Key points
- Do not analyse every breakdown. Save RCA for failures that repeat, stop a critical asset, cost a lot or put someone at risk.
- Collect the evidence before the repair destroys it: photos, the failed part, meter readings and what the operator saw.
- Stop asking why when you reach something your team can change. "The bearing wore out" is not a root cause; "the bearing was not on any greasing route" is.
- Look for causes in the system, not in people. "The technician forgot" leads nowhere; "the task was not on the checklist" leads to a fix.
Which failures deserve a root cause analysis
An analysis takes a few hours of several people's time, so it should go where it pays back. A simple rule is to trigger one when a failure meets any of these conditions:
- It repeats. The same fault on the same asset, or the same fault across several identical assets, within a set period such as 30 or 90 days.
- It stopped a critical asset. A machine the operation cannot work without, or one with no standby.
- It cost more than a threshold you set, in parts, contractor time or lost output.
- It caused or nearly caused harm to a person or the environment. These need an analysis whatever the cost.
Write the trigger down, so the decision is not made case by case in the middle of a busy week. The fastest way to find repeats is a breakdown register where every failure carries a failure code: sort by code and by asset and the repeat offenders stand out.
Step 1: collect the evidence before the repair
Most analyses fail before they start, because the evidence went in the scrap bin with the old part. Agree that for any failure that might need an analysis, the technician captures the following before or during the repair:
- Photos of the failed component in place, before anything is removed.
- The failed part itself, labelled and kept until the analysis is done.
- The meter reading at the time of failure and the date of the last service.
- What the operator saw, heard or smelled beforehand, in their own words.
- Operating conditions: load, weather, shift, and anything unusual about the job.
Step 2: ask why with the 5 Whys
The 5 Whys starts from the failure and asks why it happened, then asks why about each answer, until the answer is something your team can change. Five is a guide, not a rule: some chains end at three, some need seven. Every answer should be backed by evidence, not by a guess.
Notice that the chain did not stop at "the bearing ran without grease". That answer only leads to another repair. It also did not stop at "the technician missed it", because the technician followed the checklist exactly. The useful answer was further down.
Step 3: widen the search with a fishbone diagram
The 5 Whys follows one path, which is its weakness: if the first answer is wrong, everything after it is wrong too. When a failure has several plausible causes, draw a fishbone (also called an Ishikawa or cause and effect diagram) first. Write the failure at the head and list possible causes under a few headings. For equipment failures these six headings work well:
| Heading | Questions to ask |
|---|---|
| Machine | Design, age, wear, previous repairs on this component |
| Method | Service tasks, intervals, operating procedure, start-up and shutdown |
| Material | Spare part quality, wrong part fitted, lubricant type, fuel quality |
| People | Training, instructions available, workload, handover between shifts |
| Environment | Dust, heat, water, vibration, load beyond design |
| Measurement | Were the warning signs measurable, and was anyone measuring them? |
Then cross out the causes the evidence rules out, and run a 5 Whys on the ones that remain. The two tools work best together: the fishbone stops you fixing on the first idea, the 5 Whys takes the strongest candidate down to something you can change.
Step 4: turn the root cause into a change
An analysis that ends in a report and no change has only cost time. Each root cause should produce at least one action, with an owner and a date. Typical actions are:
- Add, change or remove a preventive task, or change its interval. See preventive maintenance intervals.
- Add a check to a daily inspection, so the warning sign is caught by an operator.
- Change the spare part, the supplier or the lubricant.
- Rewrite an operating or service procedure, and train the people who use it.
- Apply the fix to every identical asset, not just the one that failed.
Raise each action as a work order so it is scheduled, done and closed like any other job, rather than living in a meeting note. See what is a work order.
Step 5: check that the fix worked
Set a date to look again, long enough for the failure to have come back if the fix was wrong. Then check the breakdown record for the same failure code on the same assets. If it has not returned, close the analysis. If it has, the root cause was wrong or the action was not done, and the analysis starts again with the new evidence. Over a longer period the effect should show up in your reliability figures; MTBF, MTTR and availability explains how to read them.
Common root cause analysis mistakes
- Stopping at the failed part. "The seal failed" describes the failure. It does not explain it.
- Blaming a person. If a mistake was possible, the system allowed it. Ask what would have stopped it.
- Guessing instead of checking. Each why needs evidence: a photo, a reading, a record. A confident guess sends the whole chain the wrong way.
- Doing it alone. The operator, the technician who did the repair and the planner each see a different part of the story.
- No follow-up. Without a date to check, nobody knows whether the fix worked until the next breakdown.
Root cause analysis in Solmex
Solmex does not run the analysis for you, and there is no built-in 5 Whys or fishbone form. What it gives you is the record the analysis depends on. Every breakdown can carry photos and a failure code from a code list your site defines, and the diagnosis step records the root cause and findings in words. The work done on the repair records its own root cause, parts and photos. Breakdown analysis ranks failure causes and assets and lists failures that repeated within 30 days, which is where most analyses should start. The corrective actions are raised as work orders and preventive template changes like any other job. See breakdowns and reports for details.