Search for emergency lighting testing in New Zealand and you will be told, repeatedly and confidently, that the annual test is a 90-minute discharge.
For a lot of buildings that is wrong. And it is wrong in the expensive direction — you may be paying for a longer test than your building requires, or worse, running a 90-minute test on a system designed for 30 and condemning batteries that were never meant to last that long.
The 90-minute myth
Ninety minutes is a real number. It is just not the only number, and it is not the default.
The figure gets repeated because a lot of New Zealand content on this subject is lifted from Australian sources, where the maintenance regime is structured differently, and because 90 minutes is the duration that appears most often in supplier marketing. Once a number is on fifty websites it becomes the truth by weight.
The actual duration comes from the New Zealand Building Code, and it varies by building.
What clause F6 actually sets
Building Code clause F6 — Visibility in escape routes requires that features in escape routes remain reasonably visible when the main lighting fails. The Acceptable Solution, F6/AS1, sets out how long that has to last, and it does so by risk group.
In broad terms:
| Situation | Duration required |
|---|---|
| Occupants must remain until main lighting is restored, or evacuation time over 90 minutes | Maintained until supply restored |
| Escape height over 150 metres, evacuation time between 30 and 90 minutes, or occupant load over 1000 | 90 minutes |
| All other buildings | 30 minutes |
So the large office tower, the stadium, the big hospital: 90 minutes or maintained. The two-storey suburban office block, the retail unit, the small warehouse, the community hall: commonly 30 minutes.
This is a summary of the direction of F6/AS1, not a reproduction of it, and the classification of a specific building is a design decision made by the fire engineer or designer at consent stage. The number that governs your building is the one on your compliance schedule. Do not set your maintenance regime from this table — use it to know what question to ask.
Which duration applies to your building
There is one authoritative answer and it is not on the internet. It is on your compliance schedule, the document the council issued with your code compliance certificate.
The compliance schedule names, for Specified System 4:
- The performance standard the emergency lighting was designed and installed to
- The inspection and maintenance standard — commonly AS/NZS 2293.2:2019, but older buildings may sit on AS/NZS 2293.2:1995 or even NZS 6104:1981
- The frequency of each procedure and who must carry it out
If you do not have your compliance schedule, request a copy from your territorial authority. This is the single most useful thing a new building owner or property manager can do, and it is free. We have taken on buildings where the owner had been paying for 90-minute testing for years on a 30-minute system, and buildings where the reverse was true and the IQP had been signing anyway.
Older buildings are the interesting ones. A building consented in the 1980s may legitimately sit on NZS 6104:1981, and the maintenance procedures on its schedule will look nothing like a modern one. That is not non-compliance. That is the schedule doing its job.
Why the discharge test must run the full duration
Whatever your duration is, the discharge test has to run it. Here is the engineering reason, which is more persuasive than the compliance reason.
An emergency lighting battery does not fail gradually and visibly. It fails by losing capacity while continuing to work perfectly for short periods. A fitting with a badly degraded battery will illuminate instantly when you flick the test switch, hold up beautifully for five minutes, and collapse at twenty-two.
That is precisely the failure the monthly function test cannot see and the full-duration test exists to find.
Shortening the test to save an evening removes the only part of the year where that failure is detectable. A 30-minute test on a 90-minute system tells you the battery holds a third of its required capacity. It does not tell you anything about the other two thirds, which is where the failures live.
The check at the end is not "is it still on". It is whether the fitting is still producing adequate light at the end of the rated duration. A fitting limping along at a fraction of its output at minute 88 has failed, even though the lamp is lit.
What happens during and after the test
Practical points that catch people out.
The building is less protected during the test. You are deliberately running the emergency system flat. Schedule it accordingly — not before a busy day, not during an event, and ideally at a time when the building is lightly occupied.
Recharge takes considerably longer than discharge. Until the batteries are back to full charge, the system will not deliver its rated duration. Running a discharge test on Thursday evening does not give you a fully protected building on Friday morning. Two discharge tests close together is a genuinely bad idea.
Results are recorded per fitting. Not "system passed". Each luminaire and exit sign, individually, with its result. A site-level pass is not a record — it is an opinion.
Failures get replaced and re-tested. A battery replaced during the round needs its own test before it is signed off, and the replacement should be recorded against the fitting so you can see the age of every battery in the building at a glance.
Batteries have a service life regardless of test results. They are a consumable. A fitting that passes today with a battery approaching the end of its rated life is a planned replacement, not a pass to be forgotten for twelve months. Budgeting for staged replacement is far cheaper than replacing half a building's fittings in one panicked round after a failed BWoF.
Reading your compliance schedule
If you have the document in front of you, this is what to look for under Specified System 4:
- Performance standard — which version of AS/NZS 2293 or which older NZ standard the system was designed to. This tells you the duration.
- Inspection and maintenance standard — usually AS/NZS 2293.2:2019, Section 3.
- Frequency and responsibility — typically a monthly check by the owner or their agent, plus IQP inspections. Councils differ on whether the IQP attendance is six-monthly and annual, or annual only. Yours says which.
- Reporting — records retained, commonly for the previous 24 months, held on site with the schedule, plus a Form 12A from the IQP each year.
If the count of fittings on the schedule does not match what is on site — and after any fitout, it frequently does not — that is a problem to solve now rather than at BWoF time. It is dealt with by an amendment to the compliance schedule, not by quietly testing a different number of fittings.
Related: how often emergency lighting must be tested, what an IQP is, and why buildings fail emergency lighting on their BWoF.