Blog
Data centers depend on continuous electrical power to keep servers, storage systems, networking equipment, and cooling infrastructure running. Even a short power interruption can affect service availability, data integrity, and business operations. For this reason, choosing the right backup generator capacity is an essential part of data center power system planning.
However, generator sizing for data centers is not simply a matter of adding up server power consumption. The calculation must also account for cooling equipment, UPS systems, electrical losses, starting currents, redundancy requirements, and future expansion. An undersized generator may struggle to support critical loads, while an oversized unit can increase capital costs and operate inefficiently.
This guide explains how to calculate data center generator requirements, evaluate key sizing factors, and select a suitable diesel generator solution for reliable backup power.
The required generator capacity depends on the total electrical demand that must remain operational during a utility power failure. This is usually different from the IT equipment load alone because a data center also needs supporting systems to maintain a stable operating environment.
Before selecting a generator, engineers should identify the following major loads:
Not every connected load necessarily needs to be supplied by the generator. The design team should distinguish between critical loads, essential supporting loads, and non-essential loads that can be disconnected during an outage.
KEYPOWER provides diesel generator solutions for data center applications, where backup power, load capacity, scalability, and generator redundancy are important considerations.
A practical generator sizing calculation starts with the electrical demand that the backup power system must support. The following steps provide a useful starting point for preliminary planning. Final sizing should be verified through an electrical load study and generator performance analysis.
First, identify the equipment that must continue operating during a power outage. Add the real power consumption of the IT equipment, cooling systems, UPS losses, and other essential loads.
Basic formula:
Total Critical Load (kW) = IT Load + Cooling Load + UPS Losses + Other Essential Loads
For example, consider a hypothetical data center with the following electrical demand:
| Load Category | Power Demand |
|---|---|
| IT equipment | 500 kW |
| Cooling systems | 250 kW |
| UPS losses | 40 kW |
| Other essential loads | 60 kW |
| Total critical load | 850 kW |
In this example, the backup power system must support approximately 850 kW of critical electrical demand. This is the starting point, not the final generator rating.
Generator capacity is commonly expressed in both kilowatts (kW) and kilovolt-amperes (kVA). These two measurements describe different aspects of electrical power.
Real power (kW) represents the power consumed by equipment, while apparent power (kVA) accounts for the combined effect of real and reactive power.
Generator sizing formula:
Required kVA = Total Load (kW) ÷ Power Factor
Assuming a power factor of 0.8, the example above gives:
850 kW ÷ 0.8 = 1,062.5 kVA
This means the generator system needs to provide at least 1,062.5 kVA under the assumed operating conditions, before accounting for additional engineering requirements.
The actual power factor should be determined from the electrical design rather than automatically assuming 0.8 for every project.
Cooling equipment, pumps, fans, and other motor-driven systems can draw significantly higher current during startup than during normal operation. These temporary increases in demand can affect generator voltage and frequency stability.
Engineers should evaluate motor starting methods, the largest motor starting sequence, UPS rectifier characteristics, and load step acceptance. Variable frequency drives and soft starters may help reduce starting demand, but their effects should be included in the overall electrical analysis.
Generator selection should therefore consider both steady-state kW/kVA requirements and transient performance. A generator that meets the calculated running load may still be unsuitable if it cannot handle the required load steps.
Power Usage Effectiveness (PUE) is an important metric in data center energy management. It compares the total facility power consumption with the power used by IT equipment.
PUE Formula:
PUE = Total Data Center Facility Energy ÷ IT Equipment Energy
For example, if a data center has an IT load of 500 kW and a PUE of 1.6, its estimated total facility demand is:
500 kW × 1.6 = 800 kW
This provides a useful high-level estimate of facility power demand. However, PUE should not replace an equipment-level load schedule when sizing a backup generator. The actual standby load may differ depending on which cooling systems, UPS equipment, and auxiliary services are connected to the emergency power system.
For accurate generator sizing, engineers should use measured or specified electrical loads and verify the operating configuration during utility failure.
Generator capacity is only one part of a reliable data center backup power system. The number of generator sets and their operating arrangement are equally important.
An N configuration provides exactly the generator capacity required to support the designated critical load. For example, if the facility requires 2 MW of backup power, the installed generator capacity must provide at least 2 MW under the specified operating conditions.
This arrangement can meet the load requirement, but a generator failure or planned maintenance may reduce the available backup capacity.
An N+1 configuration includes one additional generator beyond the number required to support the critical load. If one generator becomes unavailable, the remaining units can still provide the required capacity, provided that the system is properly designed.
For example, a facility requiring three generators to support its critical load may install four units. The additional generator provides reserve capacity for maintenance or equipment failure.
A 2N arrangement uses two independent power paths, with each path designed to support the full critical load. This provides a higher level of redundancy than a basic N configuration, although the final level of resilience also depends on electrical distribution, switchgear, fuel supply, controls, and other shared infrastructure.
The appropriate configuration depends on the facility's availability objectives, budget, maintenance strategy, and applicable data center design requirements. Simply adding generators does not guarantee redundancy if critical components remain shared between systems.
Once the required capacity and redundancy arrangement have been established, the next step is to select generator sets with suitable performance ratings and operating characteristics.
KEYPOWER offers a range of industrial diesel generator sets across different power ranges and engine brands. For larger data center projects, generator selection can be evaluated according to prime power, standby power, voltage, frequency, power factor, fuel consumption, and installation conditions.
For example, the KEYPOWER KPX-M938P is an MTU-powered diesel generator with the following published specifications:
| Model | KPX-M938P |
| Prime Power | 750 kW / 938 kVA |
| Standby Power | 825 kW / 1,031 kVA |
| Engine | MTU 12V 2000 G85 |
| Frequency | 60 Hz |
| Power Factor | 0.8 |
These ratings illustrate the importance of distinguishing between prime and standby power. The correct rating must match the intended operating duty and project specification. A standby rating should not automatically be used as the basis for continuous or prime-power operation.
For projects requiring higher capacity, the KEYPOWER MTU diesel generator range includes larger models for evaluation. Final selection should be based on the complete load profile and required redundancy rather than rated output alone.
A generator's published output is based on specified reference conditions. Actual available power can be affected by installation altitude, ambient temperature, humidity, cooling airflow, fuel quality, and exhaust backpressure.
For data centers located in hot climates or at high elevations, engineers should request site-specific performance data from the generator manufacturer. Cooling system design, ventilation, exhaust routing, and acoustic requirements should also be checked before finalizing the generator room layout.
Other practical considerations include:
These factors can influence the final generator configuration and should be reviewed during the engineering stage, not after equipment delivery.
After installation, generator performance needs to be verified under controlled load conditions. Load bank testing allows operators to evaluate generator output, voltage regulation, frequency stability, cooling performance, and overall operating behavior.
It is particularly useful when the actual data center load is not yet available or when commissioning a newly installed generator system.
Depending on the testing plan, resistive or reactive load banks can be used to simulate different electrical load characteristics. KEYPOWER provides load bank equipment for generator testing applications.
Testing should follow the generator manufacturer's recommendations and the project's commissioning requirements. Regular testing and maintenance help identify potential issues before an actual utility outage occurs.
Even when the basic load calculation is correct, several planning mistakes can affect the reliability of a data center backup power system.
A detailed load assessment, clear redundancy strategy, and manufacturer-supported performance review can help prevent these problems.
Choosing a backup generator for a data center requires more than matching a power rating to a single load figure. IT demand, cooling requirements, UPS characteristics, power factor, transient loads, redundancy, and future expansion all influence the final system design.
KEYPOWER supplies diesel generator sets across a broad range of power capacities, including large-output MTU-powered models. By reviewing the project's load schedule, operating conditions, required voltage and frequency, and redundancy strategy, the appropriate generator configuration can be identified more accurately.
If you are planning a new data center, expanding an existing facility, or upgrading its emergency power infrastructure, contact KEYPOWER with your project requirements to discuss a suitable generator solution.
Start by calculating the total critical electrical load, including IT equipment, cooling systems, UPS losses, and essential auxiliary loads. Convert the required kW to kVA using the actual power factor, then evaluate motor starting demand, transient performance, redundancy, and site conditions before selecting the generator.
The number depends on the facility's critical load and redundancy design. An N configuration provides the required capacity, N+1 adds one additional generator, and 2N provides two independent power paths, each designed to support the full critical load.
kW measures real power, while kVA measures apparent power. The relationship is kW = kVA × power factor. For example, a generator rated at 1,000 kVA with a 0.8 power factor can provide 800 kW of real power under its specified operating conditions.
N+1 redundancy provides one additional generator beyond the number required to support the critical load. This helps maintain the required power capacity if one generator is unavailable due to maintenance or equipment failure, provided the remaining system components are properly designed.
The choice depends on the generator's operating duty and the project's specifications. Standby power is intended for emergency operation during utility interruptions, while prime power is designed for applications with varying loads and extended operating hours. The generator rating must match the expected duty.
Load bank testing simulates electrical demand to verify generator performance, including output capacity, voltage regulation, frequency stability, and cooling behavior. It is especially useful during commissioning and routine maintenance when the actual facility load may not be sufficient for testing.