Totally Integrated Power Technical Series Edition 15: DC and AC Power Supply Systems in Data Centres

Overview

The paper starts by looking at where a data centre's energy goes. Critical (ICT) consumption versus non-critical infrastructure consumption is put at about 2:1, so savings potential sits in cooling and auxiliaries as much as in the power chain. That framing shows how small a share of total consumption the DC-versus-AC chain efficiency difference represents. The core critique targets the often-cited 28.2% DC saving from LBNL.

Siemens shows the result depends on an outdated AC reference chain; other comparisons using modern high-efficiency UPS and server power supplies put the difference at roughly 5-7%. The message is that the headline figure is entirely a function of which baseline chain was modelled. The efficiency section notes that modern UPS reach 95% or better, yet UPS load utilisation in redundant systems typically sits near 40%.

Read together, the two facts mean that part-load operating point, not peak efficiency, drives the result, and any DC/AC comparison has to be made at that point. It is a manufacturer publication taking a sceptical stance on DC; it mentions DC standards such as ETSI EN 300 132-3 (380 VDC) but gives no independent field measurements. It remains a short, concrete example of how to interrogate the single number that dominates the DC debate.

Key points

1. Critical (ICT) to non-critical consumption ratio is about 2:1.
2. The widely quoted 28.2% DC saving (LBNL) rests on an outdated AC baseline; other comparisons give about 5-7%.
3. Modern UPS reach 95%+ efficiency.
4. UPS load utilisation in redundant systems is typically about 40%.
5. Related standards: IEC 60364, ETSI EN 300 132-3 (380 VDC), IEC 62040-3.

Using it in design

• When comparing AC and DC distribution, compute chain efficiency from component curves at realistic part load instead of quoting headline savings.
• Flag the 28% figure as outdated and in conflict with newer sources.
• Model the UPS at about 40% load in redundant topologies and use the efficiency at that point, not the peak.
• Show the DC advantage as a 5-7% range together with the assumed chain components in the comparison output.

Who should read it

Designers and decision makers who need to test efficiency claims in the AC-versus-DC distribution debate.

Related standards

• IEC 60364
• ETSI EN 300 132-3 (380 VDC)
• IEC 62040-3

Publication details

Authors

Siemens

Publisher

Siemens

Year

2015

Document number

Technical Series Ed. 15

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