About condition-based maintenance (CBM)
Condition-based maintenance (CBM) is a maintenance strategy that schedules work according to an asset's measured condition, not the calendar or a usage meter. Work is triggered when a monitored indicator, such as vibration, temperature, or oil quality, crosses a defined limit. The result is fewer unnecessary tasks on healthy assets and earlier action on assets that are starting to degrade.
This guide covers how CBM works, what triggers the work, and how it differs from time-based and predictive maintenance. It also covers when CBM is worth the investment and the step programs often underestimate: turning a condition signal into executed, recorded work. CBM is one of the types of preventive maintenance. Ultimo brings decades of Enterprise Asset Management (EAM) domain expertise to the work that follows the signal.
How does condition-based maintenance work?
Condition-based maintenance works as a loop: measure the asset's condition, compare it with a limit, and act only when the limit is crossed. The loop has five steps.
Monitor. Sensors, inspections, or operator readings capture an indicator of asset health, either continuously or at set intervals.
Compare. The reading is checked against a fixed alarm level or against the asset's own baseline.
Diagnose. When a limit is crossed, a reliability or maintenance professional identifies the likely failure mode and its urgency.
Act. The work is planned, staffed, supplied with parts, and executed before the asset fails.
Learn. The finding is recorded in the asset's history and used to refine the limit and the response.
Many sources describe the same flow as collect, analyze, and act. The fifth step is easy to skip. Without it, the limits never improve and the same degradation is rediscovered each time. In an EAM system, step 4 becomes a work order and step 5 becomes part of the asset's history.
What triggers condition-based maintenance work?
CBM work starts when a measured value passes a set limit. That limit can be a fixed alarm level from manufacturer guidance, or a deviation from the asset's own normal readings. Three illustrative examples show how this works. The values shown are round placeholders that cite no standard. Real limits come from manufacturer guidance and the asset's own history.
Conveyor motor current draw. A motor drawing steadily more current than normal can point to rising friction or a failing component. If draw reaches 15% above its baseline, the action is an inspection at the next planned stop.
Gearbox oil condition. Oil analysis tracks particle count and viscosity. If viscosity drifts 10% from the oil's baseline, that signals contamination or wear, and the action is a filter change, an oil change, or a deeper inspection.
Compressed-air leak detection. An ultrasonic scan can locate a leak on a compressed-air line that is hard to hear in a noisy plant. If a reading is 10 dB above background noise at a fitting, the action is a repair at the next planned stop.
In each case the work is prompted by evidence, not by elapsed time. The same motor could run well past its calendar-based service date without a problem. It could also degrade early, and a fixed schedule would miss it.
How is CBM different from time-based, preventive, and predictive maintenance?
Condition-based maintenance acts on an asset's present condition, while predictive maintenance forecasts how long the asset has left. They draw on similar data but answer different questions. CBM asks whether the asset has crossed a limit now. Predictive maintenance asks when it will.
The main strategies differ by what starts the work:
Reactive maintenance: the work starts after the asset fails.
Time-based maintenance (TBM): the work starts on a calendar interval, regardless of asset condition.
Usage-based maintenance: the work starts when a meter, such as hours or cycles, reaches a set value.
Condition-based maintenance (CBM): the work starts when measured condition crosses a limit.
Predictive maintenance (PdM): the work starts when an analytical forecast of remaining life says it is time.
Searchers often ask "what is CBM and TBM?" The two are alternatives. TBM trusts the calendar, and CBM trusts the evidence. Many sites combine both, assigning each asset the strategy its failure behavior and criticality justify.
CBM sits inside the broader family of preventive maintenance. Sources draw the boundary with predictive maintenance in different places, and some treat predictive as an extension of CBM. This article treats them as separate because predictive adds a forecasting step that plain CBM does not.
Which condition monitoring techniques support CBM?
Condition monitoring, sometimes called condition-based monitoring, is the data-collection practice that CBM relies on. CBM is the strategy that acts on the data. Several techniques feed a CBM program, and each detects a different kind of degradation.
Vibration analysis:** detects imbalance, misalignment, and bearing wear in rotating equipment.
Oil analysis: detects contamination, wear particles, and lubricant breakdown.
Thermography: detects abnormal heat in electrical panels, connections, and mechanical components.
Ultrasonic testing: detects leaks, friction, and early bearing faults that are hard to hear or see.
Motor current analysis: detects electrical and mechanical faults in motors from their current signature.
Visual inspection and operator rounds: detect leaks, corrosion, noise, and wear, and need no sensors at all.
CBM does not require a sensor network. A structured visual inspection with measurable readings can trigger work on a condition basis, which makes it a practical starting point for many teams.
When is CBM worth it?
Condition-based maintenance pays off on assets where failure is costly, degradation is detectable, and the warning arrives early enough to act. Where those conditions do not hold, a simpler strategy usually wins. Four questions work as a decision test:
What does a failure cost? Consider lost output, safety exposure, repair cost, and downstream delays. High consequences justify monitoring.
Does the asset degrade in a measurable way? CBM needs an indicator that changes before failure. Some failure modes give no warning.
Is the warning interval long enough? The time between the first detectable sign and functional failure, often called the P-F interval, must exceed the time needed to plan and execute the work.
Can the work be scheduled in time? If parts, labor, or a production window cannot be arranged within that interval, an early warning has little value.
Siemens' 2024 analysis put unplanned downtime at about $1.4 trillion a year across the world's 500 largest companies. That scale is why critical assets earn monitoring. It does not mean every asset does. Criticality rankings and failure history kept in an EAM system give the first two questions a factual starting point.
The counter-case is just as important. Low-criticality assets, cheap-to-replace parts, and failure modes that arrive randomly are often better served by run-to-failure or time-based maintenance. Monitoring them can cost more than the failures it would prevent.
When the test is met, the benefits are specific:
Fewer tasks performed on assets that are still healthy.
Fewer unplanned stops, because degradation is caught earlier.
Better use of labor and spare parts, since work is planned against evidence.
A growing body of condition data that improves later decisions.
How does a CBM program start?
A CBM program starts with a short list of critical assets, not the whole asset base. Starting narrow keeps the data volume, the alert volume, and the response workload manageable for the team. The usual sequence has five steps.
Rank assets by criticality. Focus on assets where downtime is most expensive or most dangerous.
Choose the failure modes worth monitoring. Methods such as FMECA and reliability-centered maintenance (RCM) help decide which failure modes are detectable and worth the effort.
Define the measurement and the limit. Pick the indicator, how often to read it, and the alarm level or baseline deviation that triggers action.
Set the response procedure. Decide who reviews an alert, how fast, and what work follows each type of finding.
Expand deliberately. Add assets and techniques as the team proves it can respond to what it already detects.
Thin asset data and limited capacity are the usual objections. The answer is to match scope to maturity. Ultimo's Progressive Intelligence approach lets teams adopt intelligence at their own pace, with no big-bang change. The Ultimo EAM Maturity Model places condition-based insight and failure prevention at its Proactive stage.
Where does condition-based maintenance break down?
CBM breaks down at the handoff between a condition signal and executed work. An alert that nobody turns into a planned, parts-ready, assigned, and recorded job is just another notification. Condition data that never reaches the asset's history cannot refine the limit either.
The consequence is that the signal-to-work step decides whether failures are actually prevented. Detection alone does not prevent anything. When the handoff fails, the result is unplanned stops, missing parts, and repeat failures on the same asset. Someone has to diagnose the finding, find the right parts, schedule the crew, complete the job, and capture what was found.
This is where Ultimo's EAM software with AI embedded supports the program. Teams comparing condition-based maintenance software should ask what happens after the alert. Ultimo is Intelligent Asset Management software that supports the work following a condition finding:
Work Order Management supports the planning, prioritization, execution, and tracking of the full work order process, so a finding can be planned, assigned, and followed through to completion.
AI-generated inspection lines help planners turn a flat work instruction into structured checks and measurements, so condition readings become data that can be analyzed.
Photo-based meter readings let a worker photograph a gauge, with AI extracting the value and the worker verifying it before submission.
Assisted Troubleshooting surfaces relevant work order and asset information and failure history at the point of diagnosis, which helps the diagnose step.
Completed activities assistant suggests structured wording for finished work, and the worker confirms it, so the finding becomes part of the asset's failure history.
RCA follow-up actions generator turns root cause analysis into actionable follow-up items, so lessons do not stay in a report.
Embedded AI suggests, and the human confirms. The asset professional stays in control of every decision.
Frequently Asked Questions
What is condition-based maintenance?
Condition-based maintenance (CBM) is a maintenance strategy that triggers work when an asset's measured condition crosses a defined limit. The condition might be vibration, temperature, oil quality, or a visual reading. Work happens when the evidence calls for it, not on a fixed calendar. This reduces unnecessary tasks and catches degradation earlier.
What is the difference between CBM and time-based maintenance (TBM)?
Condition-based maintenance starts work when measured condition crosses a limit. Time-based maintenance starts work at a fixed calendar interval, whatever the asset's condition. TBM is simple to schedule but can over-service healthy assets or miss early failures. CBM targets work to actual need but requires a way to measure condition.
What is the difference between condition-based and predictive maintenance?
Condition-based maintenance acts when an asset's current condition crosses a limit. Predictive maintenance forecasts remaining life or time to failure and plans work against that forecast. Both use condition data. CBM answers whether a limit has been crossed now, and predictive answers when it is likely to be crossed.
Is condition-based maintenance the same as condition monitoring?
No. Condition monitoring is the practice of collecting and reading data on asset health, such as vibration or temperature. Condition-based maintenance is the strategy that decides when to act on that data. Monitoring supplies the evidence, and CBM turns the evidence into scheduled maintenance work.
What are the four types of maintenance?
The four types most often cited are reactive, preventive, condition-based, and predictive maintenance. Reactive maintenance fixes an asset after it fails. Preventive maintenance follows a calendar or usage interval. Condition-based maintenance acts on measured condition, and predictive maintenance acts on a forecast. This guide, like many sources, treats condition-based maintenance as part of preventive maintenance and keeps predictive as its own strategy.
What are examples of condition-based maintenance?
Common examples include servicing a conveyor motor when its current draw climbs above baseline and changing gearbox oil when oil analysis shows contamination. Another is repairing a compressed-air leak after an ultrasonic scan locates it. Each example starts work from evidence, not from a date.