Predictive vs Preventive Maintenance: Practical Guide
Preventive and predictive maintenance are not competing technologies with one universal winner. The better strategy depends on asset criticality, failure behaviour, detectability, operating context and whether the information gained can support a useful maintenance decision.
Maintenance strategy should follow the equipment risk
The question is not whether predictive maintenance is more modern than preventive maintenance. The useful question is which maintenance policy controls the relevant failure mode at an acceptable technical and economic cost.
Preventive maintenance remains appropriate where servicing or replacement intervals are supported by manufacturer instructions, regulation, operating experience or a predictable age- or usage-related failure mechanism.
Condition monitoring becomes more useful where deterioration can be measured before functional failure. Depending on the machine and failure mode, monitoring can involve vibration, temperature, lubricant condition, electrical parameters, process performance or other physical indicators.
Predictive maintenance goes further when collected information is used to estimate future degradation or failure risk. It does not necessarily require artificial intelligence, cloud platforms or a digital twin. Sophisticated analytics may help, but useful predictive decisions can also come from well-understood condition trends and engineering thresholds.
For industrial maintenance materials and recurring spare-parts requirements, see our cornerstone page Supplier MRO Karawang. For maintenance instruments, tools and engineering equipment, see Supplier of Engineering Equipment & Tools.
Preventive, condition-based and predictive are different concepts
Maintenance terminology is frequently blurred in marketing material. A vibration sensor attached to a motor does not automatically mean the plant has implemented predictive maintenance.
A condition-monitoring system first provides information about equipment condition. The maintenance strategy depends on how that information is interpreted and used.
ISO 17359:2018 provides general procedures for establishing a condition-monitoring programme for machinery and remains the current ISO edition.
Choose the method from the failure mode—not from the maintenance trend
| Situation | Possible maintenance approach | Reasoning to verify |
|---|---|---|
| Known scheduled service requirement | Preventive maintenance | Manufacturer, regulation or engineering program defines a justified service interval. |
| Consumable degrades mainly with hours or cycles | Preventive or usage-based maintenance | Age or usage is a useful predictor of the required intervention. |
| Rotating equipment develops measurable deterioration | Condition monitoring / CBM | A meaningful parameter such as vibration or temperature changes before functional failure. |
| Failure develops gradually and sufficient history exists | Predictive maintenance may be useful | Condition trends provide enough warning to schedule a meaningful action. |
| Failure is sudden with no detectable precursor | PdM may provide little benefit | If the relevant failure mode cannot be detected in advance, sensing may not solve the problem. |
| Cheap, non-critical, easily replaced component | Run-to-failure may be rational | Monitoring and planned intervention could cost more than the consequence of failure. |
| Safety-critical or legally mandated maintenance | Follow prescribed inspection / maintenance program | Predictive information should not automatically replace mandatory activities. |
Measure a parameter that is connected to the failure mode
Collecting more data is not the objective. The objective is to obtain information that changes a maintenance decision.
Vibration
Often used on rotating machinery where imbalance, looseness, bearing condition or other mechanical behaviour can influence vibration characteristics.
Temperature
Trending temperature or thermography can help identify abnormal heat associated with selected electrical or mechanical problems.
Lubricant condition
Oil analysis can provide information about lubricant condition, contamination and selected wear indicators where sampling is appropriate.
Electrical parameters
Current, voltage, insulation-related measurements and other electrical data may support diagnosis for suitable equipment and failure modes.
Process performance
Pressure, flow, power, efficiency, cycle time or product quality may reveal deterioration that is not obvious from a standalone sensor.
Visual inspection
Leaks, corrosion, wear, contamination, loose parts and abnormal noise remain valuable observations even in highly instrumented plants.
ISO's condition-monitoring committee describes the discipline as periodically or continuously sensing and recording selected physical parameters associated with operating machinery so the information can support machine operation and maintenance decisions.
Predictive maintenance needs a business case
Condition-monitoring technology can improve maintenance decision-making, but its benefits should be evaluated rather than assumed.
A 2025 NIST systematic review of condition-monitoring technologies notes that organisations need to assess both engineering and financial benefits to determine whether adoption is justified.
The economic calculation can include sensor and installation cost, software, connectivity, data storage, analysis effort, training, calibration and ongoing maintenance of the monitoring system itself.
Those costs can then be compared with the consequences the program is intended to avoid or reduce: emergency downtime, secondary damage, overtime, expedited freight, lost production and unnecessary planned interventions.
Maintenance strategy should influence spare-parts planning
Maintenance and inventory decisions are connected. A predictive alert is of limited operational value if the required replacement part has a four-month lead time and procurement only starts after the machine reaches an alarm threshold.
Critical spares
Identify components whose absence can materially extend downtime after a detected fault.
Lead time
Compare expected warning time with procurement lead time. A useful prediction should create enough time for an actionable response.
Parts identity
Keep manufacturer, part number, specifications and approved alternatives clean so an alarm can become an accurate RFQ quickly.
Planned kits
Where maintenance work is predictable, prepare related seals, fasteners, lubricants and other necessary items together with the primary spare.
Supplier route
Know whether a critical part is local stock, distributor stock, import order or manufacturer-built before the equipment deteriorates.
Failure history
Use work orders and consumption history to refine both maintenance intervals and spare-parts policy.
Do not start by installing sensors on every machine
Rank the assets
Identify equipment whose failure materially affects safety, production, quality, environment or repair cost.
Define the failure modes
Determine how important components fail and which failures are currently addressed by preventive, reactive or inspection-based maintenance.
Identify detectable indicators
Establish whether vibration, temperature, lubricant, electrical or process data can provide meaningful warning for the selected failure mode.
Review existing data first
Use work orders, operator observations, breakdown records and available machine data before assuming new instrumentation is required.
Pilot a limited scope
Test the monitoring method on selected assets with clear failure modes and measurable business consequences.
Define alarm-to-action workflow
Specify who receives the alarm, who validates it, what additional inspection is required and who authorises the maintenance work.
Review the results
Compare detected faults, false alarms, avoided failures, maintenance changes and program costs before expanding the system.
What maintenance technology actually needs to be purchased?
| Requirement | Information to specify | Avoid |
|---|---|---|
| Vibration instrument / sensor | Machine type, frequency range, measurement points, output, mounting, environment and monitoring objective. | Requesting only “vibration sensor for motor”. |
| Thermal imager | Temperature range, detector resolution, measurement accuracy, viewing environment and reporting requirement. | Selecting from image resolution or price alone. |
| Temperature sensor | Range, accuracy, probe type, process connection, output and operating environment. | Assuming every temperature sensor is interchangeable. |
| Oil-analysis supplies | Lubricant, sampling point, sample bottle requirement, test program and laboratory procedure. | Taking uncontrolled samples that cannot be compared over time. |
| Online monitoring system | Number of assets, channels, sensor types, communications, cybersecurity, historian and alarm workflow. | Buying a platform before defining what maintenance decisions it must support. |
| Replacement spare | Manufacturer, part number, equipment reference, dimensions, rating and approved alternative policy. | Using the monitoring alarm itself as the part description. |
What to send when an inspection identifies a maintenance need
Equipment
Machine manufacturer, model, asset number and exact installation location.
Part identity
Manufacturer, part number, dimensions, rating and photographs of the existing component.
Observed condition
Describe the finding that triggered the requirement without asking the supplier to diagnose unsupported failure causes.
Documentation
Attach datasheet, drawing, nameplate, BOM or manual information where available.
Timing
State whether the work is planned, shutdown-related or urgent and provide the required delivery date.
Alternatives
Clarify whether equivalent products may be proposed and who must approve a substitution.
For recurring maintenance materials, bearings, seals, electrical components and other industrial spares, review our MRO supplier services in Karawang. For measurement instruments and maintenance tools, see our engineering equipment and tools supply.
Frequently asked questions
Is predictive maintenance always better than preventive maintenance?
No. The appropriate method depends on failure mode, asset criticality, detectability, maintenance requirements and economics.
Does predictive maintenance require AI?
No. AI and machine learning can support some applications, but predictive decisions can also use condition trends, established diagnostic techniques and engineering models.
Does preventive maintenance always replace parts too early?
No. A correctly designed preventive task may be appropriate where deterioration is related to age, operating hours, cycles or a prescribed servicing interval.
Should every critical machine have online sensors?
Not automatically. First determine whether the relevant failure modes produce measurable precursors and whether online monitoring offers enough advantage over periodic inspection or other maintenance methods.
Can preventive and predictive maintenance be combined?
Yes. Most industrial facilities use several maintenance strategies across different assets and components.
What is the best first asset for a predictive-maintenance pilot?
A useful candidate usually combines meaningful failure consequences with a detectable deterioration mechanism, sufficient warning time and a clear maintenance action after detection.
Condition-monitoring guidance
- ISO 17359:2018 — Condition Monitoring and Diagnostics of Machines — current ISO general guidelines for establishing a machinery condition-monitoring programme.
- ISO/TC 108/SC 5 — Condition Monitoring and Diagnostics of Machine Systems — overview of the role of measured machine parameters in supporting operation and maintenance decisions.
- NIST — Comprehensive Evaluations of Condition Monitoring-Based Technologies in Industrial Maintenance — 2025 systematic review examining engineering and financial evaluation of condition-monitoring technologies.
- NIST — Asset Condition Management — framework relating condition awareness, diagnostics and estimates of future asset health to manufacturing maintenance decisions.
Editorial note: this article provides general maintenance and MRO procurement guidance. Maintenance intervals, condition limits, alarm thresholds and equipment interventions should be established from manufacturer instructions, engineering analysis, equipment criticality, operating experience and applicable safety or regulatory requirements.
