White paperFree to read2012

Making Large UPS Systems More Efficient

Schneider Electric · Richard L. Sawyer

Schneider Electric WP 108 (Richard L. Sawyer, 2012) shows that specifying a large UPS on one full-load efficiency figure is misleading, and that cost should be computed at the real operating point.

On this page

Overview

The opening comparison: two 1 MW UPS systems with an identical published 93% full-load efficiency, in 2N at 300 kW load, have 10-year loss costs of $293,165 versus $793,021. Field surveys indicate 2N data centres operate at 20-40% of their 2N capacity, and the paper assumes a typical 30% load.

UPS losses are split into no-load (fixed), proportional and square-law components. No-load losses can represent over 40% of all UPS losses, while square-law losses become significant (1-4%) at higher loads. Manufacturers often exclude input filters that reduce quoted efficiency by 0.5-1%; at 30% load a filter costing 3 points raises the annual loss cost of a 1 MW UPS from $32,481 to $42,781. Eco-mode bypass efficiency is typically 98% to 99% versus 94% to 97% for the base UPS.

Topology matters: double conversion UPSs draw 9% to 30% input THD at 0.9 to 0.8 power factor, while delta conversion draws less than 3% at unity power factor. A right-sized modular delta UPS (400 kW at 75% load, 96.9%) compared with an oversized non-scalable double conversion unit (88.7% at 30%) shows a 75% lower 10-year cost of inefficiency at 1N and 80% at 2N. The appendix gives the loss models as fractions of rating: delta y = 0.01081x2 + 0.00640x + 0.01333; double conversion y = 0.03455x2 + 0.00959x + 0.03234. Cooling adds 0.4 kW per kW of loss. The loss curves rest on third-party measurements, and the related standard is IEC 62040-3.

Key points

  1. Two 1 MW UPSs with the same 93% full-load efficiency, 2N at 300 kW: 10-year loss $293,165 vs $793,021.
  2. 2N data centres operate at 20-40% of 2N capacity; 30% is assumed typical.
  3. No-load losses can exceed 40% of UPS losses; square-law losses reach 1-4% at higher loads.
  4. Input filters cut efficiency 0.5-1%; at 30% load a 1 MW UPS annual loss cost rises from $32,481 to $42,781.
  5. Eco-mode path 98-99% vs 94-97% for the base UPS.
  6. Double conversion input THD 9-30% (PF 0.9-0.8); delta under 3% at unity PF.
  7. Right-sized modular delta cuts 10-year inefficiency cost 75% at 1N and 80% at 2N.
  8. Cooling adds 0.4 kW per kW of loss.

Using it in design

  • Store UPS losses as a three-coefficient model (no-load, proportional, square-law) or a 0/25/50/75/100% curve, and record whether quoted efficiency includes input filters or transformers.
  • Compute per-module loading from IT load, redundancy (N, N+1, 2N) and installed modules, then annual and lifetime loss cost with the cooling penalty (default 0.4 kW/kW).
  • Flag operating points below about 30% load and suggest fewer or modular power modules.
  • Show eco-mode efficiency as a separate operating mode, not the default figure.

Monthly newsletter · free

Keep your reading list current

Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.

SoSo CAD

Review your power design in SoSo CAD

Inspect supply paths, equipment loads and UPS failure scenarios in one model.

Request a demo
Single-line diagram in the SoSo CAD power and capacity workspace

Who should read it

Electrical designers specifying large UPSs and comparing operating cost.

Related standards

  • IEC 62040-3

Publication details

Authors
Richard L. Sawyer
Publisher
Schneider Electric
Year
2012
Document number
WP 108
Document language
English
Source type
Manufacturer
Access
Free to read
Licence
Copyright retained; citation only — the publisher’s terms apply to the original document

Citation

Richard L. Sawyer (2012). Making Large UPS Systems More Efficient. Schneider Electric. WP 108. https://download.schneider-electric.com/files?p_Doc_Ref=SPD_VAVR-6LJV7V_EN

All content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.