What is reliability-centered maintenance in simple terms?

It is a structured way to decide how to maintain each asset by asking what it does, how it can fail, and whether that failure matters. Instead of putting everything on a fixed PM schedule, RCM assigns each failure mode the cheapest task that actually manages it, including a deliberate decision to run some items to failure.

What are the seven questions of RCM?

Functions and performance standards, functional failures, failure modes, failure effects, failure consequences, proactive tasks, and default actions when no proactive task is worthwhile. SAE JA1011 requires all seven; a process that skips any of them is not true RCM.

What is the difference between RCM and preventive maintenance?

Preventive maintenance is one possible task, servicing on a fixed interval. RCM is the process that decides whether preventive maintenance is even the right task for a given failure mode, or whether condition monitoring, failure-finding, redesign, or run-to-failure fits better. RCM often concludes that many assets should not be on a fixed PM at all.

Is run-to-failure allowed in RCM?

Yes, for failures with only minor, non-safety consequences where no proactive task pays for itself. RCM makes run-to-failure a documented, deliberate choice rather than an accident. It is never allowed for failures with safety or environmental consequences.

What standard defines RCM?

SAE JA1011, 'Evaluation Criteria for Reliability-Centered Maintenance Processes,' defines the minimum criteria a process must meet, chiefly the seven questions. SAE JA1012 is the companion guide that explains and illustrates how to apply them.

Why do most equipment failures not follow the bathtub curve?

The 1978 Nowlan and Heap study found that only a minority of components have a clear wear-out age. About 89% of the items studied failed randomly or were dominated by infant mortality, where failure risk is highest right after installation. That is why condition monitoring often beats fixed overhaul intervals.

How long does RCM take to pay back?

Public DOE-cited figures put a well-run predictive program at 8 to 12 percent savings over preventive alone, and a referenced NASA RCM case reported a 1.75 to 2.2 year return on investment. Payback depends heavily on how reactive the starting point was.

Reliability-Centered Maintenance (RCM) Explained

The seven questions of SAE JA1011, the task-selection logic behind them, and why most fixed PM schedules miss the failures that matter.

Reliability-centered maintenance (RCM) is a structured process for deciding the right maintenance strategy for every asset, one failure mode at a time. It answers seven questions about an asset's functions, failures, and consequences, then matches each failure mode to a proactive task or a deliberate run-to-failure decision. The method is defined by the standard SAE JA1011.

RCM is not a checklist you buy off a shelf. It is a way of reasoning about failure that replaces "how do we service this machine" with a harder, more useful question: "what does this machine do, how can it stop doing it, and does that even matter enough to act on?" This guide walks the seven questions, the task-selection logic behind them, and the numbers that explain why RCM exists at all.

What is reliability-centered maintenance?

Reliability-centered maintenance is a disciplined process that determines what must be done to keep each asset doing what its users need, in its current operating context. Instead of applying one strategy everywhere, RCM assigns each failure mode its own answer: an on-condition inspection, a scheduled restoration or replacement, a failure-finding check, a redesign, or a documented decision to run the item to failure.

That last option surprises people. RCM is famous for concluding that a large share of assets should not be on a fixed preventive schedule at all, because scheduled overhaul does nothing to prevent the way they actually fail. It is the opposite of "more PMs are always better." RCM sits one rung above raw strategy selection on the equipment reliability ladder because it tells you which strategy each asset earns.

The seven questions defined by SAE JA1011. A process that skips any of them is not RCM.

What are the seven RCM questions?

Every genuine RCM process answers seven questions for each asset, in order. SAE JA1011 exists precisely because so many vendors were selling shortcut methods that dropped one or two of them:

What are the functions and performance standards? What the asset is supposed to do, stated with a number, "pump 400 gallons per minute," not "pump water." You cannot define failure until you define success.
In what ways can it fail to deliver those functions? These are the functional failures: total loss, partial loss, or delivering out of spec.
What causes each functional failure? The failure modes, the specific mechanisms: a seized bearing, a worn seal, a plugged strainer, an operator error. This is where RCM overlaps with FMEA.
What happens when each failure occurs? The failure effects, the physical evidence, the damage, the safety exposure, and how you would notice.
In what way does each failure matter? The consequences. RCM sorts every failure into hidden, safety/environmental, operational, or non-operational, and that sort drives everything downstream.
What can be done to predict or prevent each failure? The proactive tasks: on-condition monitoring, scheduled restoration, or scheduled discard.
What should be done if no proactive task is worthwhile? The default actions: failure-finding checks, redesign, or a deliberate decision to run to failure.

Question seven is the one shortcut methods drop, and it is the one that keeps RCM honest. Many failure modes genuinely have no cost-effective proactive task. RCM forces you to name that fact and choose deliberately, rather than bolting on a PM that accomplishes nothing.

Why does RCM exist?

RCM exists because the aviation industry proved, with data, that most preventive overhaul was wasted. In 1978 Stanley Nowlan and Howard Heap of United Airlines published Reliability-Centered Maintenance for the U.S. Department of Defense. Studying failure data across the fleet, they found six distinct failure patterns and one uncomfortable headline: only a minority of components wear out on a predictable schedule.

Roughly 89% of the items studied showed no wear-out age at all, their failures were random or dominated by "infant mortality," where the risk of failure is highest right after installation or overhaul. The single largest pattern, at about 68%, was infant mortality followed by a constant failure rate. In other words, tearing a healthy machine apart on a calendar often introduces failure instead of preventing it. DOE's O&M guidance summarizes the same finding plainly: age-related failures account for less than 20% of failures, and random failures make up roughly 80%.

The old mental model, the bathtub curve where everything wears out on schedule, turned out to describe only a small slice of real equipment. That is the intellectual foundation for condition-based maintenance and predictive maintenance: if you cannot predict failure by age, you must watch the equipment's actual condition instead.

For most components there is no age at which failure suddenly becomes likely, so a fixed overhaul interval cannot catch it.

How does RCM choose a task for each failure mode?

RCM chooses tasks by consequence first, then by technical feasibility. The consequence category, set in question five, decides how hard you must try to manage a failure, and the physics of the failure decides which task can actually catch it. The four categories are:

Hidden failures. The failure is not evident to operators during normal work, a standby pump that will not start, a tripped relief valve. These need failure-finding tasks: periodic checks that confirm the protected function still works.
Safety and environmental consequences. The failure could hurt someone or breach the environment. A proactive task is mandatory; if none is feasible, the item must be redesigned. Run-to-failure is not allowed here.
Operational consequences. The failure costs production, lost output, scrap, downtime. A task is justified if it costs less than the production it protects.
Non-operational consequences. The failure only costs the repair itself. Here run-to-failure is often the correct, cheapest answer.

Only after the consequence is known does RCM ask whether a task is technically feasible and worth doing. A task is worth doing when it is both feasible, it actually detects or prevents the failure mode, and cost-effective against the consequence it avoids.

Consequence category	Question RCM asks	If no good proactive task exists
Hidden	Will loss of the protected function be evident on its own?	Failure-finding task; redesign if the multiple-failure risk is unacceptable.
Safety / environmental	Could this failure hurt someone or breach permits?	Redesign is mandatory. Run-to-failure is not permitted.
Operational	Does the failure cost production, not just repair?	Run-to-failure if a task would cost more than the downtime it prevents.
Non-operational	Does it only cost the direct repair?	Run-to-failure is usually the right choice.

RCM's default-action logic. Consequence severity, not habit, decides how much prevention a failure earns.

What tasks can RCM assign?

RCM's task menu is short and specific. Every failure mode ends up on one of these:

On-condition tasks. Inspect or monitor for a warning sign, then act before failure. This is the home of condition monitoring and predictive maintenance, and it is the preferred task whenever a detectable warning exists on the P-F curve.
Scheduled restoration. Overhaul or restore the item at a fixed interval, regardless of condition, only worthwhile when there is a real wear-out age.
Scheduled discard. Replace the item at a fixed life, again only where a genuine safe-life limit exists.
Failure-finding. Periodically test hidden, protective functions to make sure they still work.
Run-to-failure. A deliberate decision to do no proactive work, chosen when consequences are minor and no task pays for itself.
Redesign (one-time change). Change the item, process, or operating context when no maintenance task can manage the consequence.

The point is that most of these live outside the calendar. When RCM assigns on-condition tasks, the spares list and the work plan shift with it, which is why RCM feeds directly into reliability-centered spares and into your spare parts inventory management decisions.

What does RCM cost and save?

RCM analysis is front-loaded work, but the payback is well documented in public data.

Moving from reactive to a reliability-based program lowers average annual maintenance cost sharply: DOE's O&M guidance estimates reactive maintenance runs roughly a 200% cost increase over an RCM baseline, preventive about 150% and predictive about 50% (PNNL / DOE FEMP).
A properly run predictive program, the on-condition heart of RCM, saves 8–12% over preventive alone and preventive saves 12–18% over reactive (PNNL, Maintenance Approaches).
A NASA RCM case cited in the same guidance reported a return on investment of 1.75 to 2.2 years (PNNL / DOE FEMP).
The finding that anchors the whole method: age-related failures are under 20% of all failures; random failures are roughly 80% (Nowlan & Heap, 1978).

Where does RCM fit against other strategies?

RCM is the decision layer, not a competing strategy. Preventive, condition-based, and predictive maintenance are tools; RCM is the reasoning that decides which tool each asset gets and why. It is closely tied to total productive maintenance which puts operators at the front line of the same reliability goal, and it is measured with the same reliability metrics, most importantly mean time between failures (MTBF).

In practice, few plants run the full classical RCM on every asset; it is heavy. Most start on their most critical equipment and use streamlined variants elsewhere. If you want the project mechanics, team, scope, timeline, and how to load the results into a CMMS, see RCM implementation steps. And if you are still fighting fires, start with the culture shift in reactive vs proactive maintenance.

Harmony's agents sit on top of this decision layer: they read the failure history, flag the assets whose behavior is drifting, and turn an RCM task list into work that actually gets scheduled and closed. See how that plays out on a real floor in the CLS case study.