FMEA Basics & Risk Priority Number

FMEA Basics & Risk Priority Number (CSSGB BoK I.C.2 — Analyze)

Quick Answer: Failure Mode and Effects Analysis (FMEA) is a structured method to identify how a product or process can fail, what the effects and causes are, and how well current controls detect or prevent those failures. Teams rate Severity (S), Occurrence (O), and Detection (D)—typically 1–10—and calculate RPN = S × O × D to prioritize risk-reduction actions.

FMEA is a core risk tool for both Design for Six Sigma and process improvement. At the CSSGB Analyze cognitive level, you should score risk factors, compute RPN, interpret priorities, and recommend where actions belong—not merely define the acronym.

What FMEA Does

An FMEA walks a team through potential failure modes—the ways a function can go wrong—then documents:

1. Effects of each failure on the customer, downstream process, or business.
2. Causes or mechanisms that could produce the failure.
3. Current controls that prevent the cause or detect the failure before it escapes.
4. Risk ratings that drive prioritized actions.
5. Recommended actions, owners, and revised risk after improvement.

FMEA is proactive. The ideal time to run it is during design or process planning, before failures become scrap, recalls, or patient harm. It also supports change control when a design or process changes. It is not a substitute for data-driven root cause analysis after a defect already occurs, but it often informs that work and the control plan.

Core Columns (Mental Model)

Teams tailor worksheets by industry, but Green Belts should know the standard logic:

Severity, Occurrence, and Detection Scales

Most organizations use 1–10 ordinal scales with written criteria. Exact tables vary (automotive AIAG, healthcare, internal standards), but the ranking logic is consistent:

Severity (S) rates the seriousness of the effect, assuming the failure occurs.

• Low scores: minor nuisance, no real impact on function or safety.
• Mid scores: degraded performance, customer dissatisfaction, rework, delays.
• High scores: loss of primary function, regulatory noncompliance, safety or severe harm.

Severity often does not drop unless the design or process effect itself changes (for example, adding a fail-safe so the worst effect is no longer possible). Training alone rarely lowers S.

Occurrence (O) rates how frequently the cause is expected under current design and process conditions.

• Low: rare or almost prevented by design.
• Mid: occasional; process sometimes drifts.
• High: frequent; cause is poorly controlled.

Actions that strengthen process capability, poka-yoke, supplier quality, or design robustness typically reduce O.

Detection (D) rates the chance that current controls miss the failure or cause before it escapes to the customer or next process. Important convention: higher D means worse detection (harder to find the problem).

• Low D: almost certain detection (strong automated check, robust error-proofing).
• High D: failure is unlikely to be found until the customer sees it.

Better inspection, earlier tests, sensors, and mistake-proofing that force detection reduce D. Prevention is still preferred over detection when practical.

Risk Priority Number (RPN)

$\text{RPN} = S \times O \times D$

RPN ranks relative risk so teams attack the worst combinations first. With 1–10 scales, RPN ranges from 1 to 1,000. There is no universal “magic threshold” (for example, 100) that always forces action—organizations set rules—but exam logic prioritizes higher RPN items and often high severity even when RPN is moderate (a rare catastrophic failure still needs attention).

Worked numerical example

A pharmacy packaging line has a potential failure mode: incorrect medication label applied.

$\text{RPN} = 9 \times 4 \times 6 = 216$

A second mode, label slightly crooked but readable, might score S = 2, O = 5, D = 3 → RPN = 30. The team prioritizes the wrong-label mode first.

Actions that cut risk for the high-RPN mode might include barcode verification before release (lowers D and can lower effective O of escape), poka-yoke fixtures that allow only the matching label set (lowers O), and dual verification for high-risk drugs. Suppose after barcode interlock and process changes the team rescores O = 2 and D = 2 (S still 9 because wrong label is still severe if it happens):

$\text{New RPN} = 9 \times 2 \times 2 = 36$

Residual risk is much lower; remaining controls enter the control plan and training matrix.

How Green Belts Use RPN in Practice

1. Build the cross-functional team (operators, engineers, quality, customers of the process).
2. Map the process or design functions so failure modes are not missed.
3. Score consistently using the organization’s scale—not gut feelings that change mid-meeting.
4. Sort by RPN and review high-S items even if RPN is not the absolute maximum.
5. Assign actions that attack causes (O), improve detection (D), or redesign effects (S).
6. Recalculate RPN after actions and track open items to closure.

Limitations to remember for the exam: RPN is a priority aid, not a precise probability model. Different scale definitions make RPN comparisons across companies unreliable. Teams should not game scores to avoid action. Still, for CSSGB analysis questions, compute RPN correctly and select the failure mode that most needs attention.

Tie-In to DFSS and DMAIC

In DFSS, FMEA (especially Design FMEA) is used while concepts and detailed designs form, before production scale-up. In DMAIC, Process FMEA often appears in Analyze or Improve to harden solutions and feed Control plans. Either way, the math and prioritization logic—S, O, D, and RPN—stay the same.