Most weak FMEAs fail before the first failure is written down. The scope is unclear, the team is one engineer, and the process is described from memory. The first two steps of the method prevent this.
Step 1: five questions
Planning and preparation answers five questions in writing. For the new hose assembly cell H2 the answers fit on one page:
- Why? A new cell for an existing hose family, and 12 complaints in the sister cell H1.
- What? Hose family HA-20, from cutting to packing. Incoming goods and shipping are out of scope.
- Who? The process engineer as owner, a moderator, the quality engineer, the cell leader, a setter and a maintenance technician.
- When? Five sessions of two hours in three weeks, starting 10 weeks before the start of production.
- With what? Complaints, reject data and the FMEA of cell H1, plus rating guides and a priority logic that everyone accepts.
If your customer prescribes rating tables or a priority logic, those apply. If not, agree your own before the first session.
The team
Two roles are fixed. The owner answers for content and deadlines and keeps the document current after the launch. The moderator leads the method and writes on the screen. The moderator has no stake in the ratings, which is why the owner should not moderate.
The setter or operator is the member most often left out and most often right. They know what goes wrong today and is fixed without a report.
Step 2: structure analysis
Before anyone asks what can fail, the team agrees on what the process consists of. Three levels are enough:
- The process: the whole scope of this FMEA.
- Process steps: stations or operations that change or check the product, with the numbers of the process flow chart.
- Work elements: what acts on the product in each step. The usual four are machine, person, material and environment.
For cell H2 the structure has 7 steps and 20 work elements:
| Step | Machine | Person | Material |
|---|---|---|---|
| 10 Cut hose | Saw with length stop | Operator | Hose reel |
| 20 Clean hose | Cleaning nozzle | Operator | Projectile |
| 30 Insert fittings | Marking fixture | Operator | Fittings with seal rings |
| 40 Crimp fittings | Crimper | Setter | Die set |
| 50 Proof test | Test bench | Operator | Test fluid |
| 60 Cap and label | Label printer | Operator | Caps |
| 70 Pack | Packer | Box |
Why work elements matter
Work elements look like a formality. They are where causes live.
A failure mode at crimping, such as a crimp diameter that is too large, has its cause in the crimper, the die set or the setter: the wrong program, a worn die, a setting changed without release. A team that knows the work elements asks about each of them in turn. A team without them writes “process not capable” or “operator error” and has nothing to act on.
Two rules for the structure
Same numbers everywhere. The flow chart, the FMEA, the control plan and the work instruction use the same step numbers. When step 40 is crimping in one document and step 4 in another, nobody can follow a failure from the analysis to the station.
Go and look. Walk the process, or a comparable one, with the whole team before the first session. A structure made at the desk misses the two similar hose reels that stand side by side at the saw. That detail later became a row with a severity of 10.
What you have after two steps
A one-page scope, a list of process steps with their numbers and the work elements per step. No failure has been named yet, but the team has seen the process together. The next part turns each step into a function and each function into failure chains.