For most commercial facilities, a power outage is an inconvenience. For a data centre or server room, it is a crisis. Lost transactions, corrupted data, breached service level agreements and reputational damage can follow within seconds of an unplanned power interruption. When the consequences of downtime are this severe, the specification of backup power systems demands a level of engineering rigour that goes well beyond standard commercial practice.
Diesel generators remain the backbone of standby and emergency power provision in data environments. But selecting the right generator for a data centre or server room is considerably more complex than sizing a unit for a typical commercial building. The electrical characteristics of IT infrastructure present unique challenges that, if overlooked during specification, can result in unreliable performance precisely when dependable power is most needed.
Modern data centres and server rooms are dominated by non-linear electrical loads. Servers, uninterruptible power supplies (UPS), power distribution units and cooling systems all draw power in ways that differ significantly from straightforward resistive or inductive loads. This creates harmonic distortion within the electrical system - a factor that must be carefully evaluated during the load analysis phase of any generator specification project.
Harmonic distortion can cause a generator's alternator to run hotter than its nameplate rating would suggest, reducing output capacity and accelerating component wear. A generator sized purely on paper without accounting for the harmonic profile of connected equipment may perform well below its rated output under real operating conditions. Accurate electrical load calculations must therefore reflect the actual characteristics of the IT infrastructure, not simply the aggregate connected load figure.
Most data environments rely on UPS systems to bridge the gap between a mains failure and the moment the standby generator reaches stable operating voltage and frequency. The interaction between UPS equipment and diesel generators requires careful engineering consideration.
When a generator starts and takes on load from a UPS system, the sudden inrush of current can cause voltage and frequency instability during the critical transfer period. If the generator is not correctly sized and configured for this load profile, it may struggle to stabilise quickly enough, risking a UPS bypass event or even a UPS shutdown - exactly the outcome the system was designed to prevent.
Motor starting currents from cooling plant connected to the same generator supply compound this challenge further. A detailed load analysis that sequences the application of loads, accounts for UPS input characteristics and models the behaviour of all connected equipment is essential to producing a reliable generator specification for any data environment.
Server rooms and data centres generate substantial heat. Cooling systems - whether computer room air conditioning units, chillers or precision cooling equipment - represent a significant proportion of total electrical demand and must be included in any generator sizing exercise.
Cooling loads are often motor-driven, meaning they draw high inrush currents at start-up. Where variable frequency drives are used to control cooling equipment, additional harmonic considerations apply. An independent engineering specialist will model these loads individually rather than applying a blanket demand factor, ensuring the generator is sized to handle both the sustained running load and the transient demands that occur during start-up sequences.
Many data environments require a level of power redundancy that goes beyond a single standby generator. N+1 generator configurations - where one additional generator unit is available above the minimum required to carry the full site load - are common in facilities where continuous availability is contractually or operationally mandated.
Specifying generators within a redundant configuration requires careful consideration of load sharing, automatic transfer switching and the behaviour of the overall system during a fault condition or planned maintenance period. Each generator within the set must be correctly sized to handle the full required load independently, not simply a proportional share of it under normal operating conditions. This is a specification decision that benefits significantly from experienced engineering input.
The technical demands of data centre and server room applications make multi-brand generator sourcing a practical advantage. Engine manufacturers including Perkins, Cummins, Volvo, Doosan and Baudouin each offer generator sets with varying performance characteristics, alternator options and control system capabilities. The ability to evaluate equipment across multiple manufacturers - rather than being limited to a single product range - allows the most technically appropriate solution to be selected for each specific application.
For data environments, alternator specification is particularly important. High-quality alternators with low subtransient reactance characteristics handle non-linear loads and harmonic distortion more effectively, maintaining stable output voltage under the demanding conditions that IT infrastructure creates. This level of detail should form part of every generator specification for a data or server room environment.
No generator specification for a data centre or server room should begin without a thorough site survey. Engineers need to understand the physical layout, existing electrical infrastructure, UPS configuration, cooling plant arrangement, available fuel storage space and any constraints on generator installation. A professional site assessment ensures that the specified equipment can be delivered, installed and commissioned correctly, and that the final system will perform as intended under real operating conditions.
The survey also provides the opportunity to review existing backup power arrangements where a replacement or upgrade project is underway. Many data environments operate with generators that were specified years ago and may no longer reflect current IT loads or cooling requirements accurately. A fresh load analysis often reveals significant changes in actual site demand that warrant a reassessment of the installed generation capacity.
For organisations operating data centres and server rooms, backup power is not a secondary consideration - it is a fundamental component of the facility's resilience strategy. The consequences of generator underperformance in a data environment can include data loss, contractual penalties, regulatory exposure and lasting damage to client relationships.
Correct generator specification, grounded in accurate load analysis and independent engineering advice, is the most effective way to reduce that risk. An engineering-led approach that accounts for the specific electrical characteristics of IT infrastructure, UPS systems, cooling plant and redundancy requirements delivers a standby power system that performs reliably under the conditions it will actually encounter - not just under idealised assumptions.
For data environments where failure is simply not an option, the quality of the specification process is as important as the quality of the equipment itself.
