Short answer: for a cast resin dry-type transformer in a clean, indoor, ventilated room running at or below about 80% load, start from a monthly visual check, a quarterly service visit, a semi-annual electrical check and one annual outage — then move each of those intervals up or down using four site variables: dust and pollution, humidity and condensation risk, room temperature and ventilation, and the load profile. The calendar is the starting point, never the answer. This page gives you the baseline schedule, the rules for compressing or stretching it, the tests that belong on a fixed interval versus the tests that are event-driven, and the records you need so the interval can be tuned with evidence instead of habit.

Everything marked [Confirm] is site-specific and must come from the manufacturer's manual, your insulation coordination study, or a standard's current edition. Nothing here is a substitute for the OEM document that shipped with your unit.

1. Why maintenance intervals are set by environment and load, not by the calendar

A cast resin dry-type transformer has no liquid to sample, no breather to service, no oil pump or radiator to maintain. That removes a whole category of periodic work and it is the single biggest operating-cost advantage over an oil-immersed unit. What it does not remove is the exposure of the insulation system to the room it sits in.

The epoxy-encapsulated windings are very good at resisting moisture and dust. The places that fail are the places epoxy does not cover:

• Terminal connections — busbar joints, cable lugs, and the copper-to-aluminium interfaces where creep and thermal cycling loosen a joint over years.

• Surfaces — conductive dust plus humidity creates a tracking path across bushings, insulators, standoffs and the winding surface. This is the most common early-life failure mode for cast resin units in industrial and coastal sites.

• Cooling path — fan bearings, air ducts, filters on filtered enclosures, and the room ventilation itself.

• Monitoring chain — PT100 sensors, the temperature controller, the anti-condensation heater and its thermostat.

Every one of those four degrades at a rate set by the site, not by the date. A transformer in a clean, dry, conditioned electrical room at 30% load may genuinely need one visit a year. The identical model in a cement plant, a coastal substation or a humid basement may need quarterly attention and still fail early if the room is wrong. That is why the interval has to be derived, then defended with records.

If you want the constructional background — what is encapsulated, what is not, and how cooling classes AN and AF change the maintenance burden — the dry-type transformer overview covers the fundamentals, and the cooling classes guide explains what the AF fan actually adds and what it costs you in noise.

2. The baseline interval table

This is the starting point for a clean, indoor, ventilated room with a load factor at or below roughly 80%. Adjust it using sections 4 through 7.

Two things about this table are deliberate.

The "weekly" inspection has been dropped to monthly. A weekly visit is standard practice for attended industrial substations where an operator walks past the equipment anyway, and there is no harm in that — but for an unattended building or data centre room, a weekly trip to look at a dry-type transformer is cost without information. Remote monitoring buys you more than a weekly walk: a permanently installed winding temperature indication with alarm contacts and a Modbus connection to the building management system turns the monthly visit into a data point rather than the only data point. Keep weekly walk-throughs where an operator is already in the room; do not add a visit for its own sake.

The visit frequency is less important than the load at which you take readings. An infrared scan at 15% load tells you almost nothing about a joint that only gets hot at 90% load. An insulation resistance reading taken on a humid morning and one taken on a dry afternoon are not comparable. Schedule readings, not just visits.

3. The four site variables that actually move your interval

Use this as the decision table. Find your worst-case column for each variable, then take the shortest resulting interval — do not average them.

Mapping to intervals:

That "high severity" row is where the source guidance on harsh environments lands, and it is also where the original text contains an error worth flagging: it says quarterly service should become "12 months", which would lengthen the interval under conditions that demand shortening it. The intent is clearly monthly. If you are working from a Chinese-language maintenance card that says 季度保养改为1个月1次, that is the correct reading; "12个月" in that position is a transcription error.

4. Humidity and condensation: the variable most people get wrong

Relative humidity alone is not the criterion, and this is the single most useful correction on this page.

A cast resin winding at 70% RH in a warm room is fine. The same unit at 55% RH when the winding is cold and the room air hits its dew point will condense water on the coldest surface it can find — and a cold epoxy surface at start-up after a long idle period is exactly that surface. What matters is the relationship between surface temperature and dew point, not the RH number on a wall-mounted gauge.

Practical consequences:

1. Anti-condensation heaters must be energised whenever the transformer is de-energised — not only when it is running. A strip heater that is switched with the main breaker protects the transformer during the one period it actually needs protection. Verify this at commissioning; it is a frequently missed wiring detail. [Confirm: heater supply arrangement and rating per OEM.]

1. The condensation risk is highest at low load, not high load. A lightly loaded winding runs cool and stays near ambient; a heavily loaded winding runs 40–60 K above ambient and dries itself. This produces the counterintuitive maintenance rule in section 6.

1. Measure and log dew point or surface temperature, not just RH, if your site has a history of moisture problems. A differential of [Confirm: 3–5 K] between surface temperature and dew point is a widely used margin. [Confirm per OEM and site practice.]

Where humidity is genuinely high, shorten the service interval for cleaning and sealing checks first — those are the actions that prevent tracking. Adding electrical tests does not prevent anything.

5. Dust, pollution and the cleaning interval

Cleaning is the highest-value maintenance action on a cast resin transformer and the one most often scheduled by calendar instead of by condition.

Set the cleaning interval from the pollution severity of the site (the framework in IEC 60815-1 is the standard reference, even though it is written for outdoor insulators — the severity classes transfer well to indoor electrical rooms):

Method matters as much as frequency:

• Vacuum with a soft brush first. Compressed air is the second choice, not the first: it drives conductive dust deeper into winding ducts, generates static, and carries water and oil unless the supply is clean and dried.

• Never pressure-wash. Water ingress into a winding duct is very hard to reverse and impossible to inspect.

• De-energise, isolate, verify dead, ground, then clean. Every cleaning task on this list is an electrical task under OSHA 29 CFR 1910.147 (control of hazardous energy) and an approach-boundary task under NFPA 70E. [Confirm: your site's LOTO procedure and PPE category.]

• Check the enclosure, not just the windings. Gasket condition, door seals, cable gland sealing, and any filtered vent or louvre — a leaking enclosure defeats every cleaning interval you set. If the unit sits outdoors or in a semi-exposed plant room, see what an outdoor enclosed cast resin unit has to include, because the enclosure is doing the environmental work there.

6. Load factor: why high load is not the main driver, and low load can be worse

The original guidance treats load factor above 90% as the trigger for shorter intervals. That is only half right, and the missing half is more important.

What high load actually does. Continuous operation above 90% pushes the winding hotspot up. Thermal ageing of the epoxy insulation system is what consumes insulation life, and the rate is strongly non-linear with hotspot temperature — for oil-immersed cellulose the rule of thumb is roughly a doubling of ageing rate for every [Confirm: 6–8 K] rise, and epoxy has its own curve under IEC 60076-11 thermal classes. So yes, sustained high load deserves attention. But the correct response is usually not "inspect more often" — it is:

• Confirm the unit is not simply undersized. A transformer running continuously above 90% is a sizing problem, and the fix is capacity, cooling, or load management, not a tighter maintenance card.

• Instrument the hotspot. A winding temperature indicator with alarm and trip contacts, or a fibre-optic hotspot probe on large units, gives you the actual quantity that matters.

• Check ventilation first. Most "high load" temperature complaints in buildings are ventilation complaints wearing a load-factor disguise.

What low load does. This is the counterintuitive part, and it is the one worth arguing with your team about:

The dry-type transformers that fail young from environmental causes are usually the lightly loaded ones, not the heavily loaded ones. A unit at 15–25% load runs close to ambient temperature, never dries itself out, and sits in whatever humidity the room provides. A unit at 80% load runs warm enough to keep condensation away. If your site has one transformer that is oversized for a future phase and one that is near capacity, the oversized one is the one that needs the moisture and cleaning attention.

This is also why "reduce the interval when load is high" is an incomplete rule. The full rule is: reduce the cleaning and sealing interval when the environment is dirty or wet, and reduce the thermal monitoring interval when the load is high. Those are two different actions driven by two different variables.

Harmonic load deserves one line: harmonic current produces additional winding eddy loss and can push hotspot temperature well above what the RMS load factor suggests. If your load is dominated by VFDs, rectifiers or large UPS systems, size and maintain on thermal consequences, not on kVA. [Confirm: K-factor or harmonic spectrum assessment per IEEE C57.110 / IEC 61378-1.]