RCD testing schedule for data centres in Australia

RCD testing is the most frequently scheduled electrical test in a data centre, and the most awkward one. Every trip drops whatever is on that circuit, the hall is full of equipment that leaks current to earth by design, and the operator and their insurer both want to see a schedule that is followed and a record for every device.

This guide covers how often RCDs in an Australian data centre are typically tested, what operators and insurers usually ask for, what each record should hold, and how the choice between Type A, Type B and Type S devices plays out on server-rack feeds.

Where the RCDs are

A typical site has RCDs in more places than the data hall itself:

• General-purpose socket-outlet circuits in halls, corridors, offices, the network operations centre, build rooms and workshops.
• Loading docks, plant rooms, car parks and EV charging bays.
• Rack feed circuits, where the design includes them. Whether a given rack outlet circuit has a 30 mA RCD is a design decision made under AS/NZS 3000, so test what is installed and raise anything that looks wrong as an observation rather than changing it on the spot.
• Time-delayed (Type S) devices on submains, used so an upstream device does not trip before the 30 mA device on the faulty circuit.
• Portable RCDs on contractors’ leads and power boards.

Two tests, two intervals

RCD testing has two parts, and they run on different cycles.

The push-button test confirms the mechanism still trips. It is quick and needs no instrument, but it still drops the circuit.

The trip-time test uses an instrument to inject the rated residual current and measure how long the device takes to open, on both half-cycles. It is done by a competent person, and its result is the number that goes on the record. The limit at rated current depends on the device’s rating; our RCD trip time limit lookup gives it for 10, 30, 100 and 300 mA devices.

How often: the standard and the contract

AS/NZS 3760 publishes indicative intervals for both portable and fixed RCDs, by environment. For fixed RCDs in a clean, non-hostile environment like an office (which is how most of a data hall’s white space is treated) they work out to a push-button test about every six months and an instrument trip-time test about every two years. Harsher areas such as workshops, loading docks and damp plant rooms move the trip-time test to roughly yearly, and construction zones during a fit-out follow AS/NZS 3012, which is much tighter again. The standard calls those figures indicative, and a documented risk assessment by the responsible person can vary them, so check the current edition for the exact table.

In practice the operator’s maintenance contract usually sets the real schedule, and it is commonly tighter than the standard. A typical data centre schedule looks like this:

What operators and insurers ask for

In most states and territories the model WHS Regulations require RCDs at a workplace to be tested regularly by a competent person, with a record kept until the device is next tested or taken out of service. Victoria runs under its own OHS legislation, and some states add their own electrical safety rules for particular workplaces, so check your jurisdiction.

Operators and insurers tend to go further than the regulation. Expect to be asked for:

• A written schedule that names every RCD on site and its test cycle.
• A result for every device on every cycle, not a sample.
• Evidence of who tested, with which instrument, and that the instrument was in calibration on the day.
• Failed devices listed with the action taken and the retest result.
• Records kept for years, not months. The regulation only asks for a record to be kept until the device is next tested, but many operator contracts ask for five years or longer.

Testing RCDs on a live data hall

A trip on a rack feed drops that feed, so RCD testing runs under the same method of procedure as any other live work. Confirm the redundant feed is healthy and carrying the load, deal with any single-corded equipment first, and test one circuit at a time.

Measure standing leakage first. IT power supplies leak a small current to earth through their filters, and a rack of dual-corded servers can add up to a meaningful share of a 30 mA rating. The tester’s injected current adds to or subtracts from that standing leakage depending on phase, so trip times can read faster or slower than the device really is, and a half-rated no-trip check can trip. A clamp leakage reading before the test explains odd results. A common design rule of thumb keeps standing leakage below about a third of the rated residual current; circuits well over that are worth flagging to the operator as a nuisance-tripping risk.

Type A, Type B and Type S on server-rack feeds

The RCD type marking describes which kinds of residual current the device can detect. It matters in a data hall because the loads are electronic, and some can produce residual currents an older device can’t see.

Why it matters. Smooth DC residual current can stop a Type A device from tripping when it should. Where a load can produce it, the design needs a Type B device or another approved means of protection. Recording the type marking on every periodic test is how that gets noticed when a tenant changes what is plugged in.

Type S is a different axis. The S marking means selective, or time-delayed. It sits alongside the waveform marking (a device can be Type A and S, or Type B and S), and it is used upstream so that the 30 mA device closest to the fault trips first. It is never the 30 mA additional protection device on a final circuit. A healthy Type S device is allowed to take longer than the 300 ms general limit at rated current, up to 500 ms under the device standards, so judge it against the manufacturer’s data rather than failing it on time alone.

Testing note. Type B devices need a tester that can inject smooth DC test current; a standard AC or pulsed test only proves part of their function. Record which test you ran.

What each RCD record should hold

• Board and circuit, with the device’s location (hall, row, panel).
• Device rating, rated residual current and type marking.
• Whether it is general or selective (Type S).
• Push-button result, and trip times at both half-cycles.
• Any additional tests you ran, such as five times rated current or a DC test.
• Tester’s name, the instrument and its calibration due date.
• Test date, next due date, and any defect with the action taken.

Documenting the work in Field Efficient

Field Efficient records RCD results the way operators ask for them.

• One board per test session, so every RCD sits under the board it belongs to and the report is grouped the same way.
• Trip times judged as you capture. Both half-cycles are checked against the limit for the device’s rated current. The type marking (AC, A, F or B) is recorded, and Type AC carries a replacement advisory. A Type S device is recorded as selective and is not auto-failed on time.
• Any tester. Import results from your tester’s software export or a photo of the sheet, whatever the brand, or connect a supported tester directly.
• Calibration register. The instrument used is recorded with each result, and you get a reminder before its calibration falls due.
• Signed reports anyone can check. The operator can open the report’s verification link to see who signed it, when, and confirm their copy is unaltered.
• Works with no signal, and portable RCDs are covered in the test and tag workflow, with QR asset tags printed at the board.

Related guides

• Data centre periodic verification checklist
• Test and tag in data centres
• AS/NZS 3000 verification: the tests every installation needs
• Test and tag intervals: the AS/NZS 3760 schedule

This guide is general information, not compliance advice. Always work to the current editions of AS/NZS 3000, AS/NZS 3017 and AS/NZS 3760, the site’s method of procedure, and your state or territory’s WHS and electrical safety requirements.