Quality & InspectionFree referencePDF, Excel, Word export

FMEA Prevention and Detection Controls: A Reference List with Examples

The current controls columns are where FMEA ratings often become wishful. This list separates prevention controls, which reduce the chance the cause occurs, from detection controls, which find the cause or failure mode before it reaches the customer, and states the weakness of each.

Download:
Control typeControlWhat it addressesExampleLimitation to rate honestly
Prevention
Poka-yoke, physical keying
Wrong part, wrong orientation
Asymmetric fixture pin so the part only fits one way
Can be defeated if operators bypass or force the part
Prevention
Poka-yoke, sensor interlock
Missing component or step
Presence sensor blocks cycle start if a clip is absent
Sensor drift or bypass keys reduce the effect
Prevention
Preventive maintenance schedule
Wear-related cause
Scheduled stencil or nozzle replacement by cycle count
Interval must be based on data, not habit
Prevention
Tool life management
Dimension drift from tool wear
CNC tool counter forces change after set parts
Counter must be reset and actually enforced
Prevention
Error-proof recipe download
Wrong program or parameters
Parameters loaded from controlled system, not typed
Depends on revision control of the source
Prevention
Statistical process control with reaction plan
Process drift before nonconformance
X-bar and R chart with stop rules at the machine
Only prevents when someone reacts to signals
Prevention
Torque tool with count and angle control
Under or over torque
Controller counts fasteners and locks until all are done
Needs calibration and joint-specific limits
Prevention
Kitting and line-side bin control
Mixed or wrong parts
Kitted components per work order with barcode check
Kit errors move upstream rather than disappear
Prevention
Controlled work instructions with visual aids
Operator method variation
Photo-based instruction at the station
Weak alone; relies on operator discipline
Prevention
Operator training and certification
Skill-related cause
Skills matrix with sign-off before solo work
Rate as weak prevention for high-severity items
Prevention
Supplier certification and material approval
Incoming material variation
Approved supplier list with certificate of conformance
Certificate accuracy is trust, not verification
Prevention
Environmental control
Contamination, humidity, ESD
Cleanroom class, humidity logging, ESD flooring
Needs monitoring and alarm response
Prevention
Design margin or design rule
Design cause in DFMEA
Derating rule, minimum wall thickness rule
Rules must be applied and reviewed
Prevention
First-piece approval after setup
Setup error
Setter and inspector sign-off before release to run
Detects setup error but not later drift
Detection
In-process automated gauging
Dimensional failure mode
In-machine probe measures bore after cut
Probe calibration and measurement uncertainty
Detection
Automated optical inspection
Missing, misplaced, bridged parts
AOI after reflow
False calls and programmed coverage limits
Detection
Vision system with golden image
Presence, orientation, label
Camera verifies label text and position
Lighting and part variation affect reliability
Detection
End-of-line functional test
Function failure
Powered test against specification
Only tests what the test program covers
Detection
Leak test
Seal or weld leak
Pressure decay or helium test
Sensitivity set by test parameters
Detection
Continuity and hipot test
Wiring errors, insulation faults
Cable harness tester after assembly
Does not find latent defects such as weak crimps
Detection
Go and no-go gauge
Hole or thread size
Thread plug gauge at the station
Operator dependent; wear of gauge itself
Detection
CMM inspection per sample plan
Critical dimensions
Periodic CMM check of key features
Sampling can miss intermittent failures
Detection
Visual inspection by operator
Cosmetic and gross defects
Final visual check with defect samples
Rate detection poorly for rare or subtle defects
Detection
Torque audit by sampling
Fastener torque
Audit wrench check on a sample of assemblies
Sampling only; detects after the fact
Detection
Barcode scan verification
Wrong part or label
Scan component and work order match before use
Detects mismatch only if label is correct
Detection
Dimensional or material certification review
Incoming material
Receiving inspection against certificate and sample test
Sample size and test method limit coverage
Detection
Burn-in or environmental stress screening
Infant mortality
Temperature cycling before shipment
Cost and time; screens only certain failure types
Detection
Customer or field feedback loop
Escapes
Warranty data reviewed monthly
Late detection; not a manufacturing control

In the AIAG-VDA method, prevention controls support the occurrence rating and detection controls support the detection rating. Rate a control on its demonstrated effectiveness, not on its mere existence.

Link inspection plans to your FMEA

Optional: a short review of how routings, inspection plans and NCR data in your ERP could feed FMEA controls.

Optional. The tool and every export work without this. No spam.

Prevention versus detection

A prevention control acts on the cause: a sensor interlock, a keyed fixture or a tool-life counter lowers the likelihood that the failure mode occurs at all, so it supports the occurrence rating. A detection control finds the cause or failure mode after it exists, such as an end-of-line test or an AOI station, so it supports the detection rating. The same activity can be one or the other depending on where it sits. An SPC chart with an enforced reaction plan prevents; a chart nobody reads detects nothing.

How to use the list

Search for the failure mode or the cause you are analysing, then pick controls that really exist on your line. Write the control in your FMEA with its specifics: what is checked, with what gauge, at what frequency and who reacts. Avoid listing training as a strong control for a high-severity item. Then move to the limitation column and ask whether that weakness applies here, and whether it should push the rating up.

  • List only controls that already exist when you rate; planned ones belong in recommended actions
  • Name the gauge, frequency and reaction for every detection control
  • Match each control to a cause, not just to the failure mode
  • Carry the same controls through to the control plan

Common mistakes

The most common error is giving full credit to a control that is bypassed in practice, or counting a sampling inspection as if it screened every part. Another is rating detection low because a test exists, when the test cannot see the specific failure mode. Check each detection control against the failure mode: can it actually distinguish good from bad for this characteristic, at this stage, before the part leaves your control? Where the honest answer is not yet, rate it that way and add an action.

Frequently Asked Questions

Does a detection control reduce occurrence?

No. A detection control finds a problem that has already occurred, so it improves the detection rating only. Only a prevention control that acts on the cause should be used to justify a lower occurrence rating.

Is operator training a valid control?

It counts, but it is generally a weak prevention control because it depends on human consistency. For high-severity effects, teams should look for error-proofing or automatic checks rather than relying on training alone.

Should the same control appear in the control plan?

Yes. The controls you rely on in the FMEA should appear in the control plan with specification, measurement method, sample size, frequency and reaction plan. A control that is in the FMEA but not in the plan is not really controlled.

Is a sampling inspection a good detection control?

Only for failure modes that are consistent across a lot. For random or intermittent failures a sample can miss them entirely, so detection should be rated with that gap in mind.

If you want control plan methods, gauges and reaction plans traced from your ERP routings and inspection plans, we can discuss how that data could be read in place.