What New Zealand's audit of 182 hospital UPS systems found

What New Zealand's audit of 182 hospital UPS systems found

Health New Zealand has allocated about NZ$4.9 million to replace 84 UPS systems and batteries identified as renewal priorities, after an independent review of 182 large hospital UPS systems recorded ageing assets, incomplete maintenance records and rooms without fire detection. The review and two companion reports were published on 10 September 2026.

1. What was reviewed, and why

The trigger was a battery: a UPS system failed and produced a smoke event in Christchurch Hospital's Waipapa building on 20 January 2026, the second UPS battery incident to cause notable heat and smoke damage after the Canterbury Health Laboratories building on 8 January. A third followed at Middlemore's Tiaho Mai building in March. Aurecon New Zealand did the assessment; revision B of its report is dated 15 June 2026, and Health New Zealand released it in September alongside a generator review and the Nelson Hospital outage investigation.

Scope runs to systems rated 20 kVA and above, managed by the Infrastructure Investment Group — a practical filter rather than a risk boundary, the report notes, since a poorly ventilated 15 kVA unit in a confined space could present a greater risk than a well-managed 60 kVA one.

The register holds far more units than the review scored. The 20 kVA line plus a missing kVA rating put 584 assets outside it.

2. Age and condition

The report grades each unit against a 15-year assumed service life. Fifty-two are older than 11 years; 30 have passed 14 years and sit in the immediate renewal group. Sixteen carry a condition grade of 4 or 5 — visible deterioration or active failure risk — and four sit at grade 5: two in the Waipapa building, damaged in January, two in Tiaho Mai, damaged in March.

Battery design life runs from five to twelve years at 25 °C, ten years the most common, and the report takes 70% of rated design life as the point where replacement should be planned for critical applications. Thirty-eight batteries have passed 55% of design life, 25 of them past the 70% line. Ten graded 4 or 5, eight needing urgent replacement.

3. Maintenance, alarms and the operating environment

Nine UPS units show no evidence of regular maintenance — both at Burwood Hospital, five of the 17 at Christchurch, two in Tiaho Mai — and 35 more carry an unknown status: 44 units, 24% of the fleet, without confirmed maintenance assurance. On the battery side, 50 have no confirmed maintenance status (6 graded as having none, 44 unverified) and 60 no confirmed alarm coverage (42 with no working alarm, 18 unverified).

Forty-four UPS units have no alarms and 17 more are unverified, so 33% of the fleet cannot be confirmed as monitored. The review found alarms installed and later disconnected, and wants reconnection made mandatory and verified.

Seven environmental findings. Percentages use 187 entries in each source chart, not a verified count of unique rooms.

The environment charts each total 187 assessment entries; these should not be interpreted as a count of unique rooms. The scope section separately records 162 rooms inspected on site.

The report records 95 entries as confirmed non-fire-rated, 54 without observable fire detection and 36 without fire suppression. It also identifies gaps in climate monitoring (57), climate control (17), acceptable upkeep (57) and observable seismic restraint (49). Unknown statuses are recorded separately and are excluded from these counts. These findings can overlap and should not be added into a total of affected locations.

Only two installations have VESDA early-warning detection. Exposed battery terminals at several sites also prompted a recommendation for insulated covers or fully enclosed cabinets.

4. What Health NZ has committed to, and what stays open

Announced. The 11 recommendations from the Nelson Hospital outage review are complete, as are all 12 immediate generator recommendations. Twelve mid-term generator recommendations follow by June 2027, and a broader utilities and site-infrastructure assessment runs to 30 June 2027.

What follows from the report. The short-term replacement cohort is 30 UPS units past 14 years and 25 batteries past 70% of design life. The medium-term cohort adds 22 units and 13 batteries graded 4 for age. Replacement starts as a time-based programme and shifts toward a performance-based one as monitoring and load data mature.

Still open. Battery autonomy. The report records wide variance in autonomy periods and traces it to inconsistent sizing at procurement and replacement, but draws that from snapshot load readings, calls it indicative rather than definitive, and attaches no priority replacement tables. It also lists 14 units with no verified installation date, 20 with no condition rating and 39 batteries with no verified age, and treats closing those gaps as risk reduction in itself.

5. Five questions the review makes worth asking

1. What design life is the battery specified to, and how much of it has the fleet consumed? The report plans replacement at 70% of rated design life against a five-to-twelve-year band at 25 °C; ask for both against the temperature the room actually holds.
2. What does the room look like? Detection, suppression, fire rating, climate control and monitoring, housekeeping — review these provisions together, alongside seismic restraint.
3. Are the alarms connected and verified, and what is the monitoring path? Verify the sensing equipment, alarm configuration and notification destination. For related wiring context, see what a battery midpoint senses and what it carries. A midpoint connection alone does not establish that faults will be detected and reported.
4. What is the maximum recorded load, from a meter? The report asks for recorded load data before like-for-like replacement. That discipline is the UPS sizing and selection sequence: measure the load, choose the topology, then fix capacity, redundancy, runtime and room.
5. What protects the terminals and restrains the cells? Insulated covers or fully enclosed cabinets on the terminals; restraint of individual cells inside the rack, not only the rack itself.

The report records no UPS units at Timaru Hospital: 24 V DC rectifiers supply backup power to theatre procedure lighting, with capacity to meet the required two-hour runtime unverified. It rates replacement with appropriately sized UPS units as a high priority.

For a separate hospital application example, see a 120 kVA UPS installation for a CT scanner in Cameroon. This supplier case is not part of the New Zealand review.

FAQ

Does this review apply to hospitals outside New Zealand?

No. It covers Health New Zealand's own estate, its own UPS register and its own remediation programme. The findings — batteries remaining in service beyond the review's replacement-planning threshold, disconnected alarms, and gaps in environmental monitoring or fire detection — offer questions for other facilities teams to consider against their own conditions and requirements.

Were the UPS units themselves the problem?

The review grades whole UPS units on age and on-site visual condition, and does not separate inverter faults from battery faults. The failures it records are battery-related: two Waipapa units that failed to start on a battery fault during an August 2025 brownout, and battery fires at the Canterbury Health Laboratories building in January 2026. Sixteen of the 182 units graded 4 or 5 for condition, four of them at grade 5 after the January and March incidents. The report attributes the portfolio-wide risk profile to asset age, inconsistent maintenance and monitoring, and the operating environment.

What happens to the units below 20 kVA?

The review leaves them unscored: 308 records sit below the threshold and another 276 carry no kVA rating at all. The report recommends completing the missing ratings first, then a targeted desktop assessment of the larger units among them, with physical inspection where practicable. It also asks that every UPS unit get an assigned responsible business unit — 208 records have none.