A design FMEA is the cheapest way to find failure modes before they reach customers, if it is done as engineering, not paperwork. How to run one well, rate risk sensibly and turn it into design changes and tests.
A Design FMEA for an outdoor electronics enclosure was completed on schedule, signed off and filed. It listed 140 failure modes. Six months after launch, the field failure that cost the company most, water ingress through a cable gland when installers over-tightened it, was not on the list. Nobody in the FMEA session had installed a product in the field, and nobody had asked an installer.
A DFMEA is one of the most powerful tools in product development. It is also one of the most frequently reduced to a form-filling exercise.
What a DFMEA is for
A Design Failure Mode and Effects Analysis systematically asks, for each function of a design: how could it fail, what would the effect be, what could cause it, and what are we doing to prevent or detect it before release? Its purpose is to change the design and the test plan while change is still cheap, not to document risk after the fact.
Running a session that finds real failures
- Start from functions, not parts. "Seal against rain at IP65" is a function; "gasket" is a part. Failure modes come from functions not being met: no seal, partial seal, intermittent seal, seal degrades over time.
- Bring the right people: design, manufacturing, quality, service and, where possible, someone who has seen the product, or its predecessor, fail in the field.
- Use real data: warranty returns, field complaints, lessons learned and supplier quality history.
- Think through the life of the product: transport, installation, use, cleaning, maintenance and misuse.
- Keep sessions short and focused, a few hours at a time, on one subsystem.
Rating risk sensibly
Traditional FMEAs rate severity, occurrence and detection from 1 to 10 and multiply them into a Risk Priority Number. RPN is easy to calculate and easy to misuse. A high-severity failure with a low occurrence rating can score lower than a nuisance that happens often. Most modern approaches, including the AIAG-VDA method, prioritise by severity first, then occurrence, then detection, using action priority tables rather than a single number.
A practical rule: anything with a high severity rating, such as safety, regulatory or loss of primary function, gets a preventive action regardless of the other scores.
Actions that matter
The value of an FMEA is in the actions it produces. Good actions:
- Change the design to remove or reduce the cause: a stop on the cable gland that prevents over-tightening, for example.
- Add or improve verification tests that would detect the failure before release.
- Feed special characteristics into drawings and the process FMEA so manufacturing controls them.
- Have owners, dates and a re-rating after completion.
Linking DFMEA to the rest of development
A DFMEA should connect to:
- Requirements: every critical function traced to a requirement.
- Design verification: test plans that target the high-risk failure modes, as we discussed in prototype or simulate.
- Tolerance analysis: dimensional failure modes checked with stack-ups.
- The process FMEA and control plan in manufacturing.
Keep it alive
An FMEA is a living document. Update it when the design changes, when tests reveal new failure modes and when field data arrives. The next product's DFMEA should start with everything this one learned.
Done well, a DFMEA is not paperwork. It is the cheapest insurance policy a product team can buy.
