Data Center Fire Suppression Systems: Selection, Detection and Periodic Inspection

In a data centre, a fire is often noticed first as an outage rather than as a flame. A power supply overheats, cable insulation starts to smoulder, smoke mixes into the cabinet airflow and systems begin shutting down one by one. At that point, a suppression system that was poorly chosen or poorly maintained can cause more damage than the fire itself. That is why data center fire suppression systems deserve separate attention from general building fire safety.

The situation field teams run into most often looks like this: an expensive gaseous suppression system was installed when the facility was built, but over the years new cable penetrations were opened, the ceiling void changed, the rack layout grew, and nobody asked whether the system would still work as designed after all those changes. On paper the room is protected; in reality the protection is uncertain.

In this article we walk through the suppression approaches used in data centres, the criteria to weigh when choosing a system, the detection side, room integrity and what needs to be recorded during periodic inspection.

Why data center fire suppression systems need a different approach

On an office floor or in a warehouse, the priority is to knock the fire down as quickly as possible and stop it spreading. Water-based systems are effective and economical for that job. In a data centre the priority list is different: avoiding damage to servers, storage and network equipment during suppression matters as much as putting the fire out. Water splashing onto electronics, or an agent that leaves residue, can cause an outage that lasts longer than the fire itself.

The second difference is the nature of the fire. Fires in data halls are typically low energy, slow to develop and able to smoulder for a long time. Cable insulation, power supplies, uninterruptible power supply (UPS) batteries and dense cable bundles are typical sources. Such fires can develop for a long time before they reach the density that conventional point detectors can pick up. High-volume airflow diluting the smoke makes detection harder still.

The third difference is business continuity. The organisation running the data centre treats an unwanted discharge caused by a false alarm as seriously as a real fire. An unnecessary discharge creates cost and delay for refilling the agent, and leaves the room unprotected for that whole period. That is why suppression, detection and control logic are designed together in data centres; none of them can be considered in isolation.

Gaseous suppression systems: clean agents and inert gases

Gaseous suppression is the most common choice for data halls. These systems flood the protected volume with gas at a defined concentration to suppress the fire, leaving no residue on the equipment. There are two main groups.

Chemical clean agents

Chemical clean agents such as HFC-227ea (known on the market as FM-200) and FK-5-1-12 suppress fire largely by absorbing heat. The cylinders take up relatively little space and the system discharges quickly. When choosing one of these agents, look beyond technical performance to supply continuity and environmental regulation. As rules on fluorinated gases tighten, refilling some agents and sourcing spare parts may become harder. A company taking over maintenance of an existing system should discuss the long-term availability of the agent openly with facility management.

Inert gases

Inert gases such as IG-541, IG-55 and IG-01 work by lowering the oxygen level in the room to a point where the fire cannot continue but brief human presence is still possible. Because they are made of naturally occurring gases, they are not environmentally controversial. On the other hand, they need more cylinders, a larger cylinder room, and a pressure relief vent designed into the room for the pressure rise during discharge. If that vent is forgotten or later blocked, discharge can damage wall and ceiling panels.

In both groups, the system is designed according to the standards that govern gaseous extinguishing systems (for example the TS EN 15004 series or NFPA 2001). What matters for the field team is that the design calculation and project file are kept on site and that whoever performs the maintenance can access them.

Water-based alternatives: pre-action sprinklers and water mist

Some data centres, especially large multi-hall facilities, use water-based systems instead of, or alongside, gaseous suppression. The first concern that comes to mind is water damaging the equipment; special designs are used to reduce that risk.

In pre-action sprinkler systems the pipes are normally dry. Before water can enter the pipework, the detection system must first confirm a fire, and then a sprinkler head must open from heat. This double-confirmation logic prevents a hall from being flooded because of an accidentally broken head or a single false alarm. The point for maintenance teams to watch here is regular checking of the supervisory air in the pipes and the valve. Facilities where low-pressure alarms have been ignored for years are a common sight.

Water mist systems spray water as very fine droplets, giving both a cooling effect and a local oxygen-reduction effect. They use much less water than conventional sprinklers. However, their design is tightly bound to the manufacturer and to tested configurations; changing a nozzle type or layout on site at will can invalidate the system's approved performance.

Water-based systems are often chosen because they are compatible with building-wide protection. In no case, however, should the protection required by building fire regulations be swapped for another system simply on the operator's preference. The choice is made together with an authorised designer and the fire brigade approval process.

Early detection: half of suppression is detection

However good a data centre's suppression system is, it only matters if it is triggered at the right moment. That is why detection is at least as important as suppression.

Conventional optical point detectors can be slow to see smoke in halls with heavy airflow. For this reason, aspirating smoke detection is widely used in data halls. These systems continuously draw room air through sampling pipes to a sensitive detector and can spot smoke long before it becomes visible. Sampling points are usually placed at ceiling level, under the raised floor and at the return air grilles of the cooling units.

Discharge of the suppression system is usually not tied to a single detector. Most designs require confirmation from two separate detection zones or two different threshold levels. The first level raises a warning and alerts staff; the second starts the discharge countdown. This staged approach both prevents unwanted discharge and gives people time to leave the room.

The critical checks for the maintenance team are:

• Sampling holes in the aspirating pipes are not blocked, and airflow readings match the commissioning records.
• The mapping between detection zones and the suppression control panel has been updated whenever the hall layout changed.
• Manual release and abort buttons are accessible, labelled and working.
• The countdown time and the audible and visual warnings can actually be heard and seen in the room.

Room integrity and later modifications

The most overlooked part of a gaseous suppression system is the protected room itself. To extinguish the fire, the gas must stay at design concentration in the room for a certain period. If the room leaks, the gas escapes quickly and the fire can reignite.

That is why a room integrity test (also known as a door fan test) is carried out for gaseous systems. The test shows by calculation whether the room can hold the gas for long enough. Its importance continues after commissioning. Data centres are living buildings: new cable penetrations are opened, ceiling and floor tiles are lifted and replaced, cooling ducts are changed. Each of these changes can compromise room integrity.

Common mistakes seen in the field are:

• Cable penetrations not resealed with fire-rated sealant or pillows after being opened.
• Vents cut into doors afterwards, or door closers removed.
• Cooling and ventilation dampers not connected to the control panel so that they close automatically on discharge.
• Racks or panels placed in front of the pressure relief vent.

First these changes should be recorded, then the integrity test should be repeated if there is any doubt. Finally, the test result and the corrections made should be added to the facility file, which becomes one of the first documents checked at the next inspection.

Criteria for choosing a system

There is no single right answer when deciding on a system for a new data hall or when renewing an existing one. The right choice depends on the facility's conditions. These questions help guide the decision:

• Protected volume and room construction: Can the room's integrity be improved? Will the raised floor and suspended ceiling also be protected?
• Cylinder space: Is there a cylinder room large enough for inert gas, or is a more compact chemical agent needed?
• Occupancy: Are staff present in the hall continuously, or is it mostly unoccupied? How will evacuation time before discharge be ensured?
• Supply and refilling: How quickly and from where can the agent be sourced for refilling? How long would the hall be unprotected after a discharge?
• Regulation and approval: What protection does the building require under the Turkish Regulation on the Protection of Buildings from Fire, and how will fire brigade approval be obtained?
• Maintenance capacity: Are competent teams and spare parts available in the region to carry out periodic maintenance of the chosen system?

The last of these criteria is usually the least discussed, yet in the long run it is often the most decisive. A technically excellent system ends up existing only on paper within a few years if nobody is available to maintain it.

Periodic inspection, testing and record keeping

Data center fire suppression systems provide no assurance once installed unless they are inspected regularly. Inspection intervals are set according to the standard the system was designed to, the manufacturer's instructions and the facility's own risk assessment. In general terms, periodic inspection should cover:

• Checking cylinder pressures and agent quantity (by weighing or level measurement) and comparing them with previous records.
• Physical inspection of cylinder valves, actuators, manifold and pipe connections.
• Function testing of the control panel, detection zones and warning devices; a scenario test with the release circuit isolated so that no discharge occurs.
• Verifying that connected controls (interlocks) such as cooling, ventilation and power shutdown actually operate.
• Tracking the hydrostatic test dates of the cylinders.
• Asking about any changes that could affect room integrity.

The weakest link here is usually record keeping. Cylinder pressure is measured but not compared with the previous reading, so a slow leak goes unnoticed for months. Or the hydrostatic test date sits in a spreadsheet, and tracking breaks down when the person who kept it leaves. Storing each inspection per facility, with the date, measured values, the inspector and any defects found, is essential both for audit readiness and for the reliability of the system.

Another important point for the maintenance company is isolating the release circuit during testing and reconnecting it afterwards. Cases where the circuit was not reconnected after a test and the system stayed disabled for months are a serious risk. Having a second person check and record the post-test status prevents this mistake.

Conclusion

Fire safety in data centres does not end with choosing the right extinguishing agent. The system truly protects only when early detection, two-stage release logic, room integrity and regular inspection are considered together. Every change made to the hall over the years can upset that balance, which is why changes, measurements and test results need to be traceable per facility.

At Yangın Takip, we work so that fire safety companies can track the suppression systems, cylinder and test dates, periodic inspection records and field work at all their sites, data centres included, from a single place.

Related reading

• Yangın algılama sistemi bakımı ne sıklıkla yapılmalı?
• Otellerde Yangın Güvenliği: Gereklilikler ve Denetim Hazırlığı
• Fire Safety Tracking Software