A UPS system that fails during a power cut is not a maintenance problem; it is an event investigation. Facilities managers pull the service records, insurers review the test logs, and the question is always the same: was the UPS maintenance schedule actually followed?
BS 6266:2011 ("Fire protection for electronic equipment installations") is the primary UK guidance document for UPS and emergency power systems in IT and critical facilities. But the standard's maintenance framework applies equally to any site where continuous power is essential: data centres, hospitals, control rooms, broadcast facilities, pumping stations, and industrial process environments. This article covers what BS 6266 requires, how to structure a UPS maintenance schedule that meets those requirements, and what your documentation needs to show.
What BS 6266 covers and what it does not
BS 6266:2011 is a code of practice, not a statutory instrument. It provides guidance on fire protection measures for electronic equipment installations, including UPS rooms, battery rooms, generator sets, and automatic transfer switch gear. It is not the same as the Electricity at Work Regulations 1989 or BS 7671 (the IET Wiring Regulations), both of which impose statutory obligations.
In practice, BS 6266 is the document that:
- Specifies how UPS rooms and battery enclosures should be protected against fire
- Addresses ventilation requirements for VRLA and vented lead-acid battery banks
- Provides guidance on emergency power system testing and maintenance
- Is referenced by insurers, data centre auditors (particularly those working to Tier classifications), and facilities management standards
For UPS maintenance companies, BS 6266 Section 8 and Section 9 are the most operationally relevant. Section 8 covers periodic inspection and testing; Section 9 covers documentation. Understanding both in detail is what separates a service contract that holds up under scrutiny from one that creates liability.
The UPS maintenance schedule: required intervals
A compliant UPS maintenance schedule is not one annual visit. BS 6266 and the supporting guidance from the British Standards Institution, combined with UPS manufacturer requirements, point to a four-tier schedule.
Weekly visual inspection
This is typically performed by the client's in-house facilities team, not the maintenance contractor, but your service contract should define what it covers and who is accountable:
- UPS status display: no fault, alarm, or bypass indicators active
- Battery room or enclosure: no visible swelling, leakage, or corrosion on terminal connections
- Ventilation: intake and exhaust paths unobstructed; no unusual odours (hydrogen gas from vented cells has a distinctive smell)
- Generator status panel: fuel level, coolant level, battery charge indicator, no fault codes
- ATS position indicator: confirm normal supply is feeding the load
Weekly checks take 10–15 minutes at most. The value is catching a low coolant warning or a UPS on bypass before it becomes a site visit.
Monthly battery check
BS 6266 Section 8.3 requires that battery condition be monitored at intervals consistent with the battery type and manufacturer's guidance. For most VRLA (valve-regulated lead-acid) installations, this means a monthly electrical check:
- String voltage measurement under float charge: compare against expected value (typically 2.25–2.30 V per cell for VRLA at 20°C)
- Individual pilot cell voltage check (a cell reading more than ±0.05 V from the mean is a failure indicator)
- Ambient temperature of the battery room: VRLA batteries derate at elevated temperatures; every 10°C above 20°C approximately halves expected life
- Visual inspection for case distortion, terminal corrosion, and electrolyte leakage
For installations with battery monitoring systems (BMS), the monthly check involves reviewing the monitoring data rather than manual cell-by-cell measurements. However, the monitoring system itself requires annual calibration, and this is often missed.
Quarterly load bank test
Load bank testing is the only reliable way to confirm that a generator set or UPS will actually support the rated load for the required duration. An annual functional test at site load is not sufficient to verify capacity; the actual demand may be well below rated capacity, and the test duration may be too short to identify marginal cells or a declining generator under sustained load.
The quarterly schedule for generator sets should include:
- No-load run: 10–15 minutes minimum to check temperatures, oil pressure, coolant level, and exhaust condition
- Load bank test at 100% rated load for a minimum of 30 minutes (BS 6266 recommends testing against full rated load; IEC 62040-3 uses an equivalent discharge test approach for UPS)
- Fuel consumption verification: compare actual consumption during the test against the rated fuel consumption figure; significant deviation indicates injector or governor issues
For UPS systems, quarterly testing should include:
- Full mains bypass and transfer to inverter: confirm transfer time meets the specification (typically under 10 ms for online double-conversion, under 20 ms for line-interactive)
- Battery discharge test to 80% depth of discharge (DOD): 30–60 minutes depending on battery capacity and load
- Recharge time verification: confirm the rectifier returns the battery to 95% charge within the time specified in the system design (typically 10–12 hours for VRLA)
Quarterly tests must be logged with actual measured values, not pass/fail tick boxes. A result of "test passed" tells an auditor nothing about the margin.
Annual full service
The annual service is when the maintenance contractor examines everything the weekly and monthly checks observe at a distance. For a UPS and generator site, a full annual service covers:
UPS:
- Internal inspection: capacitors for bulging, bus bars for tracking, cooling fans for bearing wear
- Firmware and software version check and update
- Calibration of voltage and current measurement circuits
- Thermal imaging of all high-current connections (a loose busbar connection at rated current generates enough heat to cause a fire; BS 6266 explicitly addresses this risk)
- Battery impedance test across every cell or block: compare against the baseline taken at installation or last replacement. Impedance rise of more than 20–25% above baseline is generally treated as a trigger for replacement planning
Generator:
- Oil, fuel filter, air filter, and coolant service
- Load bank test to 110% of rated standby capacity for 60 minutes
- Fuel polishing: diesel stored in bulk tanks degrades; microbial contamination (FAME content in EN 590 diesel accelerates this), water ingress, and sludge accumulation will cause fuel injector fouling. Fuel polishing every 12 months, or after any fuel sample failure, is the correct interval for a critical generator
- Exhaust system inspection: flexible couplings, lagging condition, and backpressure check
- Governor and AVR (automatic voltage regulator) calibration
ATS (automatic transfer switch):
- Contact condition: arcing during switching causes pitting; contacts showing more than 15% impedance rise above baseline should be flagged for replacement
- Transfer time test: measure actual elapsed time from mains failure to ATS operation; verify against site design requirement (typically 10–15 seconds for non-critical loads, under 1 second for some medical and data centre applications)
- Return transfer test: confirm that the ATS returns to normal supply correctly when the mains is restored, without overshooting or hunting
- Contactor cleaning and lubrication per manufacturer schedule
DRUPS systems: additional considerations
Diesel rotary UPS (DRUPS) systems combine a flywheel energy storage unit with a diesel engine on a common shaft. They are increasingly common in data centres and industrial facilities where a conventional UPS battery room is undesirable.
The maintenance schedule for a DRUPS differs from a static UPS + generator combination in several ways:
- Flywheel bearing inspection: every 3,000–5,000 operating hours depending on the manufacturer (typically every 12–24 months); bearing failure on a DRUPS is not a gradual degradation but a sudden loss of the energy buffer
- Clutch and coupling inspection: the mechanical connection between the flywheel and diesel engine must be inspected annually for wear; coupling torque values should be checked against the as-installed specification
- Ride-through duration test: the flywheel provides bridging power for 10–15 seconds (sufficient for the diesel to start and reach operating speed); this test must be performed under full rated load, not reduced load
- Oil analysis: DRUPS units with fluid bearings require oil viscosity and contamination analysis at least annually; results outside tolerance require flushing and refill
BS 6266 does not specifically address DRUPS systems in its current edition, but the general testing principles in Section 8 apply. DRUPS manufacturers (Piller, Hitec Power, Kinolt) publish their own maintenance manuals which should be followed in addition to the standard.
Battery replacement cycles
BS 6266 Section 8.4 requires that battery replacement be planned based on test data, not on a fixed calendar interval. In practice, however, planning for replacement at the following intervals is consistent with the standard's intent and with manufacturer guidance:
| Battery type | Design life (at 20°C) | Planned replacement trigger |
|---|---|---|
| VRLA AGM | 10 years | Impedance over 25% above baseline, or capacity under 80% at discharge test |
| VRLA gel | 12 years | Same criteria as AGM |
| Vented lead-acid (OPzS) | 15–20 years | Capacity under 80% at discharge test, or cell voltage deviation |
| Lithium-ion (LFP, NMC) | 10–15 years | Cycle count and capacity fade per BMS data |
The replacement trigger is test-driven. A battery bank that tests at 95% capacity at year 9 does not need immediate replacement. A bank testing at 78% capacity at year 6, following a thermal event or a period without temperature control, does. Documenting the test data is what makes this defensible.
Documentation requirements under BS 6266
BS 6266 Section 9 specifies what must be recorded and retained. The requirements are stricter than most maintenance companies implement.
What must be recorded after every visit:
- Date and time of visit
- Name and qualifications of attending engineer
- Equipment identification: UPS serial number, battery bank ID, generator set ID, ATS reference
- Test type performed and test conditions (load level, ambient temperature)
- Measured values for each test parameter: not "passed" but the actual voltage, impedance, transfer time, or capacity percentage
- Any defects found, described specifically ("cell 14 in string B shows impedance of 3.2 mΩ against baseline of 2.1 mΩ")
- Corrective actions recommended, with priority classification
- Any work deferred and the reason
What must be retained:
- Full service history for the life of the installation
- Battery baseline test results from commissioning
- All discharge test results with load and ambient temperature data
- Fuel analysis certificates
- Calibration records for test equipment used (insulation resistance testers, battery impedance analysers, power analysers)
Where records must be held:
BS 6266 Section 9.2 requires that a record of maintenance be available at the site. For multi-site operations, this creates an immediate problem: the engineer who attended last week should not be the only person who knows what the test results showed.
The operational reality for companies servicing multiple generator and UPS sites is that paper-based or per-engineer records create compliance gaps. When an auditor or insurer requests the service history for a specific UPS, the record needs to be retrievable against the asset ID, not against a work order number or an engineer's filing system.
RemoteOps stores service records against the individual asset (UPS unit, generator set, ATS, battery bank) so that when an auditor requests the full test history for a specific installation, it can be exported directly. Engineers submit test results including measured values, not tick boxes, from the mobile app at the point of service, which closes the gap between what was measured and what the record shows.
Fuel polishing: the overlooked maintenance item
Fuel polishing is the most commonly skipped item on emergency generator maintenance schedules, and it is the item most likely to cause a generator to fail when it is actually needed.
EN 590 diesel contains up to 7% fatty acid methyl esters (FAME). FAME absorbs water and supports microbial growth. In a bulk fuel tank that is replenished infrequently (as is typical for emergency generators), contamination builds up in layers at the fuel-water interface.
A generator that passes every functional and load test can still fail to start reliably if the fuel in the tank has degraded. The signs are subtle until they are not: slightly elevated fuel filter replacement frequency, occasional rough running on no-load tests, and then a failure to sustain load during an actual power outage.
BS 6266 does not prescribe a specific fuel polishing interval, but most generator maintenance specialists align with the Standby and Emergency Power Association (SEPA) guidance: fuel sampling and analysis at least annually, with polishing triggered by any sample showing microbial contamination, water content above 200 ppm, or particle count outside ISO 4406 Class 16/14/11 for a critical generator application.
For sites where the generator is tested monthly and fuel consumption keeps the tank turning over, annual polishing may be sufficient. For rarely used backup generators or sites in humid climates, six-monthly analysis is more appropriate.
Common failures at audit
When insurers and facilities auditors review emergency power maintenance records, four deficiencies recur most often.
No load bank test results, only run records. A generator that was started and run for 20 minutes at no load has not been tested. The test record must show load applied, percentage of rated capacity, duration, and measured outcomes.
Battery records without measured values. Service reports showing "batteries checked, satisfactory" do not demonstrate that impedance measurements were taken. Without baseline and trend data, battery condition cannot be assessed.
ATS never transfer-time tested. ATS function is tested at installation and then often not again. Transfer time can drift as contacts age. A site specifying under 1 second transfer that is actually operating at 4 seconds under degraded contact conditions will not discover this until it matters.
No fuel sample records. Fuel polishing is typically contracted separately from the UPS and generator service. The result is that no one is clearly responsible, and fuel analysis simply does not happen until a failure makes it obvious.
These gaps are not obscure compliance technicalities. They are the predictable failure points in any maintenance programme that treats the emergency power system as lower priority than the systems it protects.
Frequently asked questions
Does BS 6266 apply outside the UK?
BS 6266 is a British Standard and has no formal force outside the UK. However, it is widely used as a reference document by facilities managers and data centre operators internationally, particularly in markets where a locally developed equivalent standard does not exist. Sites operating under ISO/IEC 27001, Tier certification (Uptime Institute), or EN 50600 (data centre infrastructure) will find that the maintenance principles in BS 6266 are broadly consistent with those frameworks. The testing intervals and documentation requirements in this article are practical regardless of the specific standard that applies to your market.
How often should UPS batteries be replaced?
Replacement should be driven by test data rather than a fixed schedule. VRLA batteries have a design life of 10 years at 20°C ambient, but impedance testing at each annual service provides early warning of degradation. The replacement trigger is typically impedance rising more than 20–25% above the commissioning baseline, or a discharge test showing capacity below 80% of rated. Either condition should initiate replacement planning. Waiting for a battery to fail during a discharge test is not an acceptable approach for a critical power installation.
What is the difference between a functional test and a load bank test?
A functional test confirms that the generator starts, transfers the load, and runs. It does not confirm that the generator can sustain its rated load for the required duration. A load bank test applies a controlled electrical load, typically at 75–100% of rated capacity, for a defined period. Generator manufacturers and BS 6266 guidance require load bank testing because site loads in many installations are significantly lower than rated capacity, meaning the generator never experiences rated conditions during functional tests.
What does fuel polishing involve?
Fuel polishing is a process that recirculates stored diesel through a series of filtration and water separation stages to remove particulate contamination, emulsified water, and microbial growth. It does not replace degraded fuel; diesel with severely elevated acid value or oxidised hydrocarbons requires disposal and replacement. Polishing is most effective as a preventive measure when done before contamination becomes severe. The output of a polishing process should include a before-and-after fuel sample analysis report, which becomes part of the site's maintenance record.
Who is responsible if an ATS fails to transfer during a power outage?
Liability depends on what the maintenance records show. If the ATS was due for annual service and had not been serviced, or if a previous service identified contact degradation and no corrective action was taken, the maintenance contractor faces significant exposure. The best protection is a documented service history showing that the ATS was tested under the correct conditions at the required interval, with measured transfer times recorded. A paper trail showing that a defect was identified and communicated in writing, but that the client deferred the repair, shifts liability to the client.
Related reading: FSM Software for Critical Infrastructure, How to Automate Compliance Documentation, Service Contract Management for Maintenance Companies