Electrical Preventive Maintenance (EPM) and PPM Checklists

Most PPM programmes I am asked to review have a mechanical bias. The chillers, pumps, AHUs and fans are covered in forensic detail, and the electrical side is a handful of generic tasks with words like "inspect panel" and "check operation" attached to an annual frequency. That imbalance is understandable, because mechanical assets announce themselves when they are unwell: they get noisy, they vibrate, they leak. Electrical assets stay quiet, keep working, and then fail in a way that takes out a building, an operating theatre or a data hall, and occasionally injures the person standing in front of them. Electrical preventive maintenance is the discipline of finding the developing fault while the equipment is still quiet, and doing it without becoming the incident yourself.

The message up front: a credible electrical PPM programme is built from three things, not one. A per-asset-class task list with honest frequencies. A defined test or measurement for each task, so the outcome is a number you can trend rather than a tick. And an isolation, permit and competency framework that decides who is allowed near what, in which energy state. Get the third one wrong and the first two are irrelevant, because the work will either not be done or will be done unsafely.

1. Why electrical PM is not just another PPM line item

Electrical distribution equipment has a set of properties that make it behave differently from mechanical plant, and each one has a direct consequence for how you write the PM.

• It degrades invisibly. A bolted connection loosens through thermal cycling over years. Insulation absorbs moisture and its resistance falls. A circuit breaker sits closed for a decade and its mechanism stiffens with hardened grease. None of this produces a symptom an operator will report. It is found only by deliberate inspection, testing or thermal imaging.
• Its failures are often step changes, not gradual decline. Much electrical failure has a very short window between "measurably degraded" and "catastrophic". The degradation is gradual and detectable; the failure event itself is not. That is exactly the argument for fixed-interval testing rather than waiting for a condition signal.
• The consequence is disproportionate to the asset value. A failed 400 amp moulded case breaker is a modest purchase. The arc flash event that accompanies its failure under load is not, and neither is the loss of the load it was feeding.
• Most meaningful testing requires the equipment to be dead. Insulation resistance, contact resistance, breaker timing, relay secondary injection and torque verification all need the equipment isolated. That single fact drives the whole shape of the programme, because de-energised access to live distribution is the scarcest resource in facilities maintenance.
• Competency is regulated, not assumed. Anyone with a spanner can change a belt. Not anyone can open an MV panel, and in most jurisdictions that is a legal position, not a company preference.

The practical consequence: an electrical PPM programme cannot be built by copying the OEM manual into the CMMS and setting everything to annual. It has to be built around what can realistically be isolated, how often, and what you will do about the equipment you cannot isolate. If you have not yet set up the general PM foundation, the complete guide to preventive maintenance covers the underlying frame this sits on.

2. NFPA 70B and the standards backdrop

The reference document most electrical maintenance programmes are measured against is NFPA 70B, Standard for Electrical Equipment Maintenance, published by the National Fire Protection Association . The important change worth knowing, because it affects how auditors and insurers treat it, is that NFPA 70B was for decades a recommended practice and was reissued as a standard with the 2023 edition. That is not a cosmetic relabel. A recommended practice is guidance you may choose to follow. A standard is written in mandatory language and becomes something that can be adopted, referenced in contracts, cited by an insurer, or in some jurisdictions enforced.

The practical effect for a facilities or maintenance manager is that "we do what we have always done" has become a weaker position than it was. If your electrical maintenance programme has no documented basis, no condition assessment driving frequencies, and no records of test results, you are now measurably behind a published standard rather than behind an opinion.

Around NFPA 70B sit the other documents you will encounter. NFPA 70E addresses electrical safety in the workplace and is the origin of much of the arc flash risk assessment and boundary thinking used internationally. The IEEE colour book series and its successor standards cover testing and maintenance of specific equipment classes in engineering depth, and IEC publishes the equivalent international equipment standards used across Europe, the Gulf and Asia. In the UK and much of the Commonwealth the BS 7671 wiring regulations and the associated inspection and testing regime, along with SFG20 for the PPM task library, will be the operative references instead.

3. The nine electrical asset classes worth separating

The first failure of most electrical PPM registers is granularity. "Electrical distribution" as a single asset class with a single task list is not maintainable. These nine classes each have distinct failure modes, distinct tests and distinct competency requirements, and each deserves its own PM route.

• LV switchgear and main distribution boards: the main switchboard, ACBs and MCCBs, busbar chambers, metering. The highest fault energy anyone in a normal facilities team will encounter.
• MV switchgear and ring main units: 11 kV or similar, often partly owned or controlled by the distribution utility. Specialist territory, almost never in-house.
• Transformers, dry type and oil filled: two genuinely different maintenance problems that people wrongly treat as one.
• Sub-distribution boards and final circuits: the long tail. Individually low consequence, collectively the source of most electrical fires.
• Protection relays and settings: the safety system of the electrical installation. Silent, untested in most buildings, and the thing that decides whether a fault is cleared in 100 milliseconds or becomes an incident.
• UPS systems and battery banks: the highest failure rate of anything in the list, driven almost entirely by battery degradation.
• Standby generators and changeover: the asset most likely to be maintained diligently and still fail when needed, because the PM tests the engine and not the transfer.
• Cable systems and terminations: low task count, long intervals, and the source of the most expensive single failures.
• Earthing, bonding, lightning protection and power factor correction: the two things nobody notices until an audit or a utility penalty.

Before you set frequencies, rank these by criticality per location. A distribution board feeding a car park barrier and one feeding a data hall UPS input are the same asset type and should not be on the same schedule. The asset criticality classification method is what turns one generic task list into three tiers of frequency.

4. The per-asset-class PM table

This is the core of the guide: representative tasks, a starting frequency, the measurement that actually tells you something, and the energy state the task requires. Treat the frequencies as a first draft for a normal commercial or institutional environment at moderate criticality. Harsh environments, high dust or salt, heavy cycling duty, high humidity or critical loads pull them shorter. NFPA 70B's own model is that frequency should follow a condition assessment, not a calendar habit, so revise these once you have two or three cycles of test data.

Two habits make this table worth more than the sum of its rows. First, record the measured value, not a pass or fail. A contact resistance of 42 micro-ohms means nothing in isolation and everything when the previous two readings were 28 and 33. Second, structure the failures you find with consistent codes so the register tells you which asset class is actually degrading. The problem, cause and action coding model applies directly here.

5. The tests that actually tell you something

A PPM task that says "inspect and test" produces a signature and no information. These are the measurements that produce an engineering number, and what each one is genuinely good at.

• Infrared thermography. Finds loose or corroded connections, overloaded and unbalanced circuits, failing breaker contacts and restricted cooling. Its value comes from disciplined method: survey under representative load, record the load at the time, compare phase to phase and against ambient, and report a delta T rather than an absolute temperature. A survey at 15 percent load on a Sunday morning finds nothing and proves nothing.
• Insulation resistance. The baseline health check for windings and cables. Read the trend, not the single number, and always temperature correct, because insulation resistance changes sharply with temperature. For motors and transformer windings the polarisation index, the ten minute reading divided by the one minute reading, tells you about moisture and contamination more reliably than the raw megohm value.
• Contact and connection resistance. Measured in micro-ohms with a low resistance ohmmeter. This is the direct measurement of the thing thermography detects indirectly. Rising contact resistance on a breaker means eroded or misaligned contacts; a high bolted joint reading means a connection that will eventually become a hot spot.
• Torque verification. Unglamorous and consistently skipped. Bolted connections loosen through thermal cycling, and an under-torqued joint heats, which loosens it further. Use a calibrated torque wrench to the manufacturer's value, not a technician's feel, and mark verified joints so the next visit can see what was done.
• Oil analysis and dissolved gas analysis. For oil filled transformers this is the closest thing to a window inside the tank. Basic oil quality covers moisture, dielectric strength and acidity. DGA reads the gases dissolved in the oil and the ratios between them indicate the type of fault developing: overheating of oil or paper, low energy partial discharge, or high energy arcing. Furan analysis estimates paper insulation ageing. Trend it annually and sample more frequently the moment a gas ratio moves.
• Battery impedance or conductance. The only practical way to find the weak cell in a string without a full discharge. Trend each cell against its own baseline and against the string average; a cell drifting upward by a significant margin is the one that will collapse the string. It complements rather than replaces a periodic capacity test, because impedance predicts weakness and only a discharge test proves actual delivered capacity.
• Protection relay secondary injection. Injects known current or voltage into the relay and verifies that it picks up at the set value and operates in the set time, and that the trip circuit actually trips the breaker. This is the only way to know that the protection system will do its job. In a distressing proportion of buildings it has never been done since handover, which means the discrimination study in the O&M file is a theory.
• Partial discharge measurement. For MV equipment, detects insulation breakdown activity before it becomes a flashover, using ultrasonic or transient earth voltage techniques. Non-intrusive and performed energised, which makes it one of the few useful MV condition checks that does not need an outage.

6. Isolation, permits and lock-out tag-out

This section is the reason the frequency column in the table above is not the hardest part of the job. Most valuable electrical PM requires the equipment dead, and getting equipment dead in a live building is an organisational problem, not a technical one.

The safe isolation sequence is well established and the order is not negotiable. Identify the correct point of isolation from current drawings, not from labels alone. Isolate. Secure the isolation with a lock and a personal key, one lock per person working. Apply a tag that names the person, the work and the date. Prove the test instrument works, prove the conductors dead at the point of work, then prove the instrument again. Apply earths where the equipment and the voltage require it. Only then does the work start. The person who applies the lock is the person who removes it, and nobody else, ever.

Around that sequence sits the permit system. A permit to work for electrical activity should name the isolation points, the person in charge, the limits of the work, the earthing applied, the people covered, and the conditions for return to service. This is where a CMMS earns its licence fee, by refusing to allow the work order to move to execution until the permit is raised and approved by a named authorised person, and by holding the signed record afterwards. The mechanics of wiring that into the system are covered in the permit to work integration guide.

Arc flash. Electric shock is the risk everyone thinks about; arc flash is the one that causes the severe injuries. An arc fault releases energy as heat, light, pressure and molten metal, and the incident energy at a given point depends on the available fault current and, critically, on how fast the upstream protection clears it. This is the point where two parts of this guide connect: an untested, mis-set or sluggish protection relay does not just risk equipment damage, it directly increases the incident energy a technician would be exposed to. Relay testing is arc flash risk control.

The controls are an arc flash risk assessment that establishes the incident energy and boundaries for each item of equipment, labelling that carries that information where the worker can read it, appropriately rated arc flash PPE, and, above all, a working practice that prefers de-energised work. The methodology for calculating incident energy, the boundary definitions and the PPE category approach come from NFPA 70E in the US and from equivalent national and IEC-based frameworks elsewhere. The specific requirements, the labelling format and the legal duty differ by jurisdiction, so use your local framework and do not assume an American arc flash label scheme satisfies a European or GCC inspector, or the reverse.

7. Who is competent to do what

Competency in electrical maintenance is a legal and technical question combined, and it is the one I see handled most loosely in outsourced FM. A useful practical tiering, which you should map onto your own jurisdiction's terminology and licensing:

Three rules I would hold to regardless of jurisdiction. Competency is per equipment type and per site, not a general badge: an excellent LV electrician is not qualified on your 11 kV ring main unit because he has ten years of experience. Authorisation must be written, dated, reviewed and visible to whoever raises the work order. And the CMMS should enforce it, by restricting which trade or certification group can be assigned to which PM, so scheduling cannot quietly put the wrong person in front of an MV panel because he was the one available. Getting the work order types right is what makes that enforcement possible, because permit-bearing electrical work needs to be a distinct type with its own approval path.

8. Setting frequencies you can actually defend

Frequencies in electrical PPM tend to be inherited rather than decided. A defensible method has four inputs, and NFPA 70B's condition-assessment approach formalises this thinking.

• Criticality of the load. The consequence of failure of what sits downstream. A board feeding life safety systems, a data hall or a hospital theatre is not on the same interval as one feeding a landscape irrigation pump.
• Environment. Dust, humidity, salt-laden air, high ambient temperature, vibration and corrosive atmosphere all accelerate degradation. Coastal Gulf plant rooms are a harsher environment for electrical equipment than most temperate-climate assumptions allow for, and I routinely shorten inherited intervals for that reason alone.
• Duty and loading. A breaker that operates twice a day ages mechanically far faster than one that has not moved in five years, though the second has its own problem: a stiff mechanism that will not operate when finally called on. High loading and heavy cycling both shorten intervals, for different reasons.
• Condition history. The evidence from your own test results. Three clean thermographic surveys and stable insulation resistance readings justify extending an interval. A rising trend justifies shortening it, or moving to a condition-monitoring approach. This is the input everyone claims to use and almost nobody records well enough to use.

The broader logic of choosing between fixed-interval, meter-based and condition-based triggers is set out in the preventive maintenance strategies guide, and for electrical assets the answer is usually a blend: fixed-interval for the intrusive testing that needs an outage, condition-based for the non-intrusive surveys that can run more often. Where you have the fault history and enough similar assets to see patterns, predictive techniques add real value on transformers and battery systems in particular, because both degrade along a measurable trend.

9. Writing the checklist so the result is usable

The difference between a checklist that improves reliability and one that generates paperwork is entirely in how the tasks are written. The rules I apply when rebuilding an electrical PPM library:

• One task, one verifiable action. "Inspect and test main switchboard" is not a task, it is a project. Break it into the fifteen things it actually contains, each with its own outcome.
• Capture values, not ticks. Any task with a measurement gets a numeric field with units, and where possible the previous reading shown alongside so the technician sees the trend at the point of work. A field that only accepts pass or fail throws away the entire diagnostic value of the visit.
• State the energy state on the task line. Every electrical task should carry an explicit isolated or energised flag. It removes ambiguity, it drives the permit requirement, and it makes the schedule honest about how much outage time the programme actually needs.
• State the required competency on the task line. So the planner cannot assign it to the wrong tier and the technician can decline it without a debate.
• Reference the acceptance criterion. The torque value, the minimum insulation resistance, the maximum acceptable delta T, the RCD trip time limit. If the technician has to guess what good looks like, the reading is decoration.
• Separate the outage-dependent tasks into their own PM. Do not bury a three-yearly isolated task inside an annual energised route, because the annual will get signed off and the isolated element will quietly never happen. This is the single most common structural defect I find in electrical PPM libraries.
• Require the follow-up to be a work order, not a comment. A thermographic anomaly noted in the report and not raised as corrective work is a finding that will still be there next year, hotter.

The general structure and template patterns for this are in the PM checklists, templates and examples guide; electrical simply applies them with the energy state and competency fields added.

10. Where electrical PM does not work, and what it costs

An honest guide has to say where this approach runs out, because the gap between the textbook programme and what a live building will actually permit is wide.

Three further limitations worth stating plainly.

Intrusive electrical PM can itself cause failure. Opening a panel introduces dust and moisture, disturbs settled connections, and risks a dropped tool or a reassembly error. There is a real school of thought that on well-designed, lightly-loaded, benign-environment LV equipment, a programme of non-intrusive monitoring with long intervals between intrusive work produces better availability than frequent internal intervention. I think that argument has merit and I would weigh it case by case rather than defaulting to the shortest interval the manual allows.

The cost is dominated by specialists and outages, not labour hours. Protection relay testing, MV diagnostics, transformer electrical testing and load bank hire are third-party specialist costs, and the outage itself often costs more than the test. This is why electrical PPM budgets get cut in the middle of a contract term: the annual thermography survives and the five-yearly MV service does not. If you are building the programme, defend the long-interval specialist items explicitly in the budget, because they are the first thing to disappear.

Documentation debt limits everything. Half the buildings I look at cannot produce accurate single line diagrams, a current protection coordination study, or the original commissioning test results. Without a baseline, your first round of testing produces numbers with nothing to compare them to, and without accurate drawings safe isolation itself becomes uncertain. If that is your position, drawing verification and a baseline test campaign are the first project, before any frequency discussion. It is unglamorous, it is not cheap, and there is no way around it.

11. A twelve month starting plan

If you are taking over an electrical estate with a weak programme, this is the sequence I would recommend rather than trying to implement the full table on day one.

• Months 1 to 2: verify what you have. Asset register for all nine classes with location, rating, age and the load each feeds. Single line diagram verification. Gather any existing test records and commissioning data.
• Month 2: criticality ranking. Rank every board, transformer, UPS and generator by consequence of failure. This produces your three frequency tiers and tells you where the money goes.
• Months 2 to 3: fix the safety framework first. Isolation procedure, permit system, lock-out tag-out kit and discipline, authorisation register, arc flash assessment status. None of the technical programme is deliverable without this, and it is also the part that protects people.
• Month 3: baseline thermographic survey of everything. Non-intrusive, no outage, broad coverage. It gives you a condition picture of the whole estate and a defect list to work from immediately.
• Months 4 to 6: baseline testing on the critical tier. Insulation resistance, contact resistance, transformer oil sampling and DGA, battery impedance across every string. These are your reference numbers for every future trend.
• Month 6: protection relay status review. Establish whether a current coordination study exists and whether relay settings match it. If neither, this becomes a defined project. Do not let it drift, because it is both a reliability and a personnel safety item.
• Months 6 to 8: build the PPM library in the CMMS. Task lists per asset class with value capture fields, energy state, competency and acceptance criteria. Outage-dependent tasks as separate PMs on their own frequencies.
• Months 8 to 10: negotiate and calendar the outages. Work out how much isolated access the programme needs, secure it in the operational calendar, and register the assets where it will not be granted.
• Months 10 to 12: run, measure, revise. Track PM completion, defects found per survey, repeat findings, and outages achieved against planned. Use the first year of data to adjust frequencies on evidence rather than habit.

The idea to walk away with

Electrical preventive maintenance is not a longer version of mechanical PPM. It is a programme whose design is dictated by two constraints that mechanical maintenance largely escapes: the most informative tests require the equipment dead, and the work carries a risk of serious injury that is controlled by systems rather than by care. Everything that makes an EPM programme credible follows from taking those two constraints seriously. Per-asset-class task lists so you know what to do. Measured values rather than ticks so you know whether it is getting worse. Explicit energy state and competency on every task so the right person does it in the right condition. Outage-dependent work held as its own schedule so it cannot be silently skipped. And a permit and isolation framework built before the technical programme rather than bolted on after.

The programmes that fail are not the ones with the wrong frequency. They are the ones where the annual energised walk-around is signed off every year, the three-yearly isolated service has never once happened, the relay settings have not been verified since handover, and nobody has a number to trend. That is a programme that generates records without generating reliability, and it will look compliant right up to the moment it is not.

Final thoughts

The elevation of NFPA 70B from recommended practice to standard is the clearest signal of where expectations are moving. Electrical maintenance is being treated less as engineering discretion and more as a documented, evidence-based obligation, and the organisations that will find that comfortable are the ones already recording measured values against defined acceptance criteria on a schedule they can justify. If your electrical PPM cannot currently produce last year's insulation resistance readings, the transformer's DGA trend, the battery impedance per cell, or the date the protection relays were last injection tested, that is the gap worth closing first. The frequency table can wait; the evidence base cannot.

And if you take one operational decision from this guide, make it the separation of energised and isolated work into distinct scheduled items, each with its own frequency and its own permit requirement. It is a small change in the CMMS. It is the change that stops the difficult, valuable, outage-dependent half of your electrical programme from disappearing without anyone noticing.

Related reading: Preventive maintenance: the complete guide, PM checklists, templates and examples, Preventive maintenance strategies, Permit to work integration with a CMMS, Asset criticality classification, Predictive maintenance and failure prediction.