Published on: 


The main fire hazards affecting gas turbine generators are fuel, lubrication and hydraulic oil. Even the best-protected fuel arrangement can result in an external ignition of the fuel with a resulting fire. Similarly, mineral lube and hydraulic oils can result in a fire in the event of a leak or an oil spray which ignites on hot turbine parts.

The most popular fire suppression agent in use today for gas turbines is carbon dioxide (CO2). In general, CO2 systems have a good track record for effective operation. CO2 works primarily by removing the oxygen component of the fire triangle. Its advantages comprise the lack of any residue, low cost and its electrically nonconductive properties. Its disadvantage is its asphyxiation hazard.

The current design and installation standard for CO2 systems is the 2011 Edition of National Fire Protection Association (NFPA) 12, Carbon Dioxide Extinguishing Systems. The CO2 supply can be either from individual high-pressure cylinders or from a refrigerated lowpressure storage tank (Figure 1). Cylinders offer the advantage of off-site filling and lower cost. Storage tanks advantages include no need for hydrostatic pressure testing and a generally larger available CO2 supply. As CO2 volumes increase, the price difference between cylinders and a storage tank decreases.

CO2 for gas turbines is typically applied in a “total-flooding” mode which requires a tight enclosure around the turbine to build up the necessary concentration. Although CO2 can also be applied in a “local application” method without an enclosure, this is not a preferred approach due to the possibility of re-ignition after the CO2 has dissipated. Per NFPA 12, the minimum acceptable design concentration for liquid fuels is 34% by volume. If the fuel is natural gas, the minimum concentration is raised to 37% to provide inerting in addition to flame extinguishment. NFPA 12 further breaks down total-flooding systems into two subcategories of surface fires or deep-seated fires. .

From here on, requirements become more difficult to find, mainly because they are contained in standards other than NFPA 12. Turbine fires can be both surface fires which assume prompt extinguishment (i.e., pools of burning fuel or oil) and deep-seated fires which assume that the fire is not immediately extinguished (i.e., insulation materials, smoldering conditions). This means that the more conservative approach should be used from both subcategories.

The requirement for surface fires is that the design concentration must be reached within oneminute fromthe start of discharge. In accordance with NFPA 12 for total-flooding systems, the concentration “. . . shall be achieved and maintained for a period of time to allow effective emergency action by trained personnel.” The following standards specifically address the duration of the retention time for total-flooding systems as they relate to gas turbines.

In accordance with NFPA 850, the concentration “. . . should be held as long as the hazards of hot metal surfaces above the autoignition temperature [of fuel and/or oil] and uncontrolled combustible liquid flow exist.” Unless this information is known and published by the manufacturer (unlikely), this performance-based statement is of little use to fire protection system designers.

NFPA850 further suggests that this duration is at least 30minutes for large industrialtype turbines, but does not give a quantitative value for aeroderivative turbines. However, NFPA 37 requires aminimumduration of 20 minutes for any type of turbine. Similarly, FM Global Data Sheet 7-79 establishes that the duration should be the greater of 20 minutes or the turbine rundown time plus 10 minutes. Therefore, the retention time should be at least 20 minutes for the lighter aeroderivative turbines and 30 minutes for frame turbines.

The minimum required concentration at the end of the retention time varies according to the source. In accordance with NFPA 12 and NFPA 37, the design concentration must be held for the required duration. Thismeans either 34% or 37% from the discussion above. FM Global, meanwhile, requires that the concentration be at least 30% for the extinguishing period after the initial design concentration has been achieved.

Design strategy

A system that discharges a predetermined quantity of CO2 and then counts on an initial higher concentration to decrease gradually (hopefully not less than the minimum concentration at the end of the required duration) is adequate for short retention times (i.e., 10 minutes or less). This is referred to as “singleshot.”

However, experience has shown that most turbine enclosures do not offer the level of “tightness” required for a single- shot concentration to last 20 minutes or more. Therefore, an extended discharge is a prudent method. This consists of a smaller secondary CO2 system that continues to “trickle” a smaller quantity ofCO2 into the enclosure after the initial discharge is exhausted.The quantity of CO2 provided for the extended discharge should take into account the required duration.

As mentioned above, CO2 carries a unique personnel hazard — the potential for asphyxiation, or death due to a lack of oxygen. NFPA 12 has taken an aggressive stance, beginning with the 2005 Edition, by requiring stringent personnel safeguards both for new and existing systems.

These include a proliferation of ANSI Z535-compliantwarning signs,manual lockout valves, and audible and visual alarms. Although not specifically required, it is highly recommended that procedures be implemented that require any CO2 system to be physically disabled any time personnel enter the protected space.With these safeguards in place, the risk to personnel can be reduced to acceptable levels.

The acceptance test requires a full discharge of the system, including any extended discharges. Particular attention should be paid to louvers and dampers in ventilation systems, as in the author’s experience these have been the leading cause of failed acceptance tests. To test the damper release, the pilot tubing should be disconnected and pressurized with air or CO2. It is important that the test gas pressure not exceed the normal pressure of the extinguishing gas in that particular portion of the system; otherwise, the dampers could release under test gas pressures, but not under actual discharge pressures.

A properly calibrated CO2 analyzer is necessary to measure the actual achieved concentration within the protected enclosure (Figure 2). These are usually available from fire-protection contractors or can be rented. On the day of the acceptance test, it is wise to perform a field calibration on all channels of the analyzer in the presence of the witnesses. This can be done easily and quickly by sticking all probes of the analyzer into a plastic trash bag and filling up the bag with a hand-held CO2 fire extinguisher or a small CO2 cylinder. The analyzer reading should go to 100%.With the probes removed from the bag and exposed to fresh air, the concentration should drop to 0%.

While this field calibration is not required by a standard, it is a quick and easy test which proves that the analyzer results are reliable and that the machine has not been damaged in transit. Next, the probes are placed within the protected space. The exact placement of the probes within the protected space is subjective. Again, there are no hard requirements, but the objective of a totalflooding systemis to protect the entire space, not portions of the space. Therefore, the probe(s) should be placed near the obvious hazard. If more than one probe is available, the probes should be staggered at varying heights around the hazard.

The placement of a probe at the ceiling should be avoided, as CO2 is heavier than air and hazards are seldom found near ceilings. Similarly, placing a probe directly on the floor would not yield representative results, as CO2 settles in low areas. Care should be taken to avoid any kinks or restrictions in the probe tubing.


The actuation of the extinguishing system is typically performed with a manual discharge station located outside the protected space. While actuating the system automatically would be preferred, such actuation is usually impractical and potentially hazardous as it requires personnel to be within the protected space at themoment that the systemdischarges.

During the discharge, the analyzer strip charts should be observed for the obtained concentration. It is also a good idea to thoroughly inspect the protected enclosure and the CO2 system from the exterior to look for any potential problems, such as leaks in piping and in the enclosure. Under no circumstance should any doors be opened for the required extinguishing duration, as this will affect the concentration readings. After the required extinguishing duration, all compartment doors should be opened to vent all CO2 gas fromthe enclosure.AllCO2 systemcomponents, including dampers, should be thoroughly inspected for satisfactory performance. The appropriate annotations should be made on the analyzer strip chart recordings (i.e., date, protected space, exact location of probe, and so on).

Water mist

Watermist is slowly gaining ground on other water-based fire suppression systems and over gaseous extinguishing systems. Water mist has several distinct advantages over CO2, the main one being that it is innocuous to personnel. The other advantage is that unlike CO2 or other gaseous agents, water mist systems are not dependent on a tight enclosure around a hazard.As discussed earlier, the latter is especially advantageous for older units, as enclosure tightness tends to deteriorate with age and the failure rate of mechanical louvers and dampers increases. With that said, water mist still requires an enclosure around the protected hazard. The extinguishing mechanism consists of small water droplets which are effective at cooling and the displacement of air by water vapor.

There are two methods of achieving these small droplets, and their use depends on the system manufacturer. One method consists of pressurizing the water to approximately 1,000 psiwith a positive-displacement pump. The discharge of the pressurized water through very small orifices produces a mist.

The other method consists of routing water and an atomizing gas (usually nitrogen) in separate piping systems at lower pressures to the nozzles.With either system, the mist is so fine that thermal shock to hot turbine parts has not been an issue. Unlike CO2 systems, water mist systems are applicationspecific and are listed or approved for spaces on a maximum volume basis. Most jurisdictions will only accept a water mist system if it meets their approval requirements.

The reason is that the performance of watermist is qualitative, rather than quantitative. In other words, it is not possible tomeasure a performance goal for a particular installation as is done with CO2. Rather, the water mist system is tested under full-scale conditions at an independent laboratory and has been proven to extinguish fire under given circumstances (i.e., fuel, hazard and protected volume)—a costly process.

The two recognized U.S. approval agencies formachinery enclosures are FMGlobal and Underwriters Laboratories (UL). Similarly, the onlymanufacturers with corresponding listings or approvals include Marioff, Securiplex, and Tyco Fire & Building Products. Due to the preceding factors and the extensive use of stainless steel, water mist systems tend to be the most expensive protection option, about two to three times the installation cost of CO2 systems in the author’s experience.

From a design standpoint,watermist systems are largely pre-engineered and preassembled by their manufacturers. The installing contractor merely fabricates the piping and portions of the system for local conditions and may need to perform friction loss calculations to show that minimum required pressures are achieved at the nozzles. There may also be customized plans reflecting local conditions. It is critical that all manufacturer and listing or approval requirements are followed.

Similarly to CO2, the acceptance test requires a full discharge test, but unlike the CO2 test, there is nothing to measure and the test is usually performed with the access doors open to allow visual verification. Because enclosure integrity is not critical for water mist, there is less to be concerned about with water mist than with CO2. The main parameters are discharge pressure at the manifold, discharge fromall nozzles and discharge duration.Themanufacturers’manuals typically contain step-by-step instructions on how to perform an acceptance test.

Halon 1301 was once the fire suppression agent of choice due to its suitability for use in occupied spaces as it is neither toxic nor does it result in an asphyxiation hazard when used at its normal design concentrations of 5% to 7%. However, environmental regulations passed about two decades ago have precluded the production of newHalon. Therefore, theworld’s remainingHalon supply consists ofwhat is stored inHalon banks. Although the Halon supply is finite, it is still available with a corresponding price tag. Many countries have forbidden the use of Halon 1301 and in those that have not, including the U.S., the installation of new Halon systems is generally discouraged.

Other gaseous systems

Other options include several Halon-alternative clean agent extinguishing gasses. These are generally safe for humans and have low- or no-ozone depletion potential.

Clean agents are broken down into two categories consisting of blends of inert gasses and halocarbons. The inert gasses, such as Inergen, reduce the oxygen content (but not enough to be lethal) and cool the flame temperature while others, such as FM-200, chemically and physically interrupt the chain reaction that generates flames. Clean agent systems are used in a totalflooding application, similar to CO2. The inert gas blends generally require a large supply of cylinders due to the high design concentration of 38% to 43%. This presents a significant disadvantage due to the system space requirements. Also, because of the large volume of gas discharged, enclosure over-pressurization can become a problem which may need to be addressed through pressure-relief provisions. Lastly, the availability of these specialty blends of inert gasses, especially in rural areas, can pose a supply problem.

Halocarbon agents, while requiring roughly the same agent quantity as Halon 1301, tend to be pricey.With either system, expect the installation cost to be 50% to 100% higher than CO2. The design standard for clean gasses is NFPA 2001 which provides requirements for design concentrations and other system functions. Clean agent systems are designed and installed for each specific protected space using relatively common materials.

As with CO2 systems, the integrity of the enclosure is critical. NFPA 2001 requires a retention period of 10 minutes for most protected hazards; however, this is not applicable to combustion turbines. See the discussion for CO2 systems for the necessary duration of the fire extinguishing discharge. While a “single shot” system design is usually adequate for typical hazards, it is unlikely that the single shot will meet the duration requirements of turbine enclosures. An extended discharge should thus be provided in the same fashion as the corresponding discussion for CO2.

An acceptance test is required for clean agent systems, but this test does not require discharging the fire extinguishing agent due to the high costs associated with refilling the cylinders. Rather, NFPA 2001 allows the use of a room integrity test (door fan test). The door fan test consists of pressurizing and then depressurizing the protected space using a special fan and housing which is temporarily sealed within a standard opening, usually a door - hence the name.The collected test data are entered into a computer program which, based on the volume of the protected space, predicts the agent retention time at the required concentration based on the supplied quantity of agent and the calculated equivalent leakage area. Due to many factors involved and the inability to measure actually achieved concentration, the door fan test tends to yield conservative results. Thus, it is better to overdesign a system than to have to modify it after the door fan test.

Regardless of the type of system, all systems need periodic inspections and testing (Chart). The frequencies and procedures can be found in themanufacturers’manual and in the correspondingNFPAstandards. In general, each of the aforementioned systems will need a weekly visual inspection to ensure that the system is in operational status and to check for any adverse conditions, a semiannual maintenance inspection, and an annual functional test (see Chart).

The weekly visual inspection can and should be carried out by plant personnel who are knowledgeable in the system’s operation. Since many jurisdictions require that periodic testing is performed by an appropriately licensed party, the semiannual and annual tests are best left to a specialized contractor. However, in the absence of such a requirement, properly-trained and equipped plant technicians can also perform this testing.


Dominique Dieken is a Senior Fire Protection Engineer with Starr Technical Risks Agency, Inc. (Starr Tech), a member of the Starr Companies group of companies. Starr Tech is an insurance agency serving the Power Generation, Petrochemical, Chemical, Energy and Oil & Gas industries. For more information visit or contact