A maintenance concept is the reasoned mix of maintenance types for an asset or group of assets: which failure modes you repair after failure, which you tackle at a fixed interval, which you measure and which you remove with a modification. It links your organisation's maintenance strategy to the maintenance plan with its concrete tasks. A good maintenance concept directs the budget at the risks that matter and makes more of the work plannable.
What a maintenance concept is, and how it differs from a strategy and a plan
Researchers at KU Leuven describe the maintenance concept as the mix of maintenance interventions (corrective, preventive, condition-based, opportunistic or design-based) plus the framework you use to decide on that mix. In other words, the concept says which type of maintenance goes with which failure mode and why. Some organisations also include indicative intervals; the worked-out tasks go in the plan.
Strategy, concept and plan often get mixed up. EN 13306, the European maintenance terminology standard, calls a strategy the management method used to achieve the maintenance objectives, and a plan the structured set of tasks with its activities, procedures, resources and timescale. The concept sits between the two, and the schedule comes after them:
| Term | Level | Question | Example |
|---|---|---|---|
| Maintenance strategy | Organisation | Which objectives and which approach? | Availability first, specialist work outsourced |
| Maintenance concept | Asset or asset group | Which maintenance type per failure mode? | Bearings on condition, oil on running hours, rollers after failure |
| Maintenance plan | Task | What do you do and how often? | Measure vibration on the drum bearings every month |
| Maintenance schedule | Calendar | When and by whom? | Tuesday of week 14, by the measurement crew |
You will also hear "strategy" used at asset level: "the strategy for this pump is condition-based". That is common usage. Just agree within your organisation which word you use for what.
Types of maintenance: corrective, preventive, condition-based, predictive and design improvement
EN 13306 splits maintenance into corrective and preventive. Preventive maintenance is either predetermined (at a fixed interval) or condition-based, and predictive maintenance is a form of condition-based maintenance that works from a forecast. For a maintenance concept that split is too coarse. In practice you choose from seven building blocks, which are sometimes called maintenance concepts themselves. The trade-off between the first two is covered in more depth in preventive vs corrective maintenance.
| Maintenance type | What you do | Choose it when | Example |
|---|---|---|---|
| Corrective | Repair after failure | Consequences small, repair cheap | Conveyor roller, lighting |
| Preventive, time-based | Replace or overhaul at a calendar interval | Ageing continues even when idle | Rubber coupling, hoses |
| Preventive, usage-based | The same, by running hours, cycles or tonnes | Wear follows usage | Gearbox oil, compressor filter |
| Condition-based | Measure, act at a limit value | Failure gives a measurable warning, with time to plan | Vibration on bearings |
| Predictive | Forecast the moment from a trend or model | Downtime is expensive and there is enough data | Boiler tube wall thickness |
| Failure-finding | Test the function at an interval | Failure goes unnoticed until the function is needed | Emergency stop, standby pump |
| Design improvement | Remove the cause with a modification | Recurring failure, no task helps enough | Self-aligning rollers |
Condition-based and predictive maintenance differ mainly in timing. With condition-based maintenance you respond to a limit that is being exceeded now; with predictive maintenance you calculate ahead to the moment that limit will be reached. With any of these types you can also do work while the installation is down anyway. That kind of opportunity-based maintenance replaces a wear part a little early because the next shutdown is a year away.
Which maintenance concept fits which risk and failure behaviour
The choice for each failure mode comes down to two questions: how bad is it if this happens, and how does the failure mode behave? The diagram below puts those questions in sequence. If the consequences are small, repairing after failure is often the cheapest option. If they are large, work down the five kinds of failure behaviour from top to bottom and take the first that fits.
That order follows RCM logic. Measuring comes before replacing, because you use the full life of a part and do not open up anything that is still good. Replacing at an interval only works if the likelihood of failure really does rise with usage or age. If none of those tasks fits and a failure would go unnoticed in operation, you test periodically whether the function still works. If that does not lower the risk enough either, design improvement is what remains.
Example: the main conveyor in a distribution centre
When the main conveyor in a distribution centre goes down, outbound flow stops. Here, six failure modes lead to six choices, which together form the conveyor's maintenance concept. The intervals are examples.
| Failure mode | Consequence | Failure behaviour | Maintenance type |
|---|---|---|---|
| Conveyor roller seizes | Small | Not decisive | Corrective |
| Drive drum bearing wears | Large | Gives a measurable warning | Condition-based: measure vibration every month |
| Gearbox oil degrades | Large | Deteriorates with usage | Preventive, usage-based: every 10,000 running hours |
| Rubber coupling ages | Large | Ages over time | Preventive, time-based: every 4 years |
| Pull-cord emergency stop fails | Large | Stays hidden until the function is needed | Failure-finding: function test every quarter |
| Belt keeps running off-track | Large | No task helps enough | Design improvement: self-aligning rollers |
So a single asset does not have a single maintenance type. Predictive is not on the list here: the monthly measurement leaves enough time to act, and extra data and analysis would not pay for themselves on this conveyor. Boiler tubes in a waste-to-energy plant are a different matter. There, the wall-thickness trend across several outages gives a projection of when the minimum thickness comes into view.
Methods for developing a maintenance concept
You do not have to put a concept together on gut feeling, and you do not have to analyse every asset to the same depth. These methods complement each other.
- Risk-based maintenance. Start with an asset criticality analysis, which ranks assets by the consequence and likelihood of failure and tells you how much analysis each one deserves. The Norwegian standard NORSOK Z-008 on risk-based maintenance also uses generic maintenance concepts: one analysed set of tasks and strategies for a group of similar assets under similar conditions.
- FMECA. An FMECA maps the failure modes, effects and criticality for each function. It shows the risk you run without maintenance and how far a measure lowers it.
- RCM. RCM adds the decision logic: seven questions that lead, for each failure mode, to a task that is technically feasible and worth doing. That takes time, so reserve it for installations where failure has serious consequences.
- TPM. TPM is less about which maintenance type fits and more about who does the work. Operators take on cleaning, inspection and basic lubrication, which leaves the maintenance team with time for specialist work. That makes it a good companion to the other methods.
- Reviewing your existing plan. If you already have a preventive plan, link every task to a failure mode or a requirement. Tasks with no reason get dropped, and failure modes with no task get one. This is often quicker than an analysis from scratch.
From maintenance concept to maintenance plan in six steps
A concept only pays off once it reaches the shop floor. This is how you turn the choices into a plan; the detail for each task is in how to create a maintenance plan.
- Record the choice and the reason for each failure mode. Include the decisions to do nothing. If someone wants to know in three years' time why a conveyor roller has no inspection, they can find out.
- Turn each choice into a task with a criterion. "Condition-based" only becomes a task once it is clear what you measure, where the limit lies and what happens when it is exceeded. "Failure-finding" only becomes a task once it is clear how you test the function and what you do if it does not work.
- Set the interval. The usual starting point is the manufacturer's advice. For measurements, the interval follows from the P-F interval: the time between the first signal and the failure.
- Line up the spare parts. Corrective only works if the part is on the shelf or the lead time is acceptable. Condition-based gives you time to order. So the concept helps determine what you keep in stock.
- Group tasks into rounds. Combine tasks with the same interval, the same trade and the same conditions. You can see what that looks like in practice under preventive rounds.
- Check capacity and budget, then keep reviewing. Add up the hours per trade. If they do not fit, give priority to the measures that remove the most risk per euro. Review the concept after a failure it did not see coming, after a change to the installation and when a task keeps finding nothing.
Making maintenance plannable, predictable and traceable
These three words tend to come up together when organisations want to stop firefighting. They mean different things, and a maintenance concept contributes to all three.
Plannable maintenance: you choose the moment
Plannable maintenance is work where you know in advance what it involves and you choose the moment yourself: during a planned shutdown, a cleaning window or a quiet shift, with parts and people ready. In industry, that is the difference between a line that stands still for an hour during a planned changeover and a line that fails halfway through an order. Plannable is broader than preventive. A condition-based finding is plannable as long as the warning time is longer than your preparation: if a bearing signals trouble eight weeks ahead and you measure every four weeks, the worst case still leaves four weeks to arrange the part and the downtime. A fault that can wait is plannable too; EN 13306 calls that deferred maintenance.
Predictable maintenance: you know what is coming
Predictable is not the same as predictive. Predictive maintenance forecasts when one component is likely to fail. Predictable maintenance is about your whole workload: how many hours, which shutdowns and what costs lie ahead? You rarely need a model for that. Add up the hours from the plan per week and per trade. For rounds on running hours, calculate ahead: a pump that runs 120 hours a week and has 1,800 hours to go until its next service is due in 15 weeks. Even corrective work is predictable as an average. An asset group with 24 failures a year at an average of 6 hours takes 144 hours, almost 3 hours a week, which you keep free in the schedule.
Traceable maintenance: every task and job leaves a trail
Traceable means you can search back in two directions. From the task: which failure mode, manual or legal requirement is the reason this task exists? Without that answer, nobody dares to drop a task. From the asset: who did what and when, with which parts, what was measured and what was the cause? You need that to judge whether the concept works and to show what happened. EU rules on work equipment also expect you to keep any maintenance log up to date and to record inspection results (Directive 2009/104/EC). In practice, traceability takes three things: a unique code per asset, all work through work orders and a recorded cause for every failure.
Common pitfalls when choosing a maintenance concept
- Preventive as the default. Putting a yearly service on every asset ignores failure behaviour. With random failures, replacing more often does not help, and every time you open something up there is a chance of an assembly error.
- Starting with the technology. Fitting sensors first and only then asking what they should detect. Start with the failure mode: a measurement only makes sense if the signal exists and there is time left to act on it.
- Corrective without preparation. Deliberately running to failure is not the same as arranging nothing. Without a part, a procedure and capacity, a small failure still turns into a big one.
- Forgetting hidden failures. Standby pumps and protective devices do not stand out in operation when they are defective. Without a function test, you only find out when you need them.
- A concept that stays in a report. An analysis whose outcome never reaches the plan changes nothing about what the crew does.
- No feedback. Without recorded causes and findings, you cannot see whether a choice was right, and intervals stay untouched for years.
Frequently asked questions
What is the difference between corrective, preventive and predictive maintenance?
Corrective maintenance is carried out after something has failed, preventive maintenance in advance at a fixed interval of time or usage, and predictive maintenance at the moment a forecast from measurement data indicates. Corrective uses the full service life but happens at a moment you do not choose. Preventive is easy to plan but sometimes replaces parts that are still good. Predictive gives the sharpest timing but needs measurement data and analysis, and pays off mainly where downtime is expensive.
How many types of maintenance are there?
It depends on how you count. EN 13306 starts with two main types, corrective and preventive, and splits them further. In practice many organisations work with four or five: corrective, preventive, condition-based, predictive and sometimes design improvement. This article counts seven, because time-based and usage-based preventive maintenance call for different choices and failure-finding is often forgotten. More important than the number is that every failure mode gets a deliberate choice.
Is RCM or TPM a maintenance concept?
Strictly speaking no, although the word does get used that way. RCM and TPM are ways of arriving at a concept. RCM determines which task suits each failure mode, while TPM is mainly about who does the work. The maintenance concept is what comes out: the concrete mix of maintenance types for an asset or asset group. If you say RCM is your concept, a technician still does not know what needs doing to the pump.
Does every asset need its own maintenance concept?
No. Similar assets under similar conditions can share one concept, for example all centrifugal pumps in cooling water service. That saves analysis work and keeps maintenance consistent. A separate analysis mainly makes sense for complex or highly critical installations, or when an asset is used quite differently. Intervals can also differ within a shared concept: the same pump will usually get more maintenance in a critical position than in one with a full standby unit.
Who should write the maintenance concept?
Usually a reliability engineer or maintenance engineer, together with technicians, operators and someone with a view of production and safety. Technicians know how an asset really fails, operators are the first to notice when something changes, and production knows what downtime costs. Give each asset group an owner who keeps the concept up to date. As a rule of thumb, review it once a year for critical assets, and also after any failure the concept did not see coming.