The standard PFMEA session is an exercise in structured consensus. Engineering, quality, and production representatives gather to debate severity, occurrence, and detection ratings for potential failure modes. A severity rating of 8 is negotiated down to 6. The occurrence rating is settled by a show of hands. The detection rating is assigned based on the assumption that existing controls will probably catch the defect.
The resulting spreadsheet is uploaded to the quality management system and filed until an auditor requests it. Six months later, the exact failure mode everyone rated as low risk appears on the customer's dock in the form of a warranty claim. The PFMEA had it listed, rated, and ranked. But nobody ever tested whether the assumptions behind those ratings were actually true.
This is the operational gap between identifying a risk and verifying it. Failure Mode Verification Testing (FMVT) is the discipline of converting subjective FMEA beliefs into objective engineering data. It takes hypothetical failure modes and subjects them to controlled, deliberate testing to understand exactly how they develop, how fast they propagate, and whether your detection systems will actually catch them in time.
The Unrecognized Flaw in FMEA Methodology
I have audited plants where the FMEA functioned as a collective narrative rather than an engineering document. It is a story the team constructs about what might go wrong, heavily weighted by individual experience, biased by recent memory, and moderated by group dynamics. The document captures knowledge and forces cross-functional conversation, but it remains fundamentally untested.
When an engineering lead rates the occurrence of a weld crack at 2 out of 10, they are stating a belief that the failure is unlikely based on past experience. When a quality manager rates detection at 4, she is stating a belief that current controls will intercept the defect. These are estimates, not measurements. FMEA ratings are bets placed on theoretical failure modes. FMVT is the mechanism for cashing in those bets before reality does it for you.
FMVT targets the specific conditions under which a failure would happen. It does not ask whether a part meets drawing specification. It asks what happens when the part fails, how quickly the degradation occurs, and whether your statistical process control or end-of-line testing will register the shift before the product leaves the facility. It applies scientific rigour to theoretical risk.
Selecting Failure Modes That Demand Verification
Not every row in a PFMEA warrants physical verification testing. Resources must target the failure modes where group consensus carries the highest mathematical risk of being wrong. Attempting to test every theoretical failure stretches laboratory resources and delays production timing. You must prioritise the specific scenarios where unverified assumptions carry severe downstream consequences.
The primary targets are failure modes with high Risk Priority Numbers (RPN) but low confidence in the underlying ratings. If the cross-functional team debated a severity rating for twenty minutes and settled on a number through compromise rather than data, that failure mode is a prime candidate. The rating reflects human negotiation, not engineering reality.
New designs, new processes, and recent engineering changes also require verification. Any material substitution, supplier change, or process adjustment introduces failure modes that will not behave like the legacy data your team relies upon. If you have never manufactured this exact product configuration before, every occurrence rating is a guess. Field history does not apply.
Finally, target low-occurrence, high-severity scenarios. These are the deceptive failure modes that an FMEA dismisses with a confident occurrence rating of 2, despite carrying a severity of 9 or 10. If the detection system has never been challenged against the actual failure mode it is supposed to catch, you do not have a detection system. You have an unproven theoretical control.
Designing the Verification Protocol
FMVT is not arbitrary destruction testing. It is a structured validation process that connects directly to your PFMEA and control plan. For each selected failure mode, the test protocol must answer three distinct engineering questions: what triggers the failure, how the failure propagates, and whether current detection methods will intercept it.
The FMVT Verification Sequence
- 01Trigger ReplicationDefine the specific stress, cyclic loading, or environmental condition that initiates the failure mode.
- 02Propagation MonitoringInstrument the test to capture the timeline of degradation, not just the final breakpoint.
- 03Detection ChallengeRun existing SPC and end-of-line checks against degraded samples to test their effectiveness.
- 04System FeedbackFeed verified failure behaviour back into the PFMEA, control plan, and design criteria.
First, identify the exact trigger. Replicate the conditions that would initiate the failure. Thermal stress is not a test. Twenty cycles between -40°C and +125°C with dwell times of thirty minutes at each extreme is a test. Specify the parameters with the same precision you would apply to a Design Verification Plan and Report (DVP&R).
Second, monitor how the failure propagates. Do not simply run the test until something breaks. Instrument the samples to capture the progression of the failure mode. Track crack initiation, measure performance degradation, and record dimensional shift. You need to understand the timeline of the failure to know how much warning your production process will actually receive.
Third, and most critically, challenge your detection systems. Run your existing controls against the degrading test samples. If the failure mode produces a measurable signal before catastrophic breakdown, verify whether your current SPC sampling frequency or end-of-line pass/fail thresholds will actually flag the deviation. This step exposes the operational gap between theoretical detection and actual detection.
Exposing the Detection Gap in Practice
The value of FMVT is best illustrated by how quickly it invalidates a comfortable FMEA rating. Consider an automotive supplier developing an electric motor assembly. During the PFMEA, the team identified potential bearing preload loss due to thermal cycling. Severity was rated at 7, occurrence at 2, and detection at 3 based on the assumption that end-of-line testing would catch excessive vibration. The RPN was 42.
A subsequent FMVT protocol subjected twelve motor assemblies to thermal cycling that simulated five years of seasonal temperature swings. Vibration sensors monitored continuous degradation, while end-of-line test equipment ran at regular intervals to simulate actual production inspection.
If your detection system has never been challenged against the actual failure mode, you have a detection hope, not a detection system.
Four of the twelve motors showed measurable preload loss after the equivalent of eighteen months. The critical finding, however, was that the vibration signature did not change enough to trigger a failure flag in the standard end-of-line test. The preload loss was gradual enough to fall within the pass thresholds of the existing detection method. The FMEA detection rating of 3 was mathematically wrong; verified testing proved it should have been a 7.
The recalculated RPN jumped to 98. The engineering team redesigned the bearing retention system, added a specific supplemental measurement to the end-of-line test, and implemented an accelerated thermal cycling screen in production. The failure mode never reached a customer because FMVT proved the FMEA ratings wrong before the launch.
Integrating FMVT Into APQP and Cost Management
Verification testing cannot remain a sporadic reaction to a field warranty claim. It must be integrated into your Advanced Product Quality Planning (APQP) timeline. Make it a formal gate between PFMEA completion and control plan finalization. Identify the top failure modes that warrant physical verification and build the test plan directly into the project schedule.
The Cost of Unverified Risk
Project budgets must include dedicated resources for this testing. Test fixtures, instrumented samples, laboratory time, and engineering hours should be planned during the initial cost quoting phase. When teams argue that verification testing is too expensive, direct them to the cost structure of warranty campaigns, customer plant shutdowns, and 8D containment. A thermal cycling chamber operated for one week costs a fraction of a single OEM line stoppage.
Furthermore, verify your existing DVP&R scope. DVP&R confirms that a design meets engineering requirements under specified conditions. FMVT confirms that your failure mode assumptions are correct. DVP&R proves the part functions. FMVT proves you understand exactly how and when the part will fail, and whether you will catch it. They are complementary, not redundant, processes.
Building an Organisational Knowledge Base
Every FMVT protocol generates proprietary engineering data. Organisations that capture this data build a searchable library of verified failure modes, propagation rates, and detection signatures. This library becomes your most valuable resource for informing future PFMEA iterations with real data instead of subjective consensus.
Close the feedback loop by connecting this library directly to field returns. When warranty claims occur, compare the physical failure to the original FMEA predictions and the FMVT results. If the failure mode was predicted and verified, the verification data will confirm whether the production controls were followed. If the failure mode was entirely unverified, you have identified a critical gap in your FMEA methodology.
This continuous feedback mechanism shifts the quality system from reactive firefighting to proactive risk management. Train your engineering teams to treat FMEA ratings as hypotheses that require physical testing. When a development team understands that their severity, occurrence, and detection ratings will be physically challenged in the laboratory, the rigour of their initial FMEA analysis improves immediately.
Assumptions are operational liabilities until they are converted into verified knowledge. Your FMEA identifies what might fail. Your control plan dictates how you intend to monitor it. Failure Mode Verification Testing provides the empirical data that bridges the gap between the two, ensuring your defensive controls are calibrated to actual physical behaviour rather than conference-room negotiations.