Skip to content
Six Sigma | Part 7 of 10

A Process FMEA Puts the Risks in Order

Severity, occurrence and detection rank what can go wrong in a process, so that actions go where the risk is highest.

Every process engineer knows several ways a process can fail. The problem is not the lack of ideas. It is deciding which ones deserve action this month.

A process FMEA, failure mode and effects analysis, puts them in order.

The questions per process step

For each step of the process the team asks:

  • What can go wrong? The failure mode, for example “bore too large”.
  • What happens then? The effect, for example “leak at assembly”.
  • Why could it happen? The cause, for example “thermal growth after start”.
  • What prevents or detects it today? The current controls.

Then three ratings, each from 1 to 10:

  • Severity: how serious is the effect for the customer?
  • Occurrence: how often does the cause happen?
  • Detection: how likely is it that current controls catch it before the part leaves? Here 10 means almost no chance.

The example

Failure mode Cause S O D RPN
Bore too large Thermal growth after start 7 7 4 196
Bore too small Tool wear beyond limit 7 4 5 140
Bore not round Clamping force too high 6 3 6 108
Burr in bore Worn deburring tool 5 4 4 80

The risk priority number, RPN, is severity × occurrence × detection. The warm-up cause ranks first, which matches what the data showed in Analyze.

Severity first

RPN alone can mislead. A failure with severity 10 and a low RPN may still need action, because the consequences are unacceptable. Many companies use action priority tables that weigh severity first, then occurrence, then detection. Whatever method your company uses, apply it consistently.

Actions that change the ratings

Good actions lower occurrence by removing the cause, or lower detection by catching the failure earlier. Inspection alone only changes detection and leaves the cause in place.

In the example, an automatic warm-up cycle lowered the occurrence of the thermal growth cause from 7 to 2. A tool change limit with a counter in the control did the same for tool wear. After the actions, the team re-rated each line and the highest RPN dropped below 60.

Make it a team exercise

Two hours with the setter, an operator, maintenance and quality find more failure modes than a week alone at a desk. Use the SIPOC process steps as the structure, so nothing is forgotten.

Use the scales of your company or your customer. A severity of 7 in one plant should mean the same in another.

Keep it alive

An FMEA filed after the project is a document. An FMEA updated with every process change, every complaint and every new failure is a tool. Link it to the control plan: every high-ranking cause should have a control that appears in the plan.

The FMEA does not find root causes on its own. It tells the team where to look first and makes sure that, once the project is done, the known risks stay under control.

Newsletter

One Lean idea.
Every two weeks.