Industry
UPS for Data Centres & IT Infrastructure | Australia
Data centres are the canonical UPS use case, and the most demanding. Tier classifications (I-IV) drive UPS architecture: from single-path Tier I through to fault-tolerant 2N Tier IV with full dual-bus distribution. Every design decision (topology, redundancy, battery chemistry, bypass strategy) flows from the tier target.
UPS Services designs, installs and maintains UPS infrastructure for new-build and brownfield Australian data centres including the major retail, hyperscale, and edge operators. We work across all major UPS brands, with a particular focus on modular UPS architectures (APC Galaxy VL, Eaton 9395XC, Vertiv Liebert APM2, PowerShield Modular) where pay-as-you-grow scaling is the dominant CapEx model.
Downtime costs in Australian data centres range from $5,000 to $100,000+ per minute depending on the workload class. A single UPS failure in a Tier II facility without proper redundancy can cascade into hours of recovery, not minutes. Our design approach starts with the business risk profile and works backwards to the power topology.
We support the full lifecycle: concept design, equipment specification, procurement, installation, commissioning (SAT), preventative maintenance contracts, battery replacement programmes, and emergency response. One critical-power partner from first kVA to decommission.
Sector challenges
What makes data centres & it infrastructure different.
5 critical design considerations that shape UPS architecture for this sector.
01 / 05
Concurrent maintainability
Tier III+ requires the ability to take any single component offline for maintenance without downtime; this drives dual-path power architecture, external bypass design, and UPS module redundancy. Every service visit must be possible without load interruption.
02 / 05
Lithium-ion adoption
Lithium UPS batteries provide ~3x service life (10-12 years vs 3-5 for VRLA), 40-60% smaller footprint, and better high-temperature tolerance. Increasingly the default for new builds, but compliance with AS/NZS 5139:2019 , governing location, ventilation, fire separation, and BMS monitoring.
03 / 05
High-density rack support
AI/GPU loads at 30-100kW per rack require UPS sized to handle higher inrush currents, tighter voltage regulation (±2%), and non-linear power factor characteristics. Standard UPS sizing calculators undersize for GPU workloads.
04 / 05
Modular vs monolithic
Modular UPS (hot-swappable 25-50kW modules) dominates new builds because it supports pay-as-you-grow CapEx and reduces MTTR from hours to minutes. Monolithic UPS still has a place for known, stable loads where maximum efficiency at full load matters.
05 / 05
Commissioning and SAT
Tier III/IV facilities require structured Site Acceptance Testing with load bank verification, transfer-to-battery tests, transfer-to-bypass tests, and documented oscilloscope traces. This is not a formality: it catches installation errors before they cause live incidents.
Typical configurations
UPS patterns we deploy.
- 01N+1 modular UPS (500-2000kW)
- 022N parallel-redundant arrays
- 03Lithium-ion battery cabinets
- 04External maintenance bypass panels
- 05Dual-utility + N+1 generator
- 06Static transfer switches (STS)
- 07Modular pay-as-you-grow frames
- 08Distributed-redundant topology
Equipment
Recommended for this sector.
Manufacturer-trained installation and service across all major UPS brands.
- APC Galaxy VL (modular, 200-500kW)
- APC Galaxy VS (100-150kW)
- Eaton 93PM (modular, 50-200kW)
- Eaton 9395 (standalone, 225-1100kVA)
- Vertiv Liebert APM2 (modular, 30-300kW)
- Vertiv Liebert EXL S1 (standalone, 100-1200kW)
- PowerShield Centurion Pro (standalone)
- PowerShield Modular (modular, scalable)
- Socomec MASTERYS GP4 (standalone)
When it matters
Real-world scenarios.
What goes wrong without proper UPS, and how the right architecture prevents it.
Scenario 01
Single-module failure in N+0 configuration
A Tier II data centre running without module redundancy loses a single rectifier board. The UPS transfers to bypass within 4ms, but the bypass path is raw mains, and any subsequent mains event during the repair window (2-4 hours for parts) drops the entire hall. N+1 redundancy would have kept the load on conditioned UPS power throughout.
Scenario 02
Battery end-of-life during mains outage
A 5-year-old VRLA battery bank fails to deliver rated runtime during a genuine mains outage because impedance drift was not tracked. The UPS shuts down after 3 minutes instead of the designed 10, insufficient for generator startup sequencing. Quarterly impedance testing would have flagged the degradation 6 months earlier.
Scenario 03
Lithium battery thermal event
A lithium-ion UPS battery cabinet in a poorly ventilated room exceeds thermal limits, triggering BMS shutdown and load transfer to bypass. AS/NZS 5139 compliance (including ventilation, fire separation, and BMS alarm integration) prevents this by design.
Our services
Relevant services for data centres & it infrastructure.
Frequently asked questions
4 questions answered.
Q01
What tier classification does my data centre need?
Tier classification should match your business risk tolerance, not aspirational targets. Tier I (99.671% availability, ~28.8 hours downtime/year) suits development environments. Tier II (99.741%) suits small business hosting. Tier III (99.982%, ~1.6 hours/year) is the minimum for enterprise production. Tier IV (99.995%) suits financial trading, healthcare, and mission-critical government. Each tier step roughly doubles the critical-power infrastructure cost, and we help clients right-size to actual SLA requirements.
Q02
Should I use lithium-ion or VRLA batteries?
For new builds, lithium-ion is the default recommendation: 10-12 year service life (vs 3-5 for VRLA), 40-60% smaller footprint, better performance above 25°C, and lower total cost of ownership across the asset lifecycle. VRLA still makes sense for legacy replacements where the UPS and battery enclosure were designed for VRLA dimensions, or where facility modifications for AS/NZS 5139 compliance are disproportionately expensive. We can model both scenarios with TCO analysis.
Q03
How often should data centre UPS be serviced?
Best practice is quarterly preventative maintenance (PM) with an annual comprehensive service. Quarterly PM includes battery impedance testing, thermal imaging of critical connections, and transfer/bypass functional tests. The annual service adds full load bank testing, firmware updates, and capacitor condition assessment. For Tier III+ facilities, maintenance must be non-disruptive: performed under bypass or on individual modules while the remaining N modules carry the load.
Q04
What is the lead time for three-phase UPS procurement?
Standard lead times for major manufacturers are 8-16 weeks for modular frames and 12-20 weeks for large standalone units. Supply chain constraints can extend this to 6+ months for specific models. We maintain relationships with all major manufacturers and can advise on current availability. For emergency replacements, we can sometimes source from Australian distributor stock within 1-2 weeks.
Specify data centres & it infrastructure