Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 105


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 104
105 CHAPTER TWELVE FIXED FIRE SUPPRESSION AND ITS IMPACT ON DESIGN FIRE SIZE BACKGROUND Despite a potentially huge fire and the presence of more than 400 people in the tunnel, only three people died from the PIARC, NFPA, and several European countries are rethink- traffic accident and none from the subsequent fire. The Burn- ing fixed fire suppression application for tunnels. Before the ley Tunnel incident demonstrates that fixed fire fighting sys- Alpine tunnel fire disasters, Japan and Australia were the only tems are effective in protecting tunnel infrastructure and two countries to require and use sprinkler systems in road tun- delivering human safety (71). nels. It is noted that sprinklers were installed in several other tunnels around the world, including the United States. How- Presently, there are several ongoing discussions of the ben- ever, those installations were driven by specific requirements efits of sprinklers. However, there were also some past lessons and jurisdictions (e.g., Seattle 1952). learned, which are reviewed here. Based on the literature review, all Japanese class `AA' road For example, as mentioned earlier, the Ofenegg Tunnel tunnels are required to have sprinkler systems. (Class `AA' are tests (1965) included a 500 L (132 gal) sprinkler test, sprinkler tunnels with traffic density of more than 40,000 vehicles per droplets initially evaporated into a high-temperature steam day with a length of more than 1 km or 3,280 ft.) Starting in cloud, which caused more damage than the nonsprinklered 1963, a number of full-scale tunnel fire tests have been carried fires. The open fire was apparently soon extinguished, but out in Japan. It was concluded that sprinklers are able to reduce was accompanied by a strong odor of gasoline at the portal. fire size and temperature and prevent fire from spreading. In The fire then reignited after 17 min (status of sprinkler flow Japan, sprinklers have been used in two or three tunnel fire unstated) with pronounced, but nonexplosive, wave-front incidents per year. It shall be noted that the Japanese approach propagation. However, the ultimate minimum survival dis- is to activate sprinklers with a 3-min delay. This approach dif- tance for an upright subject was judged closer than for the fers from Australia, where sprinklers are activated immedi- nonsprinkled fires. ately (it takes 30 s for the deluge system to activate). As noted in the Ofenegg Tunnel test report, during the Lessons from the Burnley Tunnel fire in Australia, where a 1,000 L (264 gal) gasoline burn tests the sprinklers were major disaster was successfully averted by a brand new suc- immediately activated after ignition. The sprinklers reduced cessfully working safety system, are currently being studied the maximum arch temperature significantly. However, the (69). In March 23, 2007, the fire in the Melbourne City Link steam apparently pushed burning gases and gasoline vapors Burnley Tunnel started with a road traffic accident involving into adjacent tunnel sections, where they continued to burn. four cars and three HGVs. The pile-up of trucks and cars inside The fire was apparently extinguished after 10 min, but the the 3.4-km (2.1-mi) long Burnley Tunnel that killed three peo- tunnel filled with gasoline vapors, which exploded in the ple burst into a wall of fire that reached temperatures of more nineteenth minute, causing extensive damage to the test than 1,000C (1,832F). However, further casualties were setups and injuring three technicians. A lesson learned is avoided. Although, according to the Sydney Morning Herald that once the sprinkler system is activated, it is not to be (70) some witnesses reported that they had not seen any sprin- turned off until the fire source is completely extinguished kler or safety system in operation, Acting Metropolitan Dire and determined safe. Brigade Chief Officer Keith Adamson said that both sprinkler and smoke extraction systems made it much easier to find the A delay in activation produces huge volumes of high tem- source of the fire. Hundreds of motorists were immediately perature steam, which can be as dangerous as the combustion advised to leave their cars with their keys in the ignition and products. If all ignition sources cannot be extinguished and the evacuate the tunnel. Most took the emergency exits, which site uniformly cooled below a safe temperature, the fire will lead to separate pedestrian tunnels, whereas some took the reignite, perhaps explosively, when the sprinklers are shut off. riskiest route by walking back to the tunnel entrance. As a con- Meanwhile, unburned vapors are propelled around the tunnel sequence, the Burnley Tunnel, which opened in late December and ventilation ducts, which can cause another significant 2000, is now widely regarded as an example of a modern hazard to those safely away from the fire, even after the fire is safety model. extinguished.

OCR for page 104
106 FIGURE 34 Schematic effect of suppression on heat release rate (71). Figure 34 schematically shows the effect of suppression with one specification being the water application density, on HRR. With timely activation of a suppression system, the measured like rainfall in millimeters//minute. Droplets from HRR is reduced. With delayed activation, the fire becomes water spray systems are generally larger than 1 mm (0.04 in.) overwhelming and the suppression system is not effective. in diameter. It is vital to have a clear understanding of the capabilities of Meanwhile, water mist systems use higher pressures, in the detection system and the lead-in times for activation of the some cases more than 100 bar (1450.4 psi), and discharge fire life safety systems. It is essential that the detection system much finer droplets, 99% of which have a diameter less than be capable of detecting a small fire (in the order of 15 MW). 1 mm (0.04 in.). Nozzles with very small orifices are used to If this is not achieved and the fire is not detected until it enters create the mist. The smaller droplets are drawn into the fire by its rapid growth phase, the resultant fire will, in all likelihood, its own ventilation and easily evaporate owing to the large- be well beyond the capabilities of a fixed fire suppression sys- surface area-to-volume ratio. The mist systems may require tem once it is activated (72). less water per zone; storage tanks, pumps, and pipes can be smaller, saving on costs. However, to protect the small nozzle Although a few automatic sprinkler systems have been orifices higher quality materials, such as stainless steel, are installed in tunnels, most systems are deluge systems. A del- required, which add to the costs. uge system has a network of open nozzles at the roof of the tunnel, divided into zones, typically of 30 m (100 ft) based on Research projects are investigating to what extent an active the length of a HGV. When there is a fire, a valve is opened in fire protection system can limit the maximum HRR and the zone above the fire and in the zones on either side. Water whether an active fire protection system combined with venti- is sprayed from all the nozzles in the activated zones. lation offers equal or better life safety. The projects are also investigating how to specify design or performance test crite- Deluge systems have been selected over automatic sprin- ria for tunnel active fire protection systems. Today, more than kler systems as a result of two concerns. First, the ventilation 100 tunnels are equipped with an active fire protection sys- system in a tunnel could spread heat initially to sprinklers that tem. Fixed fire suppression systems have been successfully are not above the fire. Second, a tunnel fire could rapidly used for more than 40 years in Japan's congested urban road develop a considerable amount of heat over a large area so tunnels and, more recently, in all of Australia's congested that too many sprinklers would open, overwhelming the water urban tunnels. supply. In contrast, a deluge system takes a fixed amount of water and, with suitable detection, it is possible to open only Road tunnel deluge systems require substantial amounts the zones above or next to the fire. of water, which can have a significant impact on the storage, delivery, and drainage systems (although water mist systems Deluge water spray nozzles take water at a typical pressure require less water per zone). One study came to the conclu- of 1.5 to 5 bar (21.8 to 72.5 psi) and discharge a pattern of sion that, although some minimum water application rates water droplets over the area below. Water spray systems are would achieve a certain objective, a marginally higher rate designed to achieve an even discharge of water over an area, would not necessarily improve the situation (79).

OCR for page 104
107 Japan and Australia each have their own specified water Does water requirement depend on ventilation and lon- application rates to use for road tunnel fixed fire suppres- gitudinal air velocity? Ventilation may have a dual effect. sion system design, which are 6 mm/min (0.15 gpm/ft2) and Ventilation may blow away or exhaust water particles 10 mm/min (0.25 gpm/ft2), respectively. In full-scale tunnel from the fire site. Ventilation may also increase the speed sprinkler tests conducted in Europe (2nd Benelux), a water of evaporation. The blow away effect may result in the application rate of 14 mm/min (0.35 gpm/ft2) has been tested. need for activation of additional fire zones. The intense These values have been added to Figure 35 to demonstrate the evaporation needs additional studies. significant variation in prescribed water application rates for which little research has been done to compare their effec- The following conclusions were drawn in the UPTUN proj- tiveness when applied under similar conditions. ect on the basis of the fire tests with the fire mitigation systems: Fire point theory shows that there are optimum rates of Validation of the performance of fire safety equip- water application that can control a fire and are signifi- ment, such as water spraying systems, requires full- cantly less than the rates generally prescribed. Furthermore, scale fire testing and cannot be trusted from model this theory suggests that there are minimum water applica- simulations. tion rates that can reduce the heat flux below certain criti- The efficiency of the water mist systems was satis- cal limits required to sustain combustion and, once these factory. limits are reached, more water offers little or no benefit. However, the efficiency was strongly dependent on the The results of the comparative analyses suggest that water size of the fire (or heat generation rate), nozzle type, loca- application rates as low as 2 mm/min (0.05 gpm/ft2) can tion, and the water discharge rates. offer some benefits by cooling exposed surfaces and assist- For the smallest fires (less than or equal to 5 MW or ing in limiting the spread of fire from the initiating point 17 MBtu/hr) the mitigation effect was minor. (see Figure 35). The best results were achieved for the largest fires (i.e., a HRR at or above 20 MW or 68 MBtu/hr). The maxi- Although the conclusions are interesting, they need to be mum reduction of the HRR was 80%. further evaluated by answering these questions: A rapid reduction of the temperatures downstream of the fire was noticed after activation of the suppression Does the water requirement depend on the design HRR? system. The efficiency of both water mist systems was Typically, the higher the FHRR, the more the water satisfactory with respect to heat stresses as well as the evaporates. toxicity of the fire effluents on human beings. Does the water requirement depend on fire size at the The visibility was not improved downstream of the fire time of fixed fire suppression system activation? It during the first minutes after activation of the suppression appears that the earlier the system will be activated, systems. However, the visibility was generally increased the lower the FHRR will be and the less water may be as the fire size and the HRR were reduced during fire required. Some previous works have already demon- suppression. strated that late fixed fire suppression system activa- The problem of backlayering (i.e., smoke spread upstream tion resulted in an inability to take the fire under control, of the fire) and the visibility upstream were also signifi- which caused the FHRR to continue to increase. cantly improved after activation of the water mist systems. FIGURE 35 NFPA 13, NFPA 15, and other International Water Application Rates ( 79).

OCR for page 104
108 FIGURE 36 UPTUN Fire Heat Release--Temperature curve for classifications of ventilation and fixed fire suppression systems (73). High pressure water mist systems are using less water Accidental activation of the system with the cause un- and suppress the fire to a higher degree in the gas phase known, which happened in Boston, is not acceptable (see of the flames. However, for the low pressure systems, Figure 37). A malfunctioning activation of the sprinkler sys- the fire extinguishing effect is mainly cooling of the fuel tem drenched the tunnel under City Square in Charlestown, surfaces. converting the 1,100-ft (335.3-m)-long tunnel into a tempo- rary car wash. The activation was inadvertent and the source Figure 36 shows Type A fires where mitigation action of the activation unconfirmed. The malfunction activation is provided, whereas Type B often represents fires out of forced State Police to close the three main ramps that lead traf- control and may provide significant heat exposure to the fic from Storrow Drive, Interstate 93 north, and Rutherford structure. Avenue into the tunnel. Type A fires are assumed to be significantly less severe It is recognized that active fire protection systems can limit than Type B fires, which may result in unbearable conditions the size and growth of a fire and prevent the fire from spread- for humans and significant thermal exposure to construc- ing. It is also recognized that active fire protection systems will tions. Small fires, Type A, are often limited to the first object limit damage to the tunnel in the event of a fire, so that even a burning and can be ranked by the HRR measured in megawatts fire involving several HGVs will not close the tunnel for long. or 1,000 Btu/hr, although more severe fires, after signifi- It could also protect tunnel lining, possibly reducing the cant flame spread, can also be measured in terms of time amount of passive structural fire protection and providing sig- temperature curves. nificant construction savings. At the control level there are a range of opportunities to For the UPTUN fire mitigation test program, the main fully integrate such systems with the ventilation, operate them focus has been on Type A fires to protect human beings, to avoid flame spread, and to provide conditions for unhindered escape and rescue. Type A fires can be considered as fires with a HRR of up to 30 MW (102 MBtu/hr), whereas higher HRR can be considered as Type B fires. To operate effectively, the fixed fire suppression system has to be properly maintained. The operator must be able to activate it correctly, and it must survive the events that have resulted in the incident requiring its activation. Automatic activation of the sprinklers by active detectors may need to be delayed because even a light spray could star- tle unaware drivers and make the roadway slippery. Water FIGURE 37 Accidental activation of the sprinkler fire squirting from the ceiling of a subaqueous tunnel would sug- suppression system in Boston CANA (Central Artery North Area gest tunnel failure and can induce panic in motorists. Tunnel) for 45 min on May 15, 2005, at 2 p.m. (54).

OCR for page 104
109 separately, fully automate them, automate them with manual imized, and, thereby, toxic gas and smoke generation vol- override, and manually operate them with auto override. How- umes contained. ever, each of these options must be carefully evaluated. Dif- ferent tunnels will require different approaches. The undesirable consequences of its activation, such as smoke de-stratification, increased humidity, and decreased One important lesson learned from the Ofenegg and other visibility, are hopefully outweighed by their other positive out- tunnel tests is that it is dangerous to turn the fire suppression comes of fire growth rate control, containment of fire spread, system off while surfaces are still hot and fuel vapors are pres- and reduced temperatures. ent, because they can ignite and cause an explosion. A well- thought out operation of the fixed fire suppression system is The Runehamar tests brought up the question: Is it possi- important because fire sizes can be very large. ble to manage a 200 MW (682 MBtu/hr) fire? A fire sup- pression industry offers to control the fire size, reducing the For an active fire protection system to be effective it is maximum HRR by applying a fixed fire suppression system. essential that fires are quickly and accurately detected. Sprin- Once a fire is early detected by a reliable fire-detection sys- kler systems are to be designed to prevent a fire from reaching tem, the fire protection system could be activated within its peak; however, the droplets will be affected by ventilation. several minutes, taking the fire in the order of about 10 MW Longitudinal airflow must be selected to ensure an appropriate (34 MBtu/hr) and under control, or suppress a small fire. droplet spread and mass flow performance for given water However, the question is what this will do for the tunnel pressures. No doubt ventilation system performance is also safety (Table 33). affected by sprinkler operation. However, the main idea is to get a well-designed system with a reliable quick fire-detection With the longitudinal ventilation system, it appears reason- system to start these systems before the fire gets too large. able to activate both systems simultaneously. As a result, the wet fixed fire suppression system will initially start before the For an effective deluge operation, activation must be longitudinal ventilation reaches full speed. Note that it takes rapid and accurate. If discharged in this way, fire growth 30 s to discharge water if it is a wet system, whereas it takes rates are likely controlled, the risk of rapid fire spread min- 3 min to achieve a full operational ventilation mode. This TABLE 33 IMPACT OF A FIXED FIRE SUPPRESSION SYSTEM (FFSS) ON TUNNEL FIRE SAFETY Advantages of FFSS Challenges of FFSS General A sprinkler is designed to react at an early Possible loss of visibility (reduced visibility) especially at stage of the fire. Takes fire under control, an early stage when people evacuate. When the sprinkler not allowing it to further grow, or grow system is activated on an already large fire, a large amount slowly, or extinguishes a small fire before of water will be evaporated and, thus, the visibility will be the fire department arrives. further diminished. Protection of tunnel users and structure. Incomplete combustion creates smoke, gases, and steam. Duration of the fire can be limited and the Studies needed on critical time to activate the FFSS to structure of the tunnel will be subjected to protect the tunnel structures. less harsh conditions. Reaching the fire: help rescue team and Creates slippery environment when water applied. May firefighters to reach the fire source. create panic when it malfunctions with an accidental water release. If a system (a normal wet sprinkler) is activated by a defect such as breaking of the glass, water will be sprinklered into a tunnel with a possibility of causing an accident. Transverse ventilation based on smoke extraction (including single-point extraction) Reduced fire size, see also general Destroys stratification of hot air, which makes ceiling extraction inefficient and evacuation difficult. Reduced fire duration Increases mass of air/water mixture to move, results in increased vent rate for sidewall extraction system. Longitudinal ventilation--unidirectional tunnel with manageable traffic Reduced fire size may result in reduced Increases mass of air/water mixture to move--increases ventilation rate vent rate Cools environment and protects fan units Overcomes water curtains created by the FFSS--increases from high temperature vent rate See general Blows the FFS substances away from the fire--increases number of FFS zones for activation. Longitudinal ventilation--unidirectional tunnel with unmanageable traffic or bidirectional tunnel Protects tunnel structure Destroys stratification making evacuation difficult (maybe impossible) to both sides of the fire once the FFSS is activated. Traffic control for low traffic tunnels is imperative.

OCR for page 104
110 allows the sprinkler system to discharge water in a low air Annex E (the explanatory material) of NFPA 502 (2008 velocity environment, thus protecting people and structures by edition) notes that the major concerns expressed in the past by taking control of a fire at an early stage of its growth. Once the tunnel designers, engineers, and authorities worldwide regard- ventilation reaches full speed, the sprinkler zones may need to ing the use and effectiveness of water-based fixed firefighting be revisited and either switched or additional activation zones systems in road tunnels, along with the current assessment of will be required to account for ventilation. those issues have been revisited as follows: With the transverse ventilation system using ceiling 1. Fires in road tunnels usually occur inside vehicles or exhaust, the sequence of activations may differ. The primary inside passenger or engine compartments designed to purpose of the fire life safety system is to save lives and be waterproof from above; therefore, water-based fixed allow for safe evacuation. Destruction of the smoke layer, firefighting systems would not have an extinguishing worsening of visibility, and potential generation of hot effect. steam, need to be considered. The Japanese approach for It is now recognized that the purpose of a water- based fixed firefighting system is not to extinguish the the transverse system may be reasonable, which allows for fire but to prevent fire spread to other vehicles so that the a minimum of a 3-min delay before the sprinkler activa- fire does not grow to a size that cannot be attacked by tion, so that people can leave the sprinkler zones. However, the fire service. sprinkler activation delay may be dangerous for the tunnel 2. If any delay occurs between ignition and the water- structure and can lead to fire spread and growth. This con- based fixed firefighting system activation, a thin water firms the need for an integrated approach to all fire life spray on a very hot fire could produce large quantities safety systems (2). of superheated steam without materially suppressing the fire. There are a number of questions that need further study: Fire tests have shown this not to be a valid concern. A properly designed water-based fixed firefighting sys- 1. NFPA 502 and other standards allow for a maximum air tem suppresses the fire and cools the tunnel environ- velocity in a tunnel of 12 m/s (2,200 fpm). Ventilation ment. Because a HGV fire needs only 10 min to exceed systems are designed for significantly smaller critical air 100 MW (341 MBtu/hr) and 1200C (2192F), which velocities, but in combination with wind, other natural are fatal conditions, it is important to operate the fixed factors, and the traffic pattern, the resultant air velocities firefighting system as soon as possible. may be that high. What will such velocities do to a fixed 3. Tunnels are long and narrow, often sloped laterally and fire suppression system's performance? longitudinally, vigorously ventilated, and never sub- 2. Once a fixed fire suppression system is activated, it will divided: therefore, heat normally will not be localized create a water curtain in the tunnel for longitudinal air over a fire. velocity. The air velocity will be reduced and could be Advances in fire-detection technology have made it less than critical for the sprinkler controlled fire HRR. possible to pinpoint the location of a fire in a tunnel with Will smoke be under control or does the ventilation sufficient accuracy to operate a zoned water-based fixed system performance need to be increased? firefighting system. 3. If a sprinkler is activated early enough, can ventilation 4. Because of the stratification of the hot gas plume along be reduced or eliminated and what will be the impact on the tunnel ceiling, a number of the activated fixed fire smoke production? suppression systems would not, in all probability, be 4. A fixed fire suppression system will increase humidity located over the fire. A large number of the activated water-based fixed firefighting systems would be located in the tunnel. How will this humidity affect the ventila- away from the fire scene, producing a cooling effect that tion and fan's performance? would tend to draw the stratified layer of smoke down 5. Other questions are related to fire detection and the toward the roadway level, thus impeding rescue and operator's control of the situation, low visibility, haz- firefighting efforts. ardous slippery conditions, system activation malfunc- Independent laboratories have commented that they tion concerns, and optimum systems activation time. do not observe smoke stratification. Any activated water-based fixed firefighting system not over the fire Critical factors such as droplet size distribution and trajec- would cool the tunnel to help rescue services to inter- tory modeling of droplets through a range of longitudinal vene. Zoned systems are released by a detection sys- velocities are essential for CFD modeling. tem that is accurate even with forced ventilation. 5. Water spraying from the ceiling of a subaqueous NFPA 502 recognizes the benefits of the fixed fire sup- tunnel could suggest tunnel failure and induce panic in pression system for road tunnels, but is concerned with the motorists. available fire-detection technology, with the further visibility This theoretical concern was not borne out in prac- reduction, and with the impact of the fixed fire suppression tice. In the event of a fire, motorists are likely to rec- system on the effectiveness of tunnel ventilation. ognize water spraying from nozzles as a fire safety