Design FMEA: Prioritizing Risk with RPN

A Failure Mode and Effects Analysis (FMEA) is how engineers decide which risks to fix first. For each way a part can fail, you score three factors 1-10 and multiply them into a Risk Priority Number (RPN = Severity x Occurrence x Detection). You then attack the highest RPNs first. It's the same tool aerospace and medical-device teams use, and it scales perfectly to an FRC robot.

The three ratings (1-10 each):

• Severity - how bad the effect is (1 = unnoticed, 10 = hazardous/injury).
• Occurrence - how likely the failure is (1 = highly unlikely, 10 = almost certain).
• Detection - how likely you are to catch it before it bites (1 = almost certain to detect, 10 = almost impossible).

Worked example - robot safety FMEA (sorted by RPN):

Failure mode | Effect | S | O | D | RPN
Pneumatic vents while serviced | finger crush | 8 | 4 | 5 | 160
Loose SB-50 -> brownout | match loss | 4 | 5 | 5 | 100
Undersized branch wire overheats | smoke/fire | 8 | 3 | 4 | 96
Battery not strapped, ejects | struck-by, short/arc | 9 | 3 | 3 | 81
Sharp edge on intake | cut during handling | 4 | 5 | 3 | 60
Flywheel guard missing | hand laceration | 9 | 2 | 2 | 36

Reading it: the pneumatic-vent failure (RPN 160) tops the list - high severity, fairly likely, and hard to detect (you can't see stored air). That tells you to invest first in a manual vent + 'verify zero energy' step in your LOTO, not in the sharp-edge item (RPN 60) you'd notice anyway. Note that a high RPN can come from poor detection even at moderate severity - which is why the loose-SB-50 brownout (RPN 100) ranks above the missing flywheel guard (RPN 36): the guard's absence is obvious, the loose connector isn't.

The process: list every failure mode, score S/O/D as a team, sort descending by RPN, then for each high-RPN item write a corrective action with an owner and a due date - and the goal of every action is to drop one of the three scores (add a guard to cut Severity, redesign to cut Occurrence, or add a check/sensor to cut Detection). Re-score after the fix to prove the risk dropped.

An FMEA spreadsheet committed to your team repo, updated each season, demonstrates real engineering rigor - and it focuses your limited build-season hours on the failures that can actually hurt someone or lose a match.