Fire Suppression Systems Bring Unexpected Risk

Overview

The article starts from an inert-gas test discharge at an ING facility on 10 September 2016 that destroyed dozens of hard drives. The damage came not from the gas itself but from the noise of the discharge, which can push a drive's read/write element off track and disable the drive. The authors use the event to show that fire suppression can itself be a source of risk.

The introduction notes that data centre fires are rare and are usually caused by human error during testing or maintenance, or by electrical failures. The chance of a suppression system operating wrongly therefore deserves attention comparable to that of a real fire. The article compares water, pre-action and inert-gas systems in a data centre context.

On mechanical sensitivity, it relays that a drive's read/write element has to stay within plus or minus 15 percent of track spacing and that noise can push it off track, citing Rawson and Green (2012). It also notes that many facilities are turning to pre-action systems because water reaches the floor only on activation. A mitigation section collects practices aimed at reducing discharge noise.

The limits are that this is a journal article resting on one specific incident, and it offers no numeric design criterion or standard requirement. NFPA 75 and NFPA 2001 are named as related standards. For engineers, its value is as a reminder that choosing a suppression system should weigh the acoustic and water-damage sensitivity of IT equipment as well as fire performance.

Key points

1. An inert-gas test discharge at an ING facility on 10 September 2016 destroyed dozens of hard drives.
2. Data centre fires are rare and usually result from human error in testing or maintenance, or from electrical failures.
3. A drive's read/write element must stay within plus or minus 15 percent of track spacing; noise can push it off track.
4. Many facilities are moving to pre-action systems, which put water on the floor only on activation.
5. The article compares water, pre-action and inert-gas systems and describes mitigation practices.
6. Related standards: NFPA 75 and NFPA 2001.

Using it in design

• In the fire module, when inert gas is selected, require low-noise nozzles or silencers and show a hard-drive acoustic risk warning.
• Offer pre-action sprinkler as the default option.
• Define test and maintenance procedures in the design documents so that the risk of accidental discharge is reduced.

Who should read it

Engineers who select and design data centre fire protection, and facility operators.

Related standards

• NFPA 75
• NFPA 2001

Publication details

Authors

K. Heslin; S. Good; P. Turner

Publisher

Uptime Institute Journal (Uptime Institute)

Year

2017

Document language

English

Source type

Industry body

Access

Free to read

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation