates and detection capabilities differ enormously. If you let occurrence or detection leak into the severity discussion, you will systematically under-rate low-frequency but lethal failure modes — exactly the ones that dominate airbag safety cases.
The AIAG-VDA severity table runs from S=10 (affects safe operation, potentially without warning, or violates a regulation) down to S=1 (no discernible effect). For occupant-protection systems, the practical working range is narrow: most credible deployment failure modes land between S=7 and S=10. The task in this case is to assign those ratings defensibly, with a written justification tied to occupant injury risk and to the regulatory framework (FMVSS 208 in the US, UN R94/R95 in Europe), rather than by intuition.
Reference Framework: Effect, Criterion, and Justification
Each severity rating needs three linked statements: the failure effect at the vehicle/occupant level, the AIAG-VDA criterion that the effect matches, and the justification that anchors the rating in injury risk or regulation. The table below is the working reference for this case; it is a condensed extract of the AIAG-VDA severity scale as applied to restraint systems, not a replacement for the full standard.
| S | AIAG-VDA effect criterion (condensed) | Airbag-specific interpretation |
|---|---|---|
| 10 | Affects safe operation of the vehicle and/or involves non-compliance with government regulation, without warning | Airbag fails to deploy in a required crash mode; inadvertent deployment at speed; non-compliance with FMVSS 208 / UN R94 occupant protection requirement |
| 9 | Affects safe operation and/or involves non-compliance, with warning | Deployment occurs but outside the required performance window with a telltale present; degraded but regulation still nominally met |
| 8 | Loss of primary function | Airbag does not deploy in a mode where deployment is the primary restraint function, but no explicit regulatory breach is demonstrated |
| 7 | Degradation of primary function | Late, weak, or partial deployment that still provides some restraint energy |
| 6 | Loss of secondary function | Loss of a comfort or convenience function (e.g., passenger airbag disable indicator) without restraint impact |
| 5 | Degradation of secondary function | Reduced performance of a non-restraint function |
| 4 | Annoyance (appearance, noise, vibration, harshness) | Audible or tactile annoyance with no functional loss |
| 3 | Annoyance (perceived, non-functional) | Cosmetic or perceived quality issue |
| 2–1 | No discernible effect | Not applicable to deployment functions |
Applying the Scale to Deployment Failure Modes
The following worked assignments use a driver frontal airbag module as the analysis object. Each row states the failure mode, the effect at the occupant level, the assigned severity, and the justification. Treat the ratings as the output of a documented engineering judgment, not as measured data.
| Failure mode | Effect on occupant / vehicle | S | Justification |
|---|---|---|---|
| Inadvertent deployment while driving | Sudden inflator firing at highway speed; loss of steering control, driver injury from module cover, potential secondary crash | 10 | Affects safe operation without warning; violates the requirement that deployment occur only in a qualifying crash. Injury risk is immediate and unbounded by crash severity. |
| Failure to deploy in a 50 km/h frontal barrier impact | Occupant loads the steering wheel and column unrestrained; head/chest injury criteria exceed regulated limits | 10 | Direct non-compliance with FMVSS 208 / UN R94 occupant protection performance; loss of the primary restraint function in a required test mode. |
| Late deployment (firing after the required time window) | Airbag reaches the occupant out of position; increased head/neck injury, possible "punch-out" injury | 9 | Primary function degraded and regulation at risk; the effect is serious but the airbag still provides some restraint, so it sits below the no-deployment case. |
| Partial deployment (insufficient inflator output) | Reduced restraint energy; occupant excursion increases, injury criteria marginally exceeded | 8 | Loss of primary function performance without a demonstrated regulatory breach in all modes; still a serious occupant-protection shortfall. |
| Deployment with excessive inflator output (over-aggressive) | Higher occupant loading; increased risk of out-of-position injury, especially for small occupants | 8 | Primary function degraded in the direction of over-restraint; injury risk elevated but restraint is present. |
| Passenger airbag disable indicator fails to illuminate | Occupant cannot confirm airbag status; no direct restraint loss | 6 | Loss of a secondary (informational) function; no restraint performance impact. |
Reading the Severity Map
The SVG below places the same failure modes on the AIAG-VDA severity axis, grouped by the effect class that drives the rating. It is intended as a quick visual check that ratings are internally consistent: modes with the same effect class should not be scattered across the scale.
Consistency Checks Before You Commit the Ratings
Severity ratings are only useful if they survive review. Run these checks on the table above before it enters the DFMEA:
- Regulatory anchor. Any mode that can cause non-compliance with FMVSS 208 or UN R94 must be S=9 or S=10. If a no-deployment mode is rated below 9, the justification is incomplete.
- Warning clause. The S=10 criterion includes "without warning." If a telltale or driver warning is present and functional, the same effect may drop to S=9. Document whether the warning is part of the design or an assumption.
- Effect-class clustering. Modes with the same effect class should sit in the same band. A partial-deployment mode rated S=10 while a no-deployment mode is rated S=8 signals an inconsistency.
- No occurrence leakage. If the justification contains "rarely happens" or "unlikely," remove it — that belongs in the occurrence rating, not severity.
- Occupant population. State whether the effect is assessed for a 50th-percentile male, a small female, or an out-of-position occupant. The same failure mode can carry different injury consequences across populations, and the rating should name the population it assumes.
Deliverable: Severity Justification Record
The following record is a fillable template for each failure mode. It forces the three-part justification and the consistency checks into one auditable row. Populate one row per failure mode; the example row uses the inadvertent-deployment case.
| Field | Content (example) |
|---|---|
| Failure mode ID | FM-DAB-01 |
| Failure mode | Inadvertent deployment while driving |
| Effect (occupant/vehicle) | Sudden inflator firing at speed; loss of steering control; driver injury from module cover; secondary crash risk |
| AIAG-VDA criterion matched | S=10 — affects safe operation without warning |
| Assumed occupant population | 50th-percentile male driver, upright seated |
| Regulatory reference | FMVSS 208 (deployment only in qualifying crash); UN R94 |
| Warning present? | No — deployment is not preceded by any driver warning |
| Assigned severity | 10 |
| Justification | Effect matches the S=10 criterion directly: safe operation affected without warning and regulatory intent breached. Injury risk is immediate and independent of crash severity. |
| Consistency check | Clustered with FM-DAB-02 (no deployment) at S=10; no occurrence language present; population stated. |
With the severity column populated and justified, the next step in the DFMEA is to rate occurrence and detection for the same failure modes — but those ratings must not feed back into the severity values recorded here.