Condition-Based Maintenance: A Practical Guide

Q: What is the difference between condition-based and preventive maintenance?

Preventive maintenance acts on a fixed schedule, by calendar or run hours, regardless of the asset's actual state, so it often replaces healthy parts early and can still miss faults that develop between intervals. Condition-based maintenance acts on a measured condition instead, such as vibration, temperature or a process signal, and triggers work only when that measurement shows a fault is developing. The result is fewer surprise breakdowns and less wasted work, because you intervene when the machine actually needs it rather than when the calendar says so.

Q: Is condition-based maintenance the same as predictive maintenance?

They are closely related but not identical. Condition-based maintenance acts when a current measurement crosses a threshold, telling you a fault is present now. Predictive maintenance goes one step further and uses the trend to forecast when failure will occur, so work can be planned further ahead. Both rely on monitoring the asset's condition; predictive adds a time-to-failure estimate on top of the condition-based trigger.

Q: What can you monitor for condition-based maintenance?

Common condition signals are vibration on rotating equipment, temperature and thermography on electrical and mechanical assets, oil analysis on gearboxes and hydraulics, and motor current or power draw. On production lines the machine's own PLC signals and machine vision are powerful condition indicators too, because rising stop frequency, lengthening cycle times and creeping micro-stops are early signs of a developing fault. The right signal is the one that gives the longest, clearest warning before functional failure.

Q: Which assets should use condition-based maintenance?

Condition-based maintenance pays most on critical assets whose failure is expensive, unsafe or hard to predict, and on assets where a developing fault is actually detectable in advance. Low-criticality, redundant or cheap-to-replace assets are often better run to failure or maintained on a simple schedule, because the cost of monitoring outweighs the benefit. A criticality assessment, as used in reliability-centered maintenance, tells you which assets earn condition monitoring.

Key takeaways

Condition-based maintenance (CBM) triggers work off a measured condition crossing a threshold, vibration, temperature, a process signal, not off a fixed calendar or run-hours schedule.
It targets the P-F interval: the window between the point a fault becomes detectable and the point it fails. Act inside that window and you avoid both the breakdown and the early swap.
CBM pays most on critical assets whose failure is expensive or hard to predict. Low-criticality assets are often cheaper to run to failure or maintain on a simple schedule.
The value is not in the sensor, it is in turning a signal into a fixed problem fast. A reading nobody triages is just noise; the loop from detection to action is what makes CBM work.

Most maintenance is scheduled by the calendar: replace the part every so many weeks or run-hours, whether it needs it or not. It is simple, and it is wasteful at both ends. You swap healthy components too early, and you still get surprised by the ones that fail between intervals. Condition-based maintenance replaces the calendar with the machine's own condition as the trigger, acting only when a measurement says a fault is actually developing. Done well, it cuts unplanned failures and unnecessary work at the same time, which is why it sits between blanket preventive schedules and full predictive forecasting as the practical middle ground for most plants.

What condition-based maintenance is

There are four broad maintenance strategies. Reactive, or run to failure, fixes things after they break. Preventive acts on a fixed schedule by time or usage. Condition-based acts when a monitored condition crosses a threshold, meaning a fault is present now. Predictive goes further still and forecasts when the failure will happen. CBM is the point where maintenance stops guessing from a calendar and starts responding to the asset itself. The preventive versus predictive maintenance guide compares the neighbours on either side; this one focuses on the condition-based trigger in the middle.

How it works: the P-F interval

Every developing failure has a curve. At some point the fault becomes detectable, call it P, the potential failure. Some time later the asset actually stops doing its job, F, the functional failure. The gap between them is the P-F interval, and it is the whole opportunity of condition-based maintenance. Monitor the right condition and you catch the fault at P, which gives you the P-F interval to plan and execute a fix before F ever happens. Too little monitoring and you miss P entirely; the wrong signal and the interval is too short to act. Choosing a condition that gives a long, clear warning is the core design decision, and it is closely tied to how you measure reliability, as covered in reliability-centered maintenance and the MTBF that quantifies it.

What to monitor

The signal has to lead the failure, not just confirm it. The classic condition signals are well established:

Vibration on rotating equipment, motors, pumps, fans and gearboxes, where imbalance, misalignment and bearing wear show up early. See vibration analysis.
Temperature and thermography on electrical connections, bearings and mechanical assets, where heat precedes failure. See infrared thermography.
Oil analysis on gearboxes and hydraulics, and motor current or power draw, which rise as a mechanism loads up or degrades.
Production signals on the line itself: rising stop frequency, lengthening cycle times and creeping micro-stops are early condition indicators that a station is heading for trouble, readable straight from the PLC and from machine vision.

That last category is often overlooked. On a production line the machine's own behaviour is a condition signal, and the micro-stops that quietly grow before a hard failure are exactly the kind of early warning CBM is meant to catch.

Where condition-based maintenance pays, and where it does not

CBM is not free, so it is not for every asset. It earns its keep on critical assets whose failure is expensive, unsafe or unpredictable, and where the fault is genuinely detectable in advance. For low-criticality, redundant or cheap-to-replace assets, the cost of monitoring can outweigh the benefit, and a simple preventive schedule or even run to failure is the smarter call. A criticality assessment decides which assets earn condition monitoring. The preventive maintenance ROI calculator helps you compare the cost of a schedule against the breakdowns it prevents, which is the same trade-off that justifies moving an asset onto condition-based triggers.

Is the schedule worth it?

Compare a maintenance program's cost against the breakdowns it avoids before you invest in monitoring.

Open the PM ROI calculator

The hard part: closing the loop from signal to fix

The barrier to condition-based maintenance is rarely the sensors. It is turning the data into the right action fast enough. A vibration alarm nobody triages, or a line that stops with no recorded cause, becomes noise that trains the team to ignore it, which is worse than no signal at all. What makes CBM actually work is a short, reliable loop: detect the condition, prioritise it against everything else, route a specific work order, do the fix, and confirm it held. Most programs stall not because they cannot detect a fault, but because the fault does not reliably become a scheduled, completed job. That gap between signal and fix is where the value leaks out.

The partner we recommend for closing that loop is Fabrico, because it is built for exactly this problem on production assets. It reads OEE and stops straight from the PLC, uses computer vision to show the true cause of each micro-stop and slow cycle on video, and closes the loop from a detected condition to an auto-routed work order, so a signal becomes a scheduled fix rather than an ignored alarm. It is EU-built with EU data residency and holds ISO 27001, 20000-1 and 9001 (which supports audit-readiness). The tools and guides here stay free either way; Fabrico is what we point to when a team wants condition-based maintenance that actually closes the loop. Book a Fabrico demo to see the true-cause capture on your lines.

FAQ

What is the difference between condition-based and preventive maintenance?

Preventive maintenance acts on a fixed schedule regardless of the asset's state, so it often replaces healthy parts early and can still miss faults that develop between intervals. Condition-based maintenance acts on a measured condition instead and triggers work only when the measurement shows a fault developing, so you intervene when the machine actually needs it.

Is condition-based maintenance the same as predictive maintenance?

Closely related, not identical. Condition-based acts when a current measurement crosses a threshold, telling you a fault is present now. Predictive uses the trend to forecast when failure will occur, so work can be planned further ahead. Predictive adds a time-to-failure estimate on top of the condition-based trigger.

What can you monitor for condition-based maintenance?

Vibration on rotating equipment, temperature and thermography, oil analysis, and motor current are the classics. On production lines the PLC signals and machine vision matter too, because rising stop frequency, lengthening cycle times and creeping micro-stops are early signs of a developing fault. The best signal gives the longest, clearest warning before failure.

Which assets should use condition-based maintenance?

Critical assets whose failure is expensive, unsafe or hard to predict, and where a developing fault is detectable in advance. Low-criticality, redundant or cheap-to-replace assets are often better run to failure or kept on a simple schedule, because monitoring them costs more than it saves. A criticality assessment decides which assets earn it.