A worked RCA, start to finish

Root cause analysis example: the gearbox that kept failing

A maintenance root cause analysis worked start to finish: a 5 whys that stopped at operator error, a Fishbone, causes proven or refuted, and actions.

Plant Floor · Practical maintenance management

Here is a root cause analysis example worked from start to finish on one maintenance failure. A conveyor gearbox fails for the third time in a year. A quick 5 whys stops at "operator error". A Fishbone session finds the real causes. Each cause gets evidence, an owner and a date. Each confirmed cause gets an action, and every action gets checked.

Every step below follows the free root cause analysis template. Download it and fill it in alongside.

Original fictional scenario. This example was created for this page to show the method. It is not a real plant, a customer result or a scene from the book. The method and the working bands come from Fix the System.

1. The failure and why it gets an RCA

Gearbox GB-14 drives the infeed conveyor on packaging Line 2. It tripped on high temperature and the output bearing was found failed. This is the third failure under the same failure code in twelve months.

All three work orders say a version of the same thing: replaced bearing, topped up oil, tested OK. Each repair was done correctly. None of them asked why.

Two triggers on the form are checked: three failures under one code in twelve months, and the technician flagged a recurrence at close-out. The planner puts GB-14 on the repeat-failure watch list and opens an RCA that week.

2. The first attempt: a 5 whys that stopped too early

The supervisor runs a quick 5 whys at the line.

Why did the bearing fail? It ran low on oil.
Why was the oil low? Nobody topped it up.
Why not? The operator did not report the low sight glass.

Written root cause: operator error. Proposed action: remind the operators.

That answer is on the template's list of answers that mean you are not done. Ask one more why. Why did the operator not report it? Because a guard added last year hides the sight glass from the walkway.

Then a second branch shows up. The technician who drained the oil says it was low and milky. Milky oil means water. Now there are two stories, and a single line of whys cannot hold both. When you catch yourself rewriting a 5 whys because there is another branch, switch tools.

3. Pick the tool and the team

Several conditions came together, so this is a Fishbone. The planner convenes it, because the planner can book the people. The room holds the technician who did the last two repairs, the Line 2 operator, the planner, the area supervisor and the reliability engineer. The maintenance manager facilitates and mostly listens.

The rule for the session: for each category, write what actually contributed, in someone's observation. Not what could have. Collect first, rank later.

4. The Fishbone, as the team filled it in

Category	What actually contributed
People	Operators stopped reporting the oil level once a new guard hid the sight glass from the walkway.
Method	The lube PM still describes the old access point. Nobody updated it when the guard went on.
Machine	The guard blocks the fill port and the sight glass. The breather is a standard vented cap.
Material	After a supplier change, the storeroom issues a different oil under the same stock code.
Measurement	No oil sampling on this gearbox. The high-temperature trip is the first warning anyone gets.
Environment	Sanitation now washes this line down every shift, and the hose reaches the drive end.

5. Rank the causes, then prove or refute them

A Fishbone is a list of suspects, not a verdict. The team ranks by confidence and impact, then gives each top cause the evidence that would settle it, an owner and a due date. The answers come back over the following weeks.

Cause	Evidence	Owner	Result
Water gets in through the vented breather during washdown.	Oil sample for water. Inspect the breather right after a washdown.	Reliability engineer	Confirmed. The oil was milky and the sample showed water.
The lube PM is signed off without oil being added.	Compare PM sign-offs with the lube log. Walk the route with the lube technician.	Planner	Confirmed. The fill port cannot be reached without removing the guard, so the step was skipped.
The substitute oil is the wrong grade.	Storeroom issue records against the manufacturer's specification.	Planner with storeroom	Refuted. The substitute meets the specification.
Operators do not report a low level.	Stand at the operator's position and look for the sight glass.	Area supervisor	Confirmed, but it is a result of the guard, not a separate root cause.

One suspect was refuted. That is normal, and it is the point of collecting evidence. Without it, the plant would have changed oil suppliers and kept the water.

6. Root cause statements and the confirmation test

Two confirmed root causes, each in one sentence:

Washdown water enters GB-14 through a vented breather that was never specified for a washdown area.
The guard added last year blocks the fill port, and the lube PM was never updated, so the PM gets signed off without oil being added.

The confirmation test asks: if we fix these, does the failure stop on this gearbox and on similar ones in similar conditions? Yes, provided every gearbox in a washdown zone gets the same fix. That is why the similar-assets list matters.

Compare the two statements with "operator error". A root cause statement names a condition you can change. Operator error names a person you can blame.

7. Actions, one channel each

Every confirmed cause gets at least one action, opened as a work order or change request with an owner and a date. Each action goes into one channel: PM, CBM, SOP or design.

Action	Channel	Owner
Fit a sealed breather rated for washdown on GB-14.	Design	Reliability engineer
Fit the same breather on every gearbox in the washdown zones of Lines 2 and 3.	Design	Reliability engineer
Add an extended fill and level line outside the guard.	Design	Maintenance supervisor
Rewrite the lube PM: new access point, a photo of the correct level, and an oil check of clear or milky.	PM	Planner
Sample oil on washdown-zone gearboxes every quarter.	CBM	Reliability engineer
Washdown procedure: no direct spray on drive ends. Operations signs off.	SOP	Area supervisor

Look at what is missing: "remind the operators". A reminder changes nothing while the sight glass stays hidden. Once the level line sits outside the guard, the operator can see it without anyone asking.

8. Verify at 60 days and at 6 months

At 60 days, the RCA owner checks each action for evidence that it is done: the work order closed, the PM revision live, the breathers fitted. A blocked action gets a reason and a new date.

At 6 months, the question is simple. Has GB-14 failed again, and are the oil samples from the washdown-zone gearboxes clean? If the failure is back, one of three things is true: an action never closed, the change did not hold, or the cause was wrong. Each one points to a different fix.

Fix the System gives two working bands for the whole RCA program. Aim for 90% or more of flagged repeat failures with a finished RCA. Aim for under 15% of those failures coming back within two quarters.

Questions planners ask

What should a root cause analysis report include?

The failure in one sentence and why it was flagged. The tool and the team. The causes found, with the evidence that confirmed or refuted each one. The root cause statements and the confirmation test. The actions with channel, owner and date. The 60-day and 6-month checks. The RCA template has a section for each.

How do you know you found the root cause?

Run the confirmation test. If fixing the cause would stop the failure on this asset and on similar ones, you have it. If the answer is a category such as operator error, part failure or wear and tear, you do not. Keep asking why, or gather more evidence.

Can a failure have more than one root cause?

Yes, and in a plant it often does. This example had two: water ingress and a PM that could not be done as written. That is why a Fishbone, which holds several branches at once, fit better than a 5 whys, which follows one line.

When is a 5 whys enough?

When the failure is one event with one straight line of causes you can walk backwards in your head. Then the 5 whys template takes about 20 minutes. If a second branch appears, as it did here, switch tools.