Analytical Tools: FMEA, Fault Tree, Fishbone, and Change Analysis
Key Takeaways
- FMEA asks how components, tasks, or process steps can fail and what the effects would be.
- Fault tree analysis works backward from an undesired event to combinations of causes and barriers.
- Fishbone diagrams organize potential causes but do not prove which cause is correct without evidence.
- Change analysis compares what changed against what was expected, making it useful after incidents or before modifications.
Choose the analysis tool that fits the question
Safety professionals use structured tools to reduce guessing. The tool should match the decision. Some tools look forward to find potential failures before an event. Some work backward from an event. Some organize causes for discussion. Some compare a changed condition with the expected condition. The ASP exam may ask which tool is most appropriate for a scenario.
Failure modes and effects analysis, or FMEA, is a forward-looking method. A team reviews a process, component, or task step and asks how it can fail, what the effect would be, what causes the failure, what controls exist, and what actions are needed. FMEA is useful during design, procedure development, equipment selection, maintenance planning, and process changes.
Fault tree analysis starts with a top event such as explosion, fall from elevation, chemical release, or unexpected startup. The team then maps the combinations of failures that could produce that event. Fault trees help show how barriers interact. A serious event may require several things to fail together, such as isolation not performed, verification skipped, and startup initiated by another worker.
Fishbone diagrams, also called cause-and-effect or Ishikawa diagrams, organize possible causes into categories. Common categories include people, methods, machines, materials, environment, and management systems. Fishbones are useful for brainstorming and organizing investigation thinking, but they are not proof. Each possible cause still needs evidence.
Change analysis compares the incident or new condition with the normal or expected condition. It asks what changed in people, equipment, materials, methods, environment, timing, workload, supervision, software, layout, production rate, maintenance, or contractors. It is especially useful when a familiar process suddenly produces an unexpected result.
| Tool | Direction | Best use | Caution |
|---|---|---|---|
| FMEA | Forward from steps or components | Anticipating failures before harm | Needs knowledgeable team and realistic ratings |
| Fault tree | Backward from top event | Understanding combinations of causes | Can miss causes if top event is poorly defined |
| Fishbone | Organizes possible causes | Brainstorming and investigation structure | Does not prove causation by itself |
| Change analysis | Compares changed to normal | Incidents, modifications, and process drift | Requires accurate baseline information |
A common error is selecting a tool because it sounds sophisticated rather than because it answers the question. If the question asks how a new process step could fail before startup, FMEA is often a good fit. If it asks what combinations of failures could lead to a catastrophic event, fault tree logic fits. If a team needs to organize many possible causes during an investigation, a fishbone can help. If the key clue is that the task worked for years until something changed, change analysis is strong.
These tools should lead to controls. Analysis without action is not risk reduction. Results should identify higher-level controls, procedure changes, training needs, maintenance changes, purchasing changes, alarms, interlocks, inspections, or management system improvements. Actions should be assigned, tracked, verified, and reviewed for residual risk.
Use evidence to avoid bias. Interviews, records, photographs, monitoring data, maintenance history, training records, inspection reports, and field observations can confirm or reject suspected causes. Do not stop at the first explanation, especially if it blames the worker while ignoring design, scheduling, supervision, or system contributors.
A team is reviewing a new automated valve system before startup and wants to identify how each component could fail and what each failure would cause. Which tool best fits?
An investigation team wants to map combinations of failures that could lead to an unexpected equipment startup. Which method is most appropriate?
A fishbone diagram lists poor lighting, rushed production, missing tools, and unclear procedures as possible contributors. What should the team do next?