Preventive, corrective and predictive maintenance are the three basic types of industrial maintenance, and the difference between them lies in the trigger for the intervention: preventive acts at scheduled intervals, corrective acts after the failure, and predictive acts when the measured condition of the equipment shows that a failure is forming. In practice, a mature plant combines up to six strategies, including planned corrective, detective and prescriptive maintenance, each applied to the assets where it pays off.
Choosing the right type of maintenance for each piece of equipment is one of the decisions that most affect a plant's cost and availability. Erring on one side means breaking down more than you should; erring on the other means spending team hours and parts on interventions the asset did not need. This guide explains the six types of maintenance with industrial examples, compares them all in a master table and presents a practical decision rule, based on three questions: how much the failure costs, how critical the asset is and what data you have today.
What are the types of industrial maintenance
Before comparing the strategies, it is worth understanding what maintenance is and why the area matters for the future of the industry. There are different classifications in the literature, but industrial operations work, in practice, with six strategies:
Unplanned corrective maintenance: the emergency repair, carried out after the breakdown, with no scheduling.
Planned corrective maintenance: the repair decided from an identified defect, but scheduled for the best moment.
Preventive maintenance: interventions at predetermined intervals of time, operating hours or cycles, also called systematic maintenance.
Predictive maintenance: interventions decided by the real condition of the equipment, measured by techniques such as vibration analysis, thermography, ultrasound and oil analysis.
Detective maintenance: periodic tests on protection and safety systems to reveal hidden failures, those that do not appear during normal operation.
Prescriptive maintenance: the evolution of predictive maintenance, in which algorithms not only predict the failure but also recommend the action and the moment to intervene.
This division is consistent with the taxonomy that ABRAMAN (the Brazilian Association of Maintenance and Asset Management) uses in its National Document survey, which tracks how Brazilian companies distribute maintenance man-hours between predictive (by condition), preventive (by time) and corrective (after failure or breakdown). The sections below explain each type, with its trigger, examples and the context in which each one is the right choice.
Corrective maintenance: acting after the failure
Corrective maintenance is the maintenance carried out after a breakdown occurs, intended to return the item to a condition in which it can perform the required function, following the definition of the ABNT NBR 5462 standard. It is the oldest type of maintenance and, contrary to what people often say, it is not always a mistake: the problem is not that corrective maintenance exists, it is that it happens without a decision.
Unplanned corrective (emergency)
It is the breakdown that interrupts production and forces the team to drop whatever they were doing. The real cost goes far beyond the repair: it includes lost production, express freight for the part, overtime and the safety risk of work done under pressure. It is the scenario that planning exists to reduce.
Industrial example: the bearing of the main conveyor motor seizes in the middle of the shift. The line stops, the technician dismantles it under pressure, the part is not in the warehouse and arrives by express freight the next day. The same bearing replacement, done in a scheduled way, would have cost a fraction and no lost production hour.
Planned corrective
Here the defect has already been identified (the equipment lost performance, vibrates more, runs hotter), but it has not stopped yet. The team consciously decides to schedule the repair: it chooses the window, sets aside the material, prepares the work permit. The failure is the same, the cost is much lower, because the intervention happens under maintenance conditions, not under emergency conditions.
There is also a legitimate case of corrective maintenance as a strategy: run to failure. For low-criticality assets, with redundancy and a small repair cost, such as an industrial lamp or a low-value backup pump, accepting the failure and replacing it afterwards can be the cheapest decision. The key point: this must be a documented choice in the plan, not an omission.
To go deeper into the difference between planned and emergency corrective maintenance, with industrial examples and the cost breakdown laid out, read the complete corrective maintenance guide.
Preventive maintenance: acting at scheduled intervals
Preventive maintenance is carried out at predetermined intervals or according to prescribed criteria, with the aim of reducing the probability of failure or the degradation of the item's operation, also according to NBR 5462. It is the maintenance of the calendar and the counter: lubrication every 500 hours, belt replacement every 6 months, monthly roller inspection.
When the frequency is based on time or usage, it is called systematic maintenance. It works well for components with predictable, age-dependent wear: filters, belts, lubricants, seals. Its limitation appears in random failure modes, which do not follow a calendar: in those cases, replacing a part by time does not reduce the probability of failure, it only increases cost and, sometimes, introduces a new defect during reassembly.
Types of preventive maintenance
Within preventive maintenance, industrial practice distinguishes three variations by the criterion that triggers the intervention:
Time-based preventive (calendar): the intervention happens on fixed dates, like the monthly inspection of an overhead crane, regardless of how much the asset ran.
Usage-based preventive (counter): the trigger is accumulated operation, such as the operating hours of a compressor or the cycles of a press. It is fairer to real wear than the pure calendar.
Opportunity-based preventive: it takes advantage of a stoppage that will happen anyway (a product changeover, a general shutdown, the breakdown of another piece of equipment on the line) to bring forward scheduled services and save a production window.
The biggest practical risk of preventive maintenance is not technical, it is managerial: the plan copied from the manual (or from another plant) that nobody reviews. Good preventive maintenance is born from the failure mode of the asset in that operation and is adjusted with the real execution history.
To build the plan in practice, with frequency by equipment family and the indicators that show whether it works, read the industrial preventive maintenance guide.
Predictive maintenance: acting on the measured condition
Predictive maintenance monitors variables of the equipment in operation (vibration, temperature, ultrasonic noise, particles in the oil) to identify the failure as it forms and schedule the intervention before it happens. Instead of asking "how long since the last replacement?", predictive maintenance asks "how is this equipment right now?".
The four most used techniques in industry are vibration analysis, thermography, ultrasound and oil analysis. A typical example: the bearing of a critical motor starts to degrade weeks before seizing, and that degradation appears first in the vibration spectrum and in the temperature. Whoever monitors it can buy the part within the normal lead time, schedule the replacement for the next window and never experience the unplanned stoppage of that motor.
Predictive maintenance reduces both unexpected breakdowns and unnecessary interventions, but it has prerequisites that are usually underestimated: a reliable asset register, a failure history recorded with discipline and a minimally mature preventive program. An expensive sensor on top of bad data only produces alarms without context.
The techniques in detail and the prerequisites that separate a serious program from a collection of alarms are in the complete predictive maintenance guide.
Detective maintenance: testing what stays hidden
Detective maintenance looks for hidden failures: those that do not appear during normal operation because the component is only required in abnormal situations. The classic example is the protection system: the safety pressure switch, the level alarm, the emergency generator, the relief valve. If one of them fails, nobody notices day to day; the failure only reveals itself at the exact moment the device is needed, together with the emergency it was supposed to contain.
That is why detective maintenance works with scheduled functional tests: simulating the abnormal condition in a controlled environment and checking whether the protection acts. It is a mandatory strategy in systems tied to safety and the environment, and it is the least known type of maintenance outside reliability teams.
Prescriptive maintenance: the next step after predictive
Prescriptive maintenance uses algorithms and models (today, increasingly, artificial intelligence) to go beyond prediction: besides indicating that the failure will occur, it recommends what to do, when to intervene and the impact of each alternative. In Brazilian industrial practice it is still a minority and depends on everything predictive maintenance already requires, with an additional layer of large volumes of historical data and integration between systems. It is worth knowing, but it is rarely the next step for those still fighting emergency corrective maintenance.
Comparison table: the 6 types of maintenance side by side
| Type | Intervention trigger | Data it requires | Cost per intervention | When it pays off |
|---|---|---|---|---|
| Unplanned corrective | Breakdown in operation | None | Very high (repair + downtime + urgency) | Never as a plan; it is the scenario to reduce |
| Planned corrective | Identified defect, scheduled repair | Inspection or recorded symptom | Medium | Low-criticality assets or failure tolerable until the window |
| Preventive | Time, hours or cycles | Plan and basic history | Low to medium, recurring | Predictable wear, dependent on age or usage |
| Predictive | Measured condition of the asset | Monitoring + reliable failure history | Low per intervention, initial investment in measurement | Critical assets with measurable failure modes |
| Detective | Scheduled functional test | List of hidden functions and test protocol | Low | Protection and safety systems with hidden failure |
| Prescriptive | Algorithm recommendation | Everything from predictive + large historical base | High investment in technology | Operations with mature predictive maintenance and high cost of downtime |
How to choose: the rule of cost, criticality and available data
No plant uses a single type of maintenance, and there is no "best" type in the abstract. The right choice is per asset (or per family of assets), and three questions solve most cases:
1. How much does this asset's failure cost?
Add up the repair, the lost production, the safety risk and the cascade damage to other equipment. If the failure costs little and does not stop production, planned corrective maintenance (or even run to failure) can be the rational decision. If the failure stops the plant, preventing is almost always cheaper than remedying.
2. How critical is the asset to the operation?
Criticality combines impact on safety, the environment, production and quality. A class A asset (total stoppage, accident risk) deserves predictive maintenance and, in hidden functions, detective maintenance. Class B usually does fine with well-calibrated preventive maintenance. Class C can be an honest candidate for planned corrective maintenance. Without a criticality matrix, the discussion about maintenance types becomes opinion.
3. What data do you have today?
This is the question that enthusiasm for technology usually skips. Predictive maintenance requires a recorded failure history, a correct asset register and reporting discipline. If work orders close without a cause, without a failure mode and days after execution, the first investment is not the sensor: it is making the data reliable in the field. The honest rule: with weak data, start with well-executed preventive maintenance and disciplined recording; with reliable data, evolve critical assets to predictive maintenance.
Crossing the three questions, the decision logic looks like this:
Expensive failure + critical asset + reliable data: predictive maintenance (and detective in protection functions).
Expensive failure + critical asset + weak data: rigorous preventive maintenance now, building the history that enables predictive maintenance later.
Moderate failure + medium criticality: preventive maintenance calibrated by usage, revised with the execution history.
Cheap failure + low criticality + redundancy: planned corrective maintenance, assumed and documented in the plan.
How the types coexist in a real maintenance plan
The result of this rule is never "one strategy for the whole plant": it is a mix. The same production line can have the critical gearbox monitored by vibration analysis, the filters and lubricants under systematic preventive maintenance, the safety pressure switches under a six-month detective test and the low-value redundant assets under assumed planned corrective maintenance. The one who organizes this mix, schedules the work orders and measures the result is PCM (Maintenance Planning and Control).
To define the strategy asset by asset with method, the reference is RCM (Reliability Centered Maintenance), which starts from the failure modes of each piece of equipment to decide which type of maintenance to apply, including the conscious decision to leave certain assets in corrective maintenance.
And it is worth measuring the mix itself: track what share of your work orders is emergency corrective, how much is preventive completed on time and how much already comes from condition monitoring. The healthy evolution is emergency losing ground, month after month, to planned work. It is this number that shows whether the strategy is leaving the paper.
The size of the prize justifies the discipline: according to ABRAMAN's National Document (2011 edition, the last one with this breakdown open to the public), maintenance consumes on average 4.47% of the gross revenue of Brazilian companies, ranging from about 1.4% in food and beverage to 8.67% in mining. In a plant that bills R$ 200 million per year, each percentage point of that bill is worth R$ 2 million: it is the order of magnitude of what is at stake when the mix of strategies is well or poorly chosen.
Frequently Asked Questions
What is the difference between preventive and predictive maintenance?
Preventive maintenance intervenes at fixed intervals of time or usage, regardless of the state of the equipment. Predictive maintenance measures the real condition of the asset (vibration, temperature, oil) and only intervenes when the data indicates a failure forming. Preventive maintenance is simpler and cheaper to implement; predictive maintenance avoids unnecessary replacements, but requires instrumentation and a reliable history.
What is detective maintenance?
It is the strategy that periodically tests hidden functions, those that are only required in abnormal situations, such as alarms, interlocks, relief valves and emergency generators. Since the failure of these items does not appear in normal operation, the scheduled functional test is the only way to ensure the protection will act when needed.
Which type of maintenance is cheapest?
Per intervention, corrective maintenance seems cheap because it has no program cost. In total cost, emergency corrective maintenance is usually the most expensive, because it adds up production downtime, urgency and cascade damage. The truly cheapest is the right mix: preventive and predictive maintenance on critical assets and planned corrective maintenance only where the failure costs little.
What are the types of industrial maintenance?
Six strategies cover industrial practice: unplanned corrective, planned corrective, preventive (systematic), predictive, detective and prescriptive. They differ by the intervention trigger: the breakdown, the identified defect, the calendar, the measured condition, the hidden-function test and the algorithm recommendation, respectively.
Defining the right strategy per asset is half the work; the other half is executing and recording every work order with discipline inside your ERP. Discover PM Run and see how to execute any of these strategies in SAP, with field reporting and data that support the decision.
