Key points
- Every maintenance type answers one question differently: what triggers the work?
- Reactive maintenance is not always wrong. For cheap, non-critical assets with a spare on the shelf, running to failure is often the sensible choice.
- Preventive maintenance on a schedule is where most teams should start. It needs only an asset list, intervals and somewhere to record what was done.
- Condition-based and predictive maintenance need measurements, history and people who act on them. Without those, they are expensive dashboards.
The six types at a glance
The table compares the types by the thing that matters most: what makes a technician go and do the job.
| Type | What triggers the work | Best for | Needs |
|---|---|---|---|
| Run to failure | The asset fails, and that was the plan | Cheap, non-critical items with a spare: light bulbs, small hand tools | A spare and a decision written down |
| Corrective (reactive) | A fault is found or reported | Unexpected failures that no schedule would have caught | A way to report faults and track repairs |
| Preventive | A calendar date, operating hours or distance | Parts that wear with use or age: oil, filters, belts, safety checks | An asset list, intervals and meter readings |
| Condition-based | A measurement crosses a limit: vibration, temperature, pressure, wear | Critical assets whose wear can be measured cheaply | Regular readings or sensors, and agreed limits |
| Predictive | A forecast says failure is likely within a window | High-value assets where a failure is very costly | Continuous data, failure history and analysis skills |
| Reliability-centred (RCM) | A structured study of how each asset fails and what that costs | Choosing the right type per asset, usually on critical systems | Time from people who know the assets |
Reactive maintenance: run to failure and corrective
Reactive maintenance means acting after something has gone wrong. It comes in two forms that are easy to confuse. Run to failure is a deliberate choice: you have decided an asset is not worth maintaining, so you replace it when it stops. Corrective maintenance is repairing a fault you did not plan for, whether it was found during an inspection or reported by an operator.
The difference matters. A site that runs its light fittings to failure is being sensible. A site where the main production pump is only looked at when it stops is not choosing reactive maintenance; it has simply never chosen anything else. The cost shows up as unplanned downtime, overtime, parts bought at short notice and repairs that are bigger than they needed to be, because a small fault was left to become a large one.
Preventive maintenance
Preventive maintenance is work done on a schedule, before anything fails. The schedule can be based on:
- Time: every week, month or year. Right for things that age whether or not they are used, such as fire extinguishers, safety inspections and many building systems.
- Usage: every 250 operating hours, every 10,000 km, every so many cycles. Right for engines, gearboxes and anything whose wear follows how hard it works.
Preventive maintenance is the right place for most teams to start, because it needs very little: a list of assets, a service interval for each (the manufacturer's manual is the usual starting point; see how to set preventive maintenance intervals), meter readings for usage-based work, and a record of what was done. Its weakness is that a schedule cannot see the asset. A filter may be changed while still clean, and a bearing may fail between two services.
Condition-based maintenance
Condition-based maintenance waits for the asset to show a sign of wear and then acts. The sign is a measurement with an agreed limit: a vibration level, a bearing temperature, a pressure drop across a filter, an oil sample, tread depth on a tyre. When the reading crosses the limit, a work order is raised.
It does not need sensors. A technician taking readings on a round and writing them down is condition-based maintenance, and for many sites it is the practical version. What it does need is discipline: readings taken regularly, limits that someone has agreed, and a rule that a reading over the limit always becomes a job. A reading that is recorded and never looked at prevents nothing.
Inspections sit close to this type. A checklist item marked failed is a condition that has crossed its limit, which is why a failed inspection item should turn into a work order rather than a note.
Predictive maintenance
Predictive maintenance goes one step further than condition-based. Instead of reacting when a reading crosses a fixed limit, it uses the trend in the data, and often the failure history of similar assets, to estimate when failure is likely and plan the work just before it. Vibration analysis, thermography and oil analysis programmes, and newer software models fed by continuous sensor data, all fall under this heading.
Done well, it lets you replace a part close to the end of its real life without letting it fail. The catch is everything it depends on: reliable data over a long period, enough recorded failures to learn from, and people with the skill to interpret the results. A site that cannot yet say which of its assets broke down last month will get little from a predictive model. It is worth considering for a small number of critical, expensive assets once the basics are in place, not as a first step.
Reliability-centred maintenance
Reliability-centred maintenance (RCM) is a method, not a trigger. For each asset, or each critical system, a team asks a structured set of questions: what does it do, how can it fail, what happens when it does, and what is the cheapest way to deal with each failure? The answer might be preventive maintenance for one failure mode, condition monitoring for another and run to failure for a third.
A full RCM study takes real time from people who know the equipment, so it is usually kept for systems where a failure is dangerous or very costly. The thinking behind it is useful everywhere, though: decide on purpose how each asset will be maintained, rather than letting it default to whatever happens.
How to choose the right mix
You do not need an RCM study to make a sensible first choice. For each asset, three questions get you most of the way:
- What happens when it fails? If the answer is a safety risk, a stopped site or a large repair bill, it needs proactive maintenance. If the answer is a short inconvenience and a cheap replacement, run to failure may be right.
- Does its wear follow time or use? If yes, a preventive schedule on the right basis (calendar for age, hours or kilometres for use) is the simplest effective choice.
- Can its condition be measured cheaply? If a reading or an inspection reliably shows wear before failure, condition-based maintenance can save work a fixed schedule would do too early.
Then measure whether it works. Two figures tell you most of the story: the share of maintenance hours spent on planned work against reactive work, and how often your critical assets fail and how long they stay down, which is what MTBF, MTTR and availability measure. See maintenance KPIs for the rest. A mix that works moves hours from the reactive column to the planned one over time.
| Example asset | Sensible starting type | Why |
|---|---|---|
| Office light fittings | Run to failure | Cheap, quick to replace, no wider impact |
| Diesel generator | Preventive by operating hours, plus a monthly test run | Engine wear follows hours; a standby unit must start when needed |
| Fire extinguishers | Preventive by calendar | They age whether used or not, and checks are often required by regulation |
| Main process pump | Preventive, plus regular vibration and temperature readings | A failure stops production, and bearing wear can be measured |
Maintenance types in Solmex
Solmex covers the types most sites run day to day. Faults are reported as breakdowns or work requests and fixed through work orders, which are typed as preventive, reactive or other so the planned against reactive split is always visible. Preventive maintenance templates schedule services by calendar, operating hours or kilometres, against the manufacturer's milestones. Inspections turn a failed checklist item into a finding and then a work order, and each asset can carry tracking meters such as pressure or temperature with an expected range, so a reading outside it is flagged. Solmex does not run predictive models; it gives you the clean history of readings, services and failures that any predictive programme would need first.
Work orders, breakdowns and work requests are on every plan, including the free Starter plan. See preventive maintenance and inspections for the details.