Data Center Load Calculation: From IT Load to Utility Demand
Every data center electrical design starts with a load number, and most early mistakes are in that number. Here is how IT load becomes facility load and then utility demand, using a worked example with stated assumptions.

The example and its assumptions
Illustrative example, not a project. IT load: 1,200 kW. Design PUE: 1.4 (an assumption to be confirmed by the cooling and electrical design). Power factor at the transformer: 0.95. Transformer sizes follow the standard IEC 60076 series (…, 1,600, 2,000, 2,500 kVA …).
Load categories
- Critical IT load: servers, storage and network equipment, fed via UPS.
- Mechanical load: chillers, pumps, CRAH/CRAC units, fans, cooling towers.
- UPS and distribution losses: UPS inefficiency, transformer and cable losses.
- Building services: lighting, small power, lifts, BMS, fire and security.
Quick estimate with PUE
Total facility power = IT load × PUE = 1,200 × 1.4 = 1,680 kW.
This is a concept-stage number. It hides the split between categories, so it cannot tell you the generator load (which excludes non-essential loads) or the transformer loading per bus. Replace it with a load schedule as the design matures.
From kW to transformer kVA
Apparent power = 1,680 ÷ 0.95 = 1,768 kVA.
For two transformers in an N+1 arrangement, each must carry the full demand when the other is out. A 2,000 kVA unit covers 1,768 kVA, so each runs at 1,768 ÷ 2,000 = 88% after a failure, and about 44% when the two share the load. Growth allowance is a separate decision: if the facility will grow, size for the final load or leave space for a later transformer. See the transformer sizing calculator.
Moving to a load schedule
A load schedule lists each load with its rating, power factor, load factor and demand factor, then sums them per distribution board and per source. Apply demand factors only where the load behaviour justifies them: continuous IT and cooling loads in a data center have high demand factors, close to 1.0 for the base load.
Common mistakes
- Using rack nameplate as the IT load without saying so.
- Using a PUE that the cooling design cannot achieve.
- Forgetting that the generator and the utility see different load sets.
- Mixing kW and kVA between load schedule and transformer sizing.
- Leaving growth, margin and redundancy unstated so they get counted twice or not at all.
Related
PUE explained, MV/LV distribution architecture and generator sizing worked example.
Written by the Vision Matrix Institute editorial team. Worked examples use stated, illustrative assumptions; check them against your project data, the applicable standards and manufacturer datasheets before use.
Frequently Asked Questions
How do you calculate the electrical load of a data center?
Start from the planned IT load in kW. Add the electrical load of cooling, UPS and distribution losses, lighting and other building services. A quick estimate multiplies IT load by the design PUE; a detailed one builds a load schedule line by line with demand factors.
What is the difference between connected, demand and design load?
Connected load is the sum of equipment ratings. Demand load is what is expected to run at once, after applying demand or diversity factors. Design load is the demand load plus the allowances (growth, margin) the design commits to.
How does PUE relate to total load?
PUE is total facility power divided by IT power, so total facility power equals IT power times PUE. A design PUE is an assumption you choose and later verify against the cooling and electrical design.
Is IT load the same as rack nameplate?
No. Nameplate is a maximum. Actual IT draw is lower, and the design should state whether the figure is nameplate, expected average or peak, and how the rack power density was derived.
Want to learn this properly?
This topic is covered in depth in our Electrical Design – Data Center Specialist program.
Ask us on WhatsAppLearn to do this on a full project in the Electrical Design – Data Center Specialist program, or read the data center design guide.