Issue 19, July 2017
Gaseous Fire Suppression Systems Applied to High Air Flow Environments
By Eric W. Forssell
Information-technology (IT) and telecommunications (telecom) facilities provide critical services that warrant early detection of fires to minimize loss of capability due to a fire. Between the cost of equipment and the potential cost of interrupted business operations, these facilities have a high value. These facilities usually contain a high-density amount of electronics equipment that must be kept cool to prevent thermal failure and to improve longevity of semiconductor devices and magnetic storage devices. Such facilities commonly use high airflow rate cooling in a variety of airflow arrangements.
In an effort to increase the cooling efficiency, a hot aisle/cold aisle configuration is becoming increasingly popular. In this configuration, the cooling air is introduced between alternating rows of equipment racks, referred to as the cold aisle. The air passes through the equipment and is withdrawn from the opposite hot aisle. This configuration limits mixing between supplied cooling air and heated exhaust, which increases cooling efficiency. The use of containment systems to prevent bypass airflow from the cold aisle to the hot aisle further increases cooling efficiency.
NFPA 75, Protection of Information Technology Equipment, and NFPA 76, Fire Protection of Telecommunications Facilities, along with NFPA 2001, Standard on Clean Agent Fire Extinguishing Systems, address the use and application of gaseous fire protection systems in these facilities, although only NFPA 75 has a section that specifically addresses hot aisle/cold aisle configurations and the use of containment systems. Section 5.6.10 of NFPA 75 states that “if the aisle containment prevents the gaseous suppression system from producing the required design concentrations throughout the entire volume served, the gaseous suppression system shall be modified to produce the required concentration throughout the volume served.”
Section 8.4.3 of NFPA 75 states that “hot aisle or cold aisle containment systems shall not obstruct the free flow of the gaseous clean agent to the IT equipment or cooling system serving the contained aisle within an information technology equipment room or zone.” Appendix A of NFPA 75, Section A.8.4.3, says that the flow of agent from a gaseous agent system applied to the entire enclosure should be able to penetrate the IT equipment and create a uniform mixture in most cases. This implies that special handling or treatment of the aisle containment zone as a separate enclosure may not be necessary.
The FM Global Loss Prevention Data Sheet on Data Centers and Related Facilities (FM Data Sheet 5-32) states that the clean system must be designed to provide proper clearance from the containment system boundaries and the proper agent design concentration, both within the containment zone and outside it. This implies the treatment of the space within the containment zone (cold aisle or hot aisle) since a separate enclosure is required.
Neither FM Data Sheet 5-32, NFPA 75, nor NFPA 76 address whether modifications or compensation should be applied to the design of clean agent systems when applied to data centers with high air flow rates. The airflow rate involved in these configurations and the obstructions to the flow of the agent within the protected enclosure may prevent or delay achieving the agent concentration required for the gaseous clean agent system to be effective. The impact of this, or the modifications required in the suppression system design to prevent this impact, has not been fully assessed and is currently left to the system designer to evaluate.
The Fire Protection Research Foundation undertook a project with the goals of (a) developing an understanding of the operational features of datacenters, especially those employing “engineered-aisle-designs,” and associated elevated air flow velocities, that may pose challenges to effective transport of gaseous fire extinguishing agents in accordance with current minimum design requirements, and current field design and installation practices; (b) performing a gap analysis on the topic; and (c) developing a research plan, including a recommended test plan for future work.
The first phase of the project surveyed two existing data centers to provide a basis for design parameters and review the findings of a previous Research Foundation project on smoke detection applied to high air flow environments. The previous project had established cooling air flow rates ranging from 8 to 100 air changes per hour, corresponding to heat loads ranging from 0.5 to 6.5 kW/m2. The surveyed data centers were found to be in the middle of this air flow range, with the smaller of the two having a cooling air flow rate of 30 air changes per hour.
The survey also revealed the presence of dead zones that were excluded from the air flow field in the data centers. This was primarily due to the presence of equipment in the data center that did not require the cooling air flow that was required for the main server racks, areas for future expansion and areas for access to the data center.
Based on these findings, a research plan was developed. This plan took a two-tiered approach. The first tier of the research program would address the effects of the continued air flow on the gaseous agent fire suppression systems without the presence of an aisle containment system. The second tier of the research program would address the effects of the continued air flow on the gaseous fire suppression system with an aisle containment system.
The primary variables to be investigated include: air flow rate, dead air zone volume, presence and configuration of an aisle containment system, and type of agent used (halocarbon or inert gas). Scaling would be addressed by examining two basic geometries. The smaller geometry would be based on a configuration consisting of three rows of servers with a 255 m3 (9,000 ft3) main space volume. The larger geometry would be based on a six-row, two-column configuration with a main space volume of 856 m3 (30,240 ft3).
Both tiers of the research program use CFD modeling to evaluate the effects of the continued air flow on the ability of the gaseous agent fire suppression system to develop a uniform agent concentration throughout the enclosure. A limited experimental program is proposed to assess the accuracy and validity of the CFD modeling performed.
The research plan would be carried out in future phases of the project. More information can be found on Fire Protection Research Foundation website at www.NFPA.org/foundation.
Eric W. Forssell is with JENSEN HUGHES