Mitigating Fire Risks in Mission Critical Facilities

Overview

The paper starts with detection. A spot smoke detector nominally covers about 900 ft2 (84 m2), but a table taken from NFPA 72 shrinks the spacing as air changes rise: 125 ft2 (11.6 m2) at 60 ACH, 250 ft2 (23 m2) at 30 ACH, 500 ft2 at 15 ACH and 900 ft2 at 7.5 ACH or less. The industry default in high-airflow rooms is one detector per 250 ft2 (23 m2). Heat and flame detectors are not recommended because they give no incipient-stage warning.

Air-sampling (VESD) systems cover roughly 2,500 to 80,000 ft2 (232 to 7,432 m2) per detector network, are up to about 1000 times more sensitive than spot detectors and detect down to about 0.003 %/ft obscuration. On suppression, foam and dry chemical are unsuitable. Among clean-agent total-flooding systems, halocarbons (FK-5-1-12, HFC-227ea, HFC-125) discharge in 10 s or less; inert gases (IG-541: 52% N2 / 40% Ar / 8% CO2; IG-55: 50% Ar / 50% N2) take about 60 s, need about 10 times the storage volume, and cylinders can sit up to about 300 ft (91 m) away.

CO2 total flooding must not be used in occupied spaces under NFPA 12, needs about 4 times the Halon 1301 storage volume and a 30 to 60 s pre-discharge warning. Sprinkler fusible links open at about 165 to 175 F; pre-action is the recommended type if the authority requires sprinklers. Installation cost was about $1 to 2/ft2 for new build and $2 to 3/ft2 for retrofit in 2010 USD. Water-mist droplets are about 100 to 120 microns, and post-discharge humidity may cause IT failures.

The best-practice list says no single detector may trigger discharge, a second confirming detector is required, detection should be both below the raised floor and at the ceiling, and the fire alarm must not trigger EPO. NFPA 75 permits IT equipment to stay energized during gaseous-agent discharge subject to a risk assessment. The paper dates from 2011, so the costs are old but the selection logic still holds.

Key points

1. Spot detector spacing shrinks with air changes: 60 ACH 125 ft2 (11.6 m2), 30 ACH 250 ft2 (23 m2), 15 ACH 500 ft2, 7.5 ACH or less 900 ft2.
2. Industry default in high-airflow rooms is one spot detector per 250 ft2 (23 m2).
3. Air-sampling systems cover 232 to 7,432 m2, are up to about 1000x more sensitive, and detect about 0.003 %/ft obscuration.
4. Heat and flame detectors are not recommended for data centers because they give no incipient-stage warning.
5. Halocarbon discharge takes 10 s or less and inert gas about 60 s; inert gas needs about 10x the storage volume, with cylinders up to about 91 m away.
6. CO2 total flooding is barred from occupied spaces (NFPA 12) and needs about 4x Halon 1301 volume and a 30 to 60 s warning.
7. Sprinkler links open at about 165 to 175 F; pre-action is recommended where sprinklers are required.
8. No single detector may trigger discharge, and the fire alarm must not trigger EPO.

Using it in design

• Default to aspirating detection for rooms above 15 ACH and spot photoelectric otherwise; compute detector count as floor area divided by the area per detector for the ACH band.
• Require separate detection zones below the raised floor and above/below the ceiling.
• Hold agent type, discharge time (halocarbon 10 s, inert 60 s), storage volume factor and occupied-space suitability (CO2 no) in the suppression catalog.
• Run rule checks for cross-zoned/sequential release, an abort station per exit, room sealing and EPO not interlocked to the fire alarm; flag foam and dry chemical as invalid for white space.

Who should read it

Design engineers specifying data center fire detection and suppression, and facility managers negotiating with the authority having jurisdiction.

Related standards

• NFPA 75
• NFPA 76
• NFPA 2001
• NFPA 72
• NFPA 12
• EN 15004
• ISO 14520
• EN 50600-2-5

Publication details

Authors

Victor Avelar

Publisher

Schneider Electric

Year

2011

Document number

WP 83

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