Types of maintenance compared: preventive, predictive, condition based, corrective and run to failure
The EN 13306 tree that sorts every kind of maintenance into preventive or corrective, each type against the others on what triggers it and where it fits, what separates predictive from condition based from condition monitoring, and how the type is chosen from what the failure looks like rather than from preference.
Key takeaways
- Every kind of maintenance is either preventive, done before a failure to reduce its chance or consequence, or corrective, done after a fault to restore the function. EN 13306 draws the tree, and the names on it are the ones worth using.
- Preventive maintenance splits into predetermined, done at a set interval without checking condition, and condition based, done when a measured or observed condition says so. Predictive maintenance is condition based maintenance with a forecast.
- Corrective maintenance splits into deferred, planned in because the fault can wait, and immediate, done at once because it cannot. Run to failure is a deliberate decision to accept corrective maintenance for an item, not an absence of a decision.
- The right type is decided by the failure, not by preference: whether it gives a warning, whether it has a wear-out age, whether it is hidden, and what it costs when it happens.
- A plant does not choose one type. It chooses, for each failure mode that matters, the type that fits it, and the mix that results is the maintenance strategy.
Ask three people on a site what predictive maintenance is and two of them will describe condition based maintenance and the third will describe a software product. Ask whether a plant does preventive maintenance and the answer is yes, meaning time-based inspections, with no mention of the condition monitoring program that is also preventive maintenance. The words are used loosely, and loose words make for loose strategy, because a task cannot be chosen for a failure if nobody agrees what kind of task it is.
There is a standard set of terms, in EN 13306, and it is a tree with two branches and a few leaves. This guide draws the tree, sets each type against the others on what triggers it, where it fits and where it does not, and then turns the question around: given what a failure looks like, which type fits it. It sits beside the preventive maintenance program guide, which covers building the program once the types are chosen, and the failure patterns piece, which is the evidence behind most of the choosing.
The tree
At the root is the split that matters most: whether the work is done before a failure or after one. Preventive maintenance is done before, to reduce the probability of the failure or the consequence of it. Corrective maintenance is done after a fault has been detected, to restore the function. Everything else hangs from one of those two.
Preventive maintenance splits on what triggers it. Predetermined maintenance is triggered by elapsed time, running hours, cycles or throughput, without checking the condition first: the oil change at 4,000 hours, the liner change at a tonnage. Condition based maintenance is triggered by the condition itself, measured or observed, crossing a limit. Predictive maintenance is a kind of condition based maintenance in which the condition is trended and the failure forecast, so that the task can be timed. Failure finding sits with predetermined maintenance, at an interval, but its purpose is different: to find out whether a hidden function has already failed.
Corrective maintenance splits on urgency. Deferred corrective maintenance is a fault found and planned into the schedule, because it can safely wait: the weeping gland fixed at the next stop. Immediate corrective maintenance is the fault that cannot wait, which is the breakdown. Run to failure is a decision, taken in advance for a particular item, to accept corrective maintenance as the strategy.
Maintenance
Preventive
Before a failure, to reduce the chance of it or the consequence
Predetermined
At set intervals, without checking condition
Includes failure finding
Condition based
On a measured or observed condition, including a forecast from it
Includes predictive
Corrective
After a fault is found, to restore the function
Deferred
Planned in, because it can wait
Run to failure lands here, by decision
Immediate
Done at once, because it cannot
The breakdown
Terms after EN 13306. The branch is decided by the failure, not by the budget
Each type against the others
The table takes the seven types a plant actually uses and sets each against the others on what triggers it, where it fits, where it does not and an example from the crusher branch used across this site. The 'does not fit' column is the useful one. Every type has a kind of failure it is wrong for, and most maintenance waste is a type applied to a failure it does not fit: time-based replacement of items with no wear-out age, condition monitoring of items that give no warning, and run to failure on items whose failure was never costed.
| Type | Branch | Triggered by | Fits | Does not fit | Example |
|---|---|---|---|---|---|
| Run to failure | Corrective | The failure itself, by decision taken in advance | Low consequence, low cost, no warning worth looking for, or where the item is cheaper to replace than to inspect | Anything whose failure hurts safety, production or a neighbouring asset | A light fitting in a warehouse, an idler on a low duty conveyor with a spare on the shelf |
| Corrective, deferred | Corrective | A fault found, not yet a functional failure, planned into the schedule | A defect found on a round or an inspection that can safely wait for a planned job | A fault that is getting worse faster than the schedule moves | A weeping gland on a slurry pump, replaced at the next planned stop |
| Corrective, immediate | Corrective | A functional failure that cannot wait | The unplanned work every plant has; the aim is to make it rare and to learn from each one | As a strategy. A plant that runs on breakdowns runs on overtime and expedited parts | A crusher drive motor trips on winding temperature and is replaced that shift |
| Predetermined | Preventive | Time, running hours, cycles or throughput, set in advance | Items with a known wear-out age: belts, filters, lubricants, seals, wear liners under a steady duty | Items with no wear-out age, which is most electronics and most complex assemblies. Replacing them on a schedule reintroduces infant failures | Gearbox oil changed every 4,000 running hours; crusher liners turned at a set tonnage |
| Condition based | Preventive | A measured or observed condition crossing a limit | Items that give a warning that can be detected, with enough interval between the warning and the failure to act | Failures with no detectable warning, or a warning too close to the failure to plan around | A bearing replaced when its vibration crosses the alert level set from its history |
| Predictive | Preventive | A forecast from the condition trend, not a single reading | Condition based maintenance with enough history to extrapolate, on items where the timing of the replacement matters | Items with a short history, erratic data or a failure mode that does not trend | Belt wear measured monthly, with the replacement scheduled for the stop before the forecast limit |
| Failure finding | Preventive | An interval set from the acceptable probability that a hidden function has already failed | Protective devices that only act when something else has gone wrong: trips, relief valves, standby pumps, alarms | Anything whose failure is evident to the operators in normal running | A conveyor pull-wire tested each month; a standby pump run each week |
The branch is decided by the failure, not by the budget, and not by what the plant has always done.
Predictive, condition based and condition monitoring
The three terms are the ones most tangled. Condition monitoring is the measuring: the vibration reading, the oil sample, the thermal image, the operator's hand on the bearing housing. Condition based maintenance is the maintenance decision that follows from it: a task triggered when the condition crosses a limit. Predictive maintenance is condition based maintenance with the extra step of forecasting, from the trend, when the limit will be reached, so the work can be scheduled for the stop before it. Monitoring without a limit and a task is data. A limit without a trend is condition based. A trend with a forecast is predictive.
The forecasting step needs history, and the history takes time to build. A site starting condition monitoring on a new asset is doing condition based maintenance for the first year or two whatever the software is called. The condition monitoring guide covers the measurement side and the software side covers getting the readings to the people who act on them.
Choosing by what the failure looks like
The choice is made per failure mode, and the failure mode decides it. Four questions sort most cases. Does the failure give a warning that can be detected, and is the interval between the warning and the failure long enough to act? Then condition based maintenance, predictive once the history supports it. Does the item wear out at a predictable age under a steady duty? Then predetermined maintenance, at an interval set from the data rather than the manual. Is the function hidden, so nobody will notice it has failed until it is needed? Then failure finding. Does the failure matter at all? If not, run to failure, decided and written down.
The evidence behind the first two questions is in the failure patterns: most items in the airline studies showed no wear-out age, which is why predetermined replacement of complex items reintroduces the infant failures it was meant to prevent. The structured way to ask the questions for the equipment that matters is reliability centred maintenance or an FMECA, after a criticality assessment has said which equipment earns the analysis.
| The failure | The type that fits | Typical examples |
|---|---|---|
| The failure has a detectable warning and enough P-F interval to act | Condition based, predictive once the history supports it | Vibration, temperature, wear, leakage, current, performance |
| The item wears out at a predictable age under a steady duty | Predetermined, at an interval set from the data, not the manual | Belts, liners, lubricants, filters, seals |
| The item fails at random with no warning, and the consequence matters | Design out the failure mode, or add protection and a failure-finding task on it | Electronics, instruments, software |
| The function is hidden: nobody notices it has failed until it is needed | Failure finding at an interval set from the acceptable risk | Trips, interlocks, relief valves, standby equipment |
| The failure has no safety, production or cost consequence worth the task | Run to failure, decided and recorded | Lighting, low duty idlers, redundant small items |
The mix is the strategy
No plant runs on one type. A crusher has predetermined liner changes at a tonnage, condition based bearing replacement on vibration, failure finding on its trips and interlocks, deferred corrective work from the operator rounds, and run to failure on its lighting. The mix across the plant, with the reasons for each choice recorded against the failure mode, is the maintenance strategy, and the maintenance plan in the CMMS is that strategy turned into tasks with intervals and triggers.
What the mix should look like in proportion is a question the measures answer. The planned maintenance percentage says how much of the work was planned before it was done, whichever branch it sits on. A rising share of immediate corrective work says the preventive side is missing failures, and the root cause analyses of those failures say which type was applied to a failure it did not fit.
Where the terms go wrong
- Preventive maintenance used to mean only time-based inspections, so the condition monitoring program is not counted as preventive and nobody owns the gap between them.
- Predictive maintenance used to mean any condition monitoring, or any software, so the forecasting that makes it predictive is never actually done.
- Reactive maintenance used as a type rather than a symptom, so a plant plans to be reactive instead of asking which failures the preventive side is missing.
- Run to failure that nobody decided, which is neglect with a name.
- Predetermined intervals taken from the manual rather than from the plant's own failure data, so the task is done too often or too late.
- Failure finding left out altogether, so the trips and the standby pumps are assumed to work until the day they are needed.
- Deferred corrective work not distinguished from immediate, so the measures cannot tell a plant that finds faults early from one that waits for them.
Common questions
What are the types of maintenance?
At the top, two: preventive, carried out before a failure to reduce its probability or its consequence, and corrective, carried out after a fault is detected to restore the function. Preventive maintenance is either predetermined, at set intervals of time or use, or condition based, triggered by a measured or observed condition. Corrective maintenance is either deferred, planned into the schedule, or immediate, done at once. Predictive maintenance, failure finding and run to failure all fit inside those branches. EN 13306 is the terminology standard that draws the tree.
What is the difference between preventive and predictive maintenance?
Preventive maintenance is the branch and predictive maintenance is a leaf on it. Preventive maintenance is anything done before a failure to reduce its chance or consequence, whether at a set interval or on condition. Predictive maintenance is condition based maintenance where the condition is trended and the failure is forecast, so the task can be timed. All predictive maintenance is preventive. Not all preventive maintenance is predictive: a time-based oil change is preventive and not predictive.
What is the difference between corrective and reactive maintenance?
Corrective maintenance is the standard's term for work done after a fault is found to restore the function, and it includes both deferred work that was planned in and immediate work that could not wait. Reactive maintenance is the common name for the immediate kind, and often for a plant whose maintenance is mostly that kind. Breakdown maintenance means the same. A plant can have a high proportion of corrective work that is planned and deferred, which is healthy, or of immediate work, which is not.
What is run to failure?
A deliberate decision to do no preventive maintenance on an item and to repair or replace it when it fails, taken because the failure has no safety, production or cost consequence worth the task, or because no task would reduce the chance of it. It is corrective maintenance chosen in advance and recorded. The difference between run to failure and neglect is that the decision was made, with the consequence understood, and the spare is on the shelf.
What is condition based maintenance?
Preventive maintenance triggered by the condition of the item rather than by elapsed time or use: a reading, an inspection, a performance measure crossing a limit set in advance. It fits items that give a detectable warning with enough interval between the warning and the failure to plan the work. The condition can be measured by an instrument, read from the control system or observed on an operator's round. Vibration, temperature, oil condition, wear and performance are the common ones.
Which type of maintenance is best?
None of them, for a plant. Each fits a kind of failure. Predetermined maintenance fits items with a wear-out age under a steady duty. Condition based maintenance fits items that give a warning. Failure finding fits hidden functions. Run to failure fits items whose failure does not matter. The question is asked per failure mode, usually through a criticality assessment and a reliability centred maintenance or FMECA analysis for the critical equipment, and the mix that results is the plant's maintenance strategy.
Key terms
Plain-language definitions from our glossary for the concepts this article leans on.
Standards and further reading
- BS EN 13306:2017 Maintenance: maintenance terminology (BSI)
- IEC 60300-3-11:2009 Dependability management, application guide: reliability centred maintenance (IEC)
- SAE JA1011 Evaluation criteria for reliability-centered maintenance (RCM) processes (SAE International)
- ISO 14224:2016 Collection and exchange of reliability and maintenance data for equipment (ISO)
- ISO 17359:2018 Condition monitoring and diagnostics of machines: general guidelines (ISO)
- Best practices, metrics and guidelines for maintenance and reliability (SMRP)
Related case studies and tools
- Reliability engineering and asset strategy program (case study)
Related reading
Working through something like this?
See how we approach these initiatives, or tell us what you are dealing with.