Preventive Maintenance Examples Across Industries

When someone asks me for preventive maintenance examples, what they usually want is not a list. They want to know whether the programme they are about to build looks anything like what a competent operation in their sector actually runs. A monthly filter change means something very different in a shopping mall than it does in a tier III data centre, and a quarterly generator test that is adequate in an office tower is negligent in a hospital theatre block. The tasks look similar on paper. The consequence profile behind them is what makes the programme.

The message up front: the same preventive maintenance logic applied to different consequence profiles produces very different programmes. The task list is largely transferable across sectors. The frequencies, the redundancy assumptions, the evidence requirements and the escalation rules are not. Copy tasks between sectors if you like, but never copy frequencies.

1. How to read a preventive maintenance example properly

A preventive maintenance example is only useful if it carries four pieces of information. Most published lists carry one. Before you lift anything from this article, or from any vendor library, insist on all four:

• The asset, specified properly. Not "AHU" but "constant-volume AHU, belt-driven, 8,000 CFM, serving open-plan office, with DX coil". The maintenance requirement changes with drive type, coil type and duty far more than with brand.
• The task, written as an action with an acceptance criterion. "Check belt" is not a task. "Measure belt deflection, replace if outside 10 to 15 mm at mid-span, record tension" is a task. If a technician can close it without recording anything, it will eventually be closed without being done.
• The frequency, with its justification. Manufacturer requirement, statutory requirement, or observed failure interval. Those three sources produce different intervals and they are not interchangeable. A frequency with no stated source is an inherited guess.
• The failure mode it defends against. This is the one almost always missing, and it is the one that tells you whether the task is worth doing at all. If you cannot name the failure the PM prevents or detects, delete the PM.

That last point is the discipline that separates a designed programme from an accumulated one. I have opened CMMS systems carrying four thousand active PM tasks where nobody could explain the origin of more than a few hundred. The rest were inherited from a previous contractor, duplicated during a migration, or added after an incident and never reviewed. For the full argument on keeping a PM library lean, see the PM programme design pillar, and for the underlying strategy choices, the complete guide to preventive maintenance.

2. The through-line: consequence profile, not task list

Every sector in this article runs chillers, pumps, air handling units, generators, panels, fire systems and lifts. The engineering is common. What differs is what happens in the first hour after one of them stops, and that single variable drives almost every design decision in the programme.

Four questions define the consequence profile, and the answers explain the differences you will see in the sector sections below:

• How long can the service be unavailable? An office tolerates a warm afternoon. An operating theatre tolerates nothing. A data centre white space tolerates seconds.
• Is there redundancy, and is it proven? N+1 does not reduce PM demand, it changes it: now you must prove the standby works, which means a testing regime the single-unit site does not need.
• When can you take a shutdown? A school gives you eight weeks of empty building a year. A hotel and a hospital give you almost nothing. That constraint reshapes the entire annual plan.
• Who audits the evidence? Civil defence, a health regulator, an accreditation body, a client contract, an insurer. The auditor determines how much of the PM cost is documentation rather than wrench time.

Work those four questions per asset class before you write a single task and the programme designs itself. Skip them and you end up with a uniform frequency applied across a non-uniform risk landscape, which is the most common structural fault I encounter. The ranking method for this sits in the asset criticality classification pillar.

3. Commercial real estate and offices

The baseline sector, and the one most published PM libraries were originally written for. Occupancy is predictable, the building empties at night and at weekends, and the dominant consequences are tenant comfort, lease compliance and energy cost rather than safety or production loss. That combination gives you something no other sector in this list has in the same measure: plannable access.

Assets that actually matter: air-cooled or water-cooled chillers, AHUs and FCUs, cooling towers, domestic and fire water pumps, the LV switchboard and distribution panels, lifts, the BMS itself, and the fire alarm and suppression systems.

What I would prioritise in an office portfolio is not the mechanical plant, which is generally over-maintained, but the two things that generate the most tenant escalations: condensate management on FCUs above ceilings, and the BMS. Blocked drain pans cause more visible damage in commercial buildings than compressor failures do, and an unmaintained BMS quietly drifts into manual overrides until nobody knows what the sequences are supposed to be. A quarterly PM that simply verifies control sequences, setpoints and sensor calibration against the original design intent is one of the highest-return tasks in the whole library, and it is almost never in the inherited PM list.

4. Hospitality: hotels and resorts

Hospitality is where the access constraint becomes the design driver. The building never empties, the guest is physically inside the asset you need to maintain, and a maintenance failure is not a work order, it is a review. The plant room asset list is close to a hotel-shaped office building. The programme is not.

Assets that actually matter: guest room FCUs and bathroom extracts, domestic hot water generation and the recirculation loop, kitchen equipment and the kitchen extract system, laundry plant, pool filtration and dosing, lifts, and the guest-facing electrical and lighting infrastructure.

The structural difference in hospitality is that guest rooms are the largest asset population and they are only accessible when vacant. The pattern that works is a rolling room PM tied to the property management system: the CMMS receives out-of-order or vacant-room availability and releases room PM work orders against it, so the programme consumes vacancy rather than fighting occupancy. Where that integration does not exist, room PM compliance is usually the worst-performing line in the whole KPI pack, and it degrades quietly because nobody escalates a filter.

5. Healthcare: hospitals and clinics

Healthcare is the sector where the consequence profile stops being commercial and becomes clinical, and the whole programme reorganises around that. Two features dominate. First, the building cannot be shut down, so almost all work happens live with formal isolation and infection-control controls. Second, an accreditation body will read your maintenance records, which means the evidence trail is part of the deliverable rather than an administrative afterthought.

Assets that actually matter: theatre and isolation-room air handling, medical gas plant and pipelines, essential-supply generators and UPS, the IPS and isolated power systems in theatres, sterilisation equipment, nurse call, medical refrigeration, and imaging chiller and power support.

The thing healthcare teaches that every other sector should borrow is verification as the task, not the service as the task. In a hospital the theatre PM is not "service AHU", it is "prove the theatre still meets its pressure cascade and air change specification", with a measured number recorded. The service work exists to achieve that number. Once you frame PM that way, the acceptance criterion is unavoidable, and the PM becomes auditable rather than assertable. Standards worth reading directly here rather than through a contractor summary include the ASHRAE healthcare ventilation guidance and the life-safety and essential-electrical provisions published by the NFPA .

6. Manufacturing and process plant

Manufacturing is the only sector in this list where the maintenance department has a directly measurable output cost attached to every hour of downtime, and that changes the culture as much as the schedule. PM here is not primarily about comfort or compliance, it is about throughput, and the programme is built around production-critical equipment rather than around the building.

Assets that actually matter: production line motors, gearboxes and drives, conveyors, compressed air generation and distribution, process pumps and agitators, hydraulic and pneumatic systems, dust extraction, CNC and machining centres, and steam or thermal fluid plant.

Two habits from manufacturing are worth exporting to every other sector. The first is running-hours and cycle-based triggers rather than calendar triggers, which matches the PM to actual duty instead of to the almanac. The second is the operator-level daily check, a short autonomous routine done by the person running the machine, which catches the obvious things before they need a technician. Facilities teams rarely use either, and both would improve most building programmes. Platforms handle meter-based PM with very different levels of grace, so if duty-based triggering matters to you, test it properly during selection: IBM Maximo, SAP PM, Hexagon EAM and Infor EAM all support it natively, while some of the lighter-weight tools such as MaintainX, Limble, Fiix and UpKeep vary in how well meter readings and multi-trigger PMs behave at scale.

7. Data centres and critical facilities

The data centre is the extreme case in this article: near-zero tolerance for interruption, deep redundancy, and a maintenance philosophy where the risk of the maintenance activity itself is treated as seriously as the risk of the failure it prevents. Everything in the programme is shaped by one fact, which is that the most dangerous moment in a resilient facility is usually a person working on it.

Assets that actually matter: UPS modules and battery strings, generators and fuel systems, switchgear and automatic transfer switches, CRAC and CRAH units, chilled water plant and pumps, fire detection and suppression, leak detection, and the monitoring layer itself.

The two disciplines that define a good critical-facility PM programme have almost nothing to do with the task list. The first is concurrent maintainability: every PM is written to be executed on one path while the other carries the load, with the isolation sequence stated in the task rather than improvised at the panel. The second is the method statement and change-control wrapper, where high-risk PMs go through review, get a scripted script-of-work with abort criteria, and are executed in a maintenance window with rollback defined. Redundancy also creates a PM category other sectors forget entirely, which is exercising the standby: pumps, valves, breakers and transfer switches that sit idle for months are exactly the components that fail on demand, so periodic operation is itself the preventive task.

8. Retail and shopping malls

Retail runs on footfall, and footfall creates a maintenance profile driven by public safety, crowd-facing equipment and trading hours rather than by engineering criticality. The building is busy exactly when the plant is most loaded, and the quiet window is short. What fails here is rarely exotic. It is the equipment the public touches.

Assets that actually matter: escalators and travelators, lifts, automatic doors, car park ventilation and barrier systems, mall HVAC and toilet extract, refrigeration in the food and grocery zones, grease management, signage and emergency lighting, and the fire and evacuation systems.

The practical constraint in retail is that almost nothing intrusive can happen during trading hours, and trading hours in a Gulf mall run long. That pushes the programme into a narrow overnight window and makes labour planning, not task definition, the hard part. Malls also have a split-responsibility problem that quietly wrecks compliance figures: landlord plant and tenant plant on the same asset tree, where tenant kitchen extracts and refrigeration are maintained by whoever the tenant hired. If the CMMS does not model that ownership boundary explicitly, the landlord reports compliance on assets it does not control and discovers the truth during an incident. Work order routing and ownership matter enormously here, which is the subject of the work order types pillar.

9. Education: schools and university campuses

Education has the most favourable maintenance calendar of any sector in this article and the most challenging asset diversity. The favourable part is the shutdown: a campus is genuinely empty for weeks, which allows intrusive work no hospital or hotel could ever schedule. The difficult part is that a university estate contains offices, laboratories, residences, catering, sports facilities, workshops, a district cooling plant and sometimes a small clinic, each with its own profile, spread across dozens of buildings.

Assets that actually matter: central plant and district cooling, lecture-theatre AV and ventilation, laboratory fume cupboards and gas supplies, residence hot water and FCUs, catering and kitchen plant, sports and pool plant, lifts, and the fire and access control systems across a large building count.

The pattern I would recommend for a campus is a two-speed annual plan: a light, non-intrusive routine that runs through term time, and a heavy intrusive programme concentrated in the vacation, planned and resourced months in advance because the window cannot slip. Under-using that vacation window is the single most common failure I see on campus estates, because the planning must be done during the busiest part of term and it is easy to defer. Multi-building campuses also need a hierarchy that mirrors site, building, floor and system rather than a flat asset list, or reporting becomes meaningless above the individual asset. That is covered in the CAFM for universities pillar and, for the estate-wide structure, the multi-site CAFM architecture pillar.

10. Fleet and logistics

Fleet is the purest example of meter-driven preventive maintenance, and for that reason it is the sector facilities teams learn the most from. Nothing about a vehicle PM is calendar-based by preference: it is kilometres, engine hours, or a regulatory inspection date. The asset also moves, which means the maintenance system has to deal with location, driver reporting and off-site work in a way building CMMS rarely does.

Assets that actually matter: light vehicles, heavy trucks and trailers, refrigerated units, forklifts and material-handling equipment, mobile plant and cranes, tail lifts, and the depot infrastructure including fuel systems, wash bays and charging points.

The transferable lesson from fleet is the pre-use check. A driver walkaround, completed on a phone in two minutes and logged against the vehicle, catches lights, tyres, leaks and damage before they become breakdowns, and it costs nothing in technician time. The building equivalent, an operator or attendant daily round on plant rooms and critical equipment, is cheap and consistently under-used. The other lesson is that fleet teams have long accepted whichever comes first logic on triggers, mileage or time, where facilities teams tend to pick one and ignore duty entirely.

11. Cross-sector comparison

Laid side by side, the pattern is clear. The asset classes repeat. The frequencies, drivers and access constraints do not.

Read across the generator column in particular. It is the same asset, doing the same job, in every sector. The test regime ranges from monthly off-load running to a weekly run plus monthly on-load transfer plus an annual load bank, and the difference is entirely consequence, not engineering. That column is the whole article in one row.

12. What transfers between sectors, and what does not

Because sector libraries are commonly bought, inherited or lifted, it is worth being explicit about what you can safely reuse.

Transfers well:

• Task content for common plant. A chiller service is a chiller service. Vibration routes, thermographic survey methods, water treatment regimes and filter change procedures are portable across every sector here.
• Structural practices. Meter-based triggering from manufacturing, pre-use checks from fleet, verification-with-acceptance-criteria from healthcare, and standby exercising from data centres all improve any programme that lacks them.
• Asset hierarchy and naming discipline, which is sector-independent and consistently worth the effort.

Does not transfer:

• Frequencies. This is the important one. A frequency encodes a consequence judgement and an access reality specific to the site it came from.
• Evidence and documentation depth. A hospital or data centre records far more per task than an office needs, and importing that overhead into a commercial portfolio inflates cost without reducing risk.
• Shutdown assumptions. Any PM written around an available shutdown will quietly fail in a sector that has none.
• Redundancy assumptions. A task written for an N+1 estate may leave a single-path site exposed, and a task written for a single-path site misses the standby-exercising requirement entirely.

13. Building your own sector example set

If you are writing a programme from scratch, or rebuilding an inherited one, the sequence that works is short and it is deliberately not software-led:

• Verify the asset register first. A PM programme built on an inaccurate register produces work orders against assets that do not exist and misses ones that do. Physically verify the critical asset classes before writing tasks.
• Rank by consequence, not by cost or size. A small extract fan serving an isolation room outranks a large chiller serving an atrium. Consequence, not kilowatts.
• For each critical asset class, list the failure modes you have actually seen or can credibly expect. Then write only tasks that prevent or detect one of them, and name the mode in the task record.
• Set the frequency from the strongest available source, in this order: statutory requirement, manufacturer requirement under your duty conditions, observed failure interval, then recognised industry guidance such as the SFG20 schedules for building services. Record which source you used.
• Test the frequency against access reality before you publish it, and reduce it to something achievable rather than defending a number you will miss.
• Give every task an acceptance criterion and a required reading, so completion means something measurable was observed.
• Review annually against failure history. Tasks that have never found anything in three years are candidates for extension or deletion; asset classes still failing between PMs are candidates for a shorter interval or a different strategy entirely.

The review step is where a PM programme either matures or ossifies, and it is the step almost always skipped. A programme that never deletes anything is not cautious, it is undisciplined, and within a decade it will be spending most of its labour on tasks that defend against nothing. For choosing between time, meter and condition triggers per asset class, see the preventive maintenance strategies pillar, and for measuring whether any of it is working, the FM KPI framework pillar.

The idea to walk away with

Preventive maintenance tasks are close to universal. Preventive maintenance programmes are not. The office, the hotel, the hospital, the plant, the data centre, the mall, the campus and the depot all service pumps and fans and panels and generators, and they all do it on radically different cadences, with radically different evidence, because the first hour after a failure looks radically different in each.

So borrow tasks freely and frequencies never. When you look at any preventive maintenance example, including the tables above, read it as a question rather than an answer: what consequence profile produced this interval, and is mine the same? Where the answer is yes, take it. Where it is not, the example has still done its job by telling you which failure mode you need to defend against, and you can set your own interval from there.

Final thoughts

The most useful preventive maintenance example set is the one you generate from your own estate: verified assets, ranked by consequence, with named failure modes, sourced frequencies, honest access assumptions and measurable acceptance criteria. Everything published, this article included, is scaffolding to get you to that point faster. The moment your own failure history is good enough to argue with the published interval, you have a real programme rather than a borrowed one.

If I had to name the one habit that most improves a PM programme regardless of sector, it would be writing the defended failure mode into every task. It forces the justification at creation time, it makes the annual review possible, and it quietly prevents the slow accumulation of tasks nobody can explain. It costs a sentence per task and it is the difference between a library and a liability.

Related reading: Preventive maintenance: the complete guide, Preventive maintenance strategies, PM programme design: quality over quantity, Asset criticality classification, Work order types in a CMMS, CAFM for universities.