Why the Same Machine Keeps Failing: A Manufacturing RCA Case for Bad Actor Analysis

Every plant has one. The pump that fails every few months no matter how many times maintenance rebuilds it. The gearbox that gets replaced on a schedule instead of a failure curve. The conveyor drive that everyone just expects to go down during the third shift.

Reliability professionals call these bad actors: assets or components that fail repeatedly, consume disproportionate maintenance hours, and never seem to get fixed for good. Bad actor analysis is the discipline of using root cause analysis to find out why, instead of accepting repeat failure as the cost of doing business.

This article walks through why bad actors survive normal troubleshooting, what a real RCA on a bad actor looks like, and how to keep the fix from wearing off in six months.

What Makes an Asset a Bad Actor

A bad actor isn’t just an old or unreliable machine. It’s a piece of equipment where the same failure mode, or a small family of failure modes, keeps recurring despite repair. The bearing gets replaced and fails again in four months. The seal gets swapped and the leak comes back. The motor gets rewound and burns out again by next quarter.

Plants track this informally all the time. Someone on the team can usually name the worst three or five assets without looking at a report. The problem is that naming the bad actor isn’t the same as fixing it. Most bad actors have already been “fixed” multiple times. That’s exactly what makes them bad actors, and it’s the same pattern behind why most equipment failures happen again, whether the plant makes cement, chemicals, or car parts.

Why Normal Troubleshooting Doesn’t Break the Cycle

The reason bad actors survive is almost always the same: every repair addressed the physical cause and stopped there.

A technician replaces a bearing that failed from fatigue. The work order says “bearing failed, replaced bearing.” Case closed, until it fails again. What never got investigated is why the bearing fatigued in the first place. Was it misaligned? Was it the wrong bearing spec for the load? Was there a lubrication schedule that got skipped because the lube tech was pulled onto a shutdown crew for three weeks?

This is the core failure of shallow troubleshooting on a bad actor: it treats a physical root cause as if it were the whole answer. In PROACT® terms, a physical root is the tangible failure mechanism, the bearing that fatigued, the seal that extruded, the coupling that sheared. But every physical root exists because of a human decision or action somewhere upstream, and every human root exists because of a system or process that allowed it, a latent root.

Fixing the physical root fixes one failure. Fixing the human root changes one person’s decision pattern. Fixing the latent root can eliminate that failure mode across every similar asset in the plant, sometimes across every plant in the network.

A Bad Actor Case: The Chronic Pump

A useful real-world example is a chronic pump failure that shut down operations at a facility in Aberdeen after repeated breakdowns, at a cost of roughly $180,000 in a single event. The pump wasn’t new to the maintenance team. It had a documented history of repeat failure, and each prior repair had targeted the immediate physical damage without asking why the pattern kept repeating.

A structured RCA using the PROACT® methodology traced the failure past the physical damage to the human and systemic factors that had let the same failure mode return again and again. Once those systemic gaps were closed, the chronic pattern stopped, not because the pump got a better bearing, but because the conditions that kept breaking it were finally addressed.

That’s the difference between a repair and an investigation. A repair returns the asset to service. An investigation asks why the asset keeps needing repair.

How to Actually Investigate a Bad Actor

1. Confirm it’s chronic before you build a team. Pull the maintenance history. If the same failure mode shows up three or more times in a defined period, or the cumulative cost of “small” repeat failures rivals a single catastrophic event, it qualifies for a real investigation, not another work order.

2. Build a logic tree, not a 5-Why list. 5-Why works for simple, single-path failures. A bad actor with a repeat history usually has multiple contributing paths: a design issue plus a maintenance practice plus a training gap. A logic tree lets the team branch out every plausible hypothesis at each level (physical, human, latent) instead of following one narrow chain of “why” questions to a single, convenient answer.

3. Verify every hypothesis with data, not opinion. “We think it’s misalignment” isn’t a finding. Laser alignment readings, vibration data, oil analysis, and failed-part inspection are what turn a guess into a verified cause. Any hypothesis the team can’t verify gets crossed off the tree, no matter how plausible it sounds in the room.

4. Follow the human root to the latent root. If a technician skipped a step, ask why. Was the procedure unclear? Was there time pressure from a production schedule? Was the tool needed for the job checked out to another crew? The human root is rarely the end of the investigation. It’s usually the signpost pointing at the real systemic issue.

5. Track the corrective action to completion. This is where most bad actor investigations quietly fail. The team finds the real cause, writes a solid corrective action, and then the action sits in a spreadsheet nobody reopens. If the corrective action isn’t tracked to closure with an owner and a date, the bad actor comes back, and the next investigation starts from zero.

When a Bad Actor Investigation Pays for Itself

Chronic failures rarely get evaluated on their true annual cost because each individual event looks small next to a catastrophic failure. But three $15,000 repairs a year on the same pump, for five years, is $225,000 spent maintaining a problem instead of solving it. Aggregating chronic failure costs over time is usually the fastest way to get management support for a proper investigation instead of another quick fix.

The same logic applies at scale. A single-site fix is useful. A fix that gets replicated across every identical asset in a multi-site operation is where bad actor analysis produces its biggest return, because the same latent root is often quietly driving the same failure at more than one location.

Breaking the Cycle for Good

A bad actor doesn’t stay a bad actor because it’s poorly designed or inherently unreliable. It stays a bad actor because every repair before this one stopped at the physical cause. Structured root cause analysis, done with a logic tree and carried through to a tracked corrective action, is what actually breaks the pattern.

If your team is still treating repeat failures as routine maintenance instead of an investigation trigger, that’s the first thing worth changing. PROACT® RCA training builds the skills to run that investigation correctly the first time, and to make sure the fix sticks. For teams managing bad actors across multiple lines or sites, EasyRCA gives you a way to track recurring failure patterns and corrective actions in one place instead of a scattered folder of old work orders.