The 4 main types of preventive maintenance
Preventive maintenance is scheduled inspection, servicing, and component replacement performed before an asset fails, not after. The goal is lower unplanned downtime, longer asset life, fewer safety incidents, cleaner audit trails, and predictable maintenance spend. For a deeper definition, see our pillar article on what is preventive maintenance.
Preventive maintenance breaks into four canonical types, each defined by the trigger that initiates the work. The right type, or more often the right mix, depends on asset criticality, wear pattern, data availability, and the consequence of failure. The four below are listed in order of data intensity.
1. Time-based preventive maintenance (TBM)
Time-based preventive maintenance (TBM), also called calendar-based maintenance, schedules work at fixed intervals such as monthly, quarterly, or annually, regardless of how much an asset has been used. The trigger is the clock. Typical tasks include lubrication routes, filter changes, calibration cycles, and inspection rounds.
Examples include an HVAC chiller serviced quarterly across a healthcare campus, and a food-grade conveyor lubricated every two weeks on a beverage filling line. TBM is best for assets with predictable wear, safety-critical equipment with manufacturer-prescribed intervals, and regulatory-driven regimes under FDA, OSHA, EPA, or ISO.
TBM lives inside the work order layer, where a calendar engine auto-generates the work order ahead of the due date and pulls the right job plan, spares, and labor capacity.
2. Usage-based preventive maintenance (UBM)
Usage-based preventive maintenance (UBM), also called meter-based maintenance, schedules work based on operating statistics: hours run, cycles completed, miles driven, or units produced. The trigger is the meter, not the clock. Work generates only after the asset has been used to a defined threshold.
Examples include a forklift battery replaced every 1,500 cycles in a distribution warehouse, and an engine oil change every 5,000 operating hours on a fleet vehicle. UBM is best for assets where wear correlates with use: fleet equipment, rotating machinery, conveyors, pumps, and compressors with variable duty cycles.
UBM requires meter ingestion (manual entry, OT integration, or IoT) tied to the work order layer, where defined meter thresholds trigger the work order automatically.
3. Condition-based maintenance (CBM)
Condition-based maintenance (CBM) schedules work based on the current condition of the asset, measured by sensors or inspections. Common indicators include vibration amplitude, temperature, oil quality, ultrasonic readings, and motor current draw. The trigger is a threshold crossing, not a date or a meter reading. The work happens only when the asset shows a degradation signal.
Examples include a vibration sensor on a centrifugal pump triggering a bearing replacement when amplitude exceeds the alarm limit, and an infrared scan flagging a hot electrical connection at a utility substation. CBM is best for high-criticality assets with measurable degradation signals, and where time-based or usage-based intervals either over-service or under-service the asset.
CBM ingests sensor and inspection data into the work order layer, with conditional logic that converts a threshold crossing into a work order without manual intervention.
4. Predictive maintenance (PdM)
Predictive maintenance (PdM) uses condition data combined with analytics or machine learning to forecast when an asset will fail and schedules work just before that point. PdM extends CBM with prediction. The trigger is a forecast of remaining useful life, not a current threshold crossing.
Examples include a turbine remaining-useful-life model on a power-generation asset triggering a bearing replacement two weeks ahead of predicted failure, and an anomaly-detection model on a packaging-line servo flagging a gearbox issue from current-draw patterns. PdM is best for assets with rich sensor data and high failure consequence: mining haul trucks, hospital imaging equipment, aviation ground support, water-treatment pumps.
PdM lives in an analytics layer with model outputs (anomaly score, remaining-useful-life, probability of failure) that feed the work order layer. Ultimo's Predictive maintenance insights are the named capability for this type.
Beyond the four: prescriptive and failure-finding maintenance
Some sources extend the canonical list with two additional types. Prescriptive maintenance (PxM) uses AI to recommend the specific action rather than only predicting failure. Instead of "pump P-204 will fail in 14 days," PxM says "replace bearing on pump P-204 between 02:00 and 04:00 on Sunday using kit BRG-19." It is the bridge into agentic AI. Failure-finding maintenance (FFM) is scheduled inspection of hidden-failure assets such as standby generators and fire-suppression systems that show no symptom until they are called on. Most practitioners treat the four-type list as canonical and these two as extensions.
How to choose the right preventive maintenance type per asset
Choosing the type-mix per asset is a decision that scales with the asset register. A few criteria carry most of the weight.
Criticality: high-criticality assets warrant CBM or PdM, where failure consequence justifies the data investment. Low-criticality assets stay on TBM.
Wear pattern: predictable wear suits TBM. Use-correlated wear suits UBM. Condition-signal wear suits CBM, and forecastable patterns suit PdM.
Data availability: CBM and PdM need sensor or inspection data. TBM and UBM run without it.
Failure consequence: safety, environmental, regulatory, and production consequence pushes the type-mix toward the data-rich end.
Cost of over-servicing: when TBM intervals over-service an asset, CBM saves labor, parts, and downtime.
Reliability-Centered Maintenance (RCM) is the formal methodology for choosing the type-mix per asset, and Failure Mode, Effects, and Criticality Analysis (FMECA) is the risk-ranking input that drives it. Kisuma Chemicals demonstrated the impact: "The FMECA strategy in Ultimo has helped us reduce downtime by 40% and realize considerable cost savings," said Jan Wolf, Reliability Engineer, Kisuma Chemicals.
How EAM software supports the full PM type-mix
Each PM type has its own trigger: calendar, meter, sensor, or model output. When the type-mix lives across spreadsheets, OT systems, and stand-alone tools that only support one or two triggers, schedulers spend the day reconciling instead of preventing failure. The outcome operators look for is a single work order layer that supports all four triggers, one prioritization queue, one mobile-execution surface, and one reporting view.
Ultimo's AI-embedded Enterprise Asset Management (EAM) software supports the full type-mix through named modules:
Work Order Management supports TBM, UBM, CBM, and PdM in one queue, with AI-assisted work order prioritization sequencing the queue when types compete for the same crew.
The Proactive Maintenance module is the named bridge that embeds AI suggestions into the scheduling layer, supporting reliability-centered and risk-based strategies.
The Mobile App executes PM inspections at the point of need, with offline mode, photo capture, and voice input.
Stock Management and Purchasing ties spare parts to scheduled PM jobs. Montanwerke Brixlegg quantified the impact: "Even with just improvements in purchasing alone, we are looking at annual savings of 5% of our purchasing volume," said Silvio Turri, Head of Maintenance, Montanwerke Brixlegg.
Reporting and Dashboards surfaces PM compliance rate, MTBF, MTTR, and downtime via Power BI, with cross-departmental visibility.
The HSE Suite supports work permits and Lockout/Tagout for PM jobs on hazardous assets.
Ultimo software supports 2,500+ organizations, 154,000+ active users, and 22M+ assets globally, deployed at 99.98% average availability on Microsoft Azure, with Certified SAP S/4HANA integration, SOC 2 Type II, ISO certification, and FDA compliance for NA.
How AI is changing preventive maintenance
Asset digitalization creates data volume schedulers cannot process by hand. Workforce retirements accelerate. Capital constraint demands sharper prioritization. Compliance load grows. The outcome operators look for is faster type-mix decisions, tighter PdM forecasts, junior asset professionals who ramp faster, and audit trails that generate themselves.
AI is embedded directly in the PM scheduling and execution layers in Ultimo EAM. The named capabilities map to PM workflows:
Scheduling: AI-assisted work order prioritization sequences the queue when types compete for the same crew.
Forecasting: Predictive maintenance insights produce remaining-useful-life and anomaly-detection outputs for PdM.
Execution: Assisted Troubleshooting offers junior-to-senior decision support when a PM inspection turns up an abnormal condition, and the AI Work Instruction Generator produces tailored step-by-step PM job plans.
Asset register: Automated asset cataloging keeps the hierarchy that underpins all four types clean at scale.
Compliance: Autonomous HSE incident reporting routes safety findings during PM rounds without manual entry.
Beyond preventive: Digital workers, AI-driven capabilities that perform specific tasks alongside human teams, including autonomous triage of PM work orders at higher stages of the Ultimo Maturity Model(/about-us/eam-maturity-model).
Ysco demonstrated the impact of AI-driven workflow improvement across PM execution: technical efficiency improved from 94% to 96%, with hundreds of thousands in annual cost savings. Ultimo was the first EAM vendor to bring agentic AI to industrial maintenance in production. The philosophy behind it is Collaborative Intelligence: human workers, AI, and robotic workers operating as one workforce.
Frequently Asked Questions
What are the 4 types of preventive maintenance?
The four types of preventive maintenance are time-based (TBM), usage-based (UBM), condition-based (CBM), and predictive (PdM). TBM is triggered by the calendar. UBM is triggered by a meter such as cycles, hours, or miles. CBM is triggered by a condition signal crossing a threshold. PdM is triggered by an analytics or machine-learning forecast. Most operators run a hybrid mix per asset.
What are the 5 types of preventive maintenance?
Some sources extend the four canonical types with one of two additions. Prescriptive maintenance (PxM) uses AI to recommend the specific action, including timing, spare, and labor, not only predicting failure. Failure-finding maintenance (FFM) is scheduled inspection of hidden-failure assets such as standby generators. The AI Overview convention is four types, with PxM and FFM as extensions.
What are the 7 elements of preventive maintenance?
The seven elements typically referenced in a preventive maintenance program are an asset register and equipment hierarchy, a PM schedule, work orders, spare-parts and inventory management, inspections and checklists, KPIs and reporting (MTBF, MTTR, PM compliance), and a compliance audit trail. Together they form the operating model that turns scheduled work into measurable reliability outcomes.
What is the difference between preventive and predictive maintenance?
Preventive maintenance is the umbrella for scheduled work before failure, triggered by a calendar, a meter, a condition, or a forecast. Predictive maintenance (PdM) is the data-driven sub-type that uses analytics or machine learning to forecast failure timing and schedule work just-in-time. All predictive maintenance is preventive; not all preventive maintenance is predictive.
How does TPM relate to preventive maintenance?
Total Productive Maintenance (TPM) is an operator-led maintenance philosophy that builds preventive maintenance into daily operations, with operators owning routine PM tasks alongside the maintenance team. TPM is a culture and operating model anchored in pillars including autonomous and planned maintenance. Preventive maintenance is one of the strategies it relies on.