What it is for
This analysis is preventive: it is done on a design or process before the failure occurs, to find out where it makes sense to add protection, redundancy or a check. It is also used retrospectively to organize the known failures of a piece of equipment.
Steps of an analysis
- Define the scope: system, functions, boundaries and operating conditions.
- List the elements or functions to be analyzed.
- List the failure modes of each element (open, short, drift, seizure, leak).
- Describe the effects at the local level, at the next level and on the complete system.
- Identify the causes of each mode and the current controls that prevent or detect it.
- Rate the risk using three criteria: severity of the effect, probability of occurrence and ability to detect it before it causes damage.
- Prioritize and define actions, and rate again after applying them.
How it is presented
The work is done in a table with one row per failure mode and columns for effect, cause, controls, ratings and actions. The rating scales are defined by the team and must be applied consistently.
How to interpret the result
The product of the ratings is useful for ranking, but it should not be taken as an absolute measure: two modes with the same product may require very different treatment. A severe effect with low probability usually deserves attention even if its overall score is low.
Limitations
- It normally considers one failure at a time, so combinations of failures are better captured with a fault tree.
- Its quality depends on the team’s experience and the data available.
- It is a living document: it is reviewed when the design changes or new field failures appear.