top of page

Data Centre Resilience Starts at the Switchboard

3 hours ago
6 min read
Technician in hard hat and safety vest adjusts circuit breakers in a bright server room, holding a tablet.

A UPS transfer event may keep servers online, yet it does not automatically resolve the condition that caused the event. A sustained undervoltage, repeated voltage steps or distorted supply can continue to affect switchgear, cooling plant, controls and battery charging systems. Effective data centre resilience therefore begins with the complete electrical path - not only the final ride-through system connected to the IT load.

For operators, designers and project teams, the practical objective is clear: maintain supply continuity while keeping voltage and power quality within the operating limits of critical equipment. Achieving that objective requires coordinated decisions about incoming supply conditions, distribution architecture, voltage regulation, power conditioning, UPS topology and ongoing monitoring.

Why data centre resilience is an electrical-system issue

Data centres are often discussed in terms of redundancy: N, N+1, 2N and distribution paths A and B. Those arrangements remain fundamental, but redundancy alone cannot compensate for poor supply quality. Two independent paths can both be exposed to a common upstream disturbance, particularly where they share a utility connection, generator bus, main switchboard or environmental condition.

Voltage variation is a common example. A short-duration sag may be managed by a suitably designed UPS system. However, chronic low voltage can increase current demand for a given load, place additional stress on motors and transformers, and create poor operating conditions for refrigeration and mechanical services. Overvoltage can shorten the life of sensitive electronic components and increase thermal stress. Rapid voltage changes can be just as problematic where controls, contactors and power conversion equipment react unpredictably.

The impact is broader than servers. Cooling is often the first operational dependency to become exposed. Chillers, CRAC or CRAH support equipment, pumps, fans, building management controls, fire systems, security infrastructure and communications equipment each have their own voltage tolerances and response characteristics. A data hall may retain IT supply while its supporting plant begins cycling, alarming or operating outside preferred conditions.

This is why resilience assessments should examine power quality at the point of common coupling, main distribution board, generator interface, UPS input and output, and critical mechanical loads. The question is not simply whether power is present. It is whether the supplied power is suitable for the connected assets under normal operation, transfer events and abnormal network conditions.

Designing data centre resilience around voltage stability

Voltage regulation should be selected after measurement and engineering review, not as a standard add-on. The relevant issues include nominal supply voltage, measured maximum and minimum voltage, load profile, starting currents, fault levels, generator characteristics, harmonic environment, neutral loading and required downstream tolerance.

An automatic voltage regulator can support installations where input voltage moves outside the acceptable range of downstream equipment and a controlled output is required. The response behaviour, regulation accuracy, capacity, bypass arrangement and short-circuit coordination all need to suit the application. In a critical facility, maintainability matters as much as nominal performance. The design should allow inspection, isolation and replacement without creating an unnecessary single point of failure.

Magnetic induction voltage stabilisers may be considered where continuous correction, electrical isolation characteristics and long service life are priorities. Their suitability depends on the operating voltage range, load type, required correction speed and physical installation constraints. They are not a substitute for a UPS where stored energy and outage ride-through are required, but they may form part of an upstream voltage-stability strategy.

Voltage drop compensators address a different but related problem. Long cable runs, heavily loaded distribution systems and remote infrastructure can experience voltage reduction between the supply source and point of use. Where this occurs, compensating for the drop at the right location may improve the operating conditions of connected plant. The alternative - simply increasing conductor size or accepting lower terminal voltage - may be impractical or costly in an existing facility. A proper study should confirm the load variation and ensure compensation does not produce excessive voltage under light-load conditions.

Condition the supply, not just the symptom

Voltage magnitude is only one measure of power quality. Electrical noise, transients, waveform distortion and switching disturbances can affect sensitive controls and communications interfaces even when RMS voltage appears acceptable. AC power line conditioners can be relevant where the supply environment requires additional filtering or isolation, subject to the specific disturbance type and the equipment’s stated performance characteristics.

This distinction is important during fault investigation. Replacing failed power supplies without identifying the upstream electrical condition can lead to recurring incidents. Site records should correlate equipment alarms with voltage data, generator transfers, switching operations and major load changes. Event data from UPS systems, power meters and protection devices can help establish whether an issue is isolated to a load, a distribution board or the incoming supply.

UPS systems remain a critical layer, not the whole design

UPS systems provide essential ride-through during short outages and transfers, and can deliver a regulated supply to protected IT loads depending on topology and operating mode. Their design must account for actual load growth, battery autonomy, recharge requirements, harmonic performance, bypass capacity, maintenance access and coordination with standby generation.

A common mistake is to size the UPS only against current server load. Data centre demand also includes network equipment, storage, security systems and, in some designs, selected control or mechanical loads. Future rack density and staged build-out can materially alter the load profile. A UPS operating comfortably at commissioning may approach a less favourable operating range within a few years if expansion is not incorporated into the original design.

Generator and UPS interaction requires particular attention. Generator voltage and frequency response during step loading can affect UPS input behaviour, battery charging and bypass operation. Frequency converters may be relevant in specialised applications where equipment requires a stable output frequency that differs from, or must be isolated from, the available source. They should be applied based on a defined electrical requirement rather than as a general resilience measure.

Variable transformers also have a role during commissioning, testing, laboratory environments and controlled voltage adjustment applications. They are generally not a replacement for automatically regulated equipment where a critical load needs ongoing correction without manual intervention.

Engineering decisions that prevent hidden weak points

The most resilient data centre electrical systems are designed around realistic operating scenarios. This includes utility voltage excursions, loss of a transformer, generator start and synchronisation, maintenance bypass operation, battery recharge after an outage, high ambient temperatures and progressive load growth.

A disciplined design review should address several connected issues:

  • whether the incoming supply voltage remains within the tolerance required by critical and supporting loads;

  • where voltage regulation and conditioning should sit so they protect the intended equipment without compromising selectivity or redundancy;

  • how UPS, generator, automatic transfer switch and bypass modes behave during transitions;

  • whether harmonic currents, power factor correction and non-linear loads affect transformers, cables and neutral conductors; and

  • how critical equipment will be monitored, maintained and safely isolated over its service life.

These considerations are interdependent. For example, placing a voltage stabiliser upstream of a UPS may improve input conditions, but the arrangement must be assessed for bypass operation, fault clearance, generator compatibility and the effect on both A and B paths. Locating equipment downstream may simplify one part of the system while leaving mechanical loads exposed. There is no universal arrangement that suits every facility.

Specify measurable operating requirements

Performance-based specifications are more useful than generic requests for “clean power”. They should define the expected input supply range, required output voltage tolerance, load capacity, overload capability, efficiency expectations, ambient conditions, ingress protection where relevant, bypass requirements, alarm interfaces and maintenance provisions.

For existing facilities, power-quality monitoring before specification is particularly valuable. A week of data may show little, while several months can reveal seasonal demand effects, recurring utility events or generator-related behaviour that would otherwise be missed. Real-time monitoring and automated alerts can also give facilities teams earlier warning of voltage trends before they progress into an operational incident.

Ashley-Edison Asia's voltage-control and power-protection technologies can be considered across these layers, including automatic voltage regulators, voltage stabilisers, AC power line conditioners, voltage drop compensators, UPS systems, variable transformers and frequency converters. Selection should always be based on documented site conditions, load criticality and coordinated electrical design. As the authorised Australian and New Zealand distributor, OzGreen Energy provides product information and technical support to help engineers and project teams consider suitable Ashley-Edison Asia solutions for their specific requirements.

The practical test of resilience is not whether a data centre has a long list of electrical assets. It is whether those assets operate as a coordinated system when supply conditions are least forgiving. Start with measured voltage behaviour, map the consequences across IT and supporting plant, then specify each layer of protection for a defined purpose.

 
 
 

Comments


bottom of page
Ask me anything