Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 29


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 28
29 CHAPTER SIX FIRE TESTS--LITERATURE REVIEW Fire tests are of vital importance in the understanding of the Measurement errors (e.g., low-velocity measurement physics of tunnel fires, understanding the impacts of fires, by inappropriate instrumentation). and for verifying calculations, assumptions, computer models, Raw data processing algorithms and subjective judgments and tunnel design. They are also important for tunnel opera- (e.g., visibility judgments based on video recording). tors and emergency responders in their efforts to coordinate and verify in practice the emergency response plans. Each full-scale program had its own objectives and goals, which drove methodology and resultant conclusions, mak- The fire tests that have been performed can be classified as: ing it difficult to generalize findings from the historical test data results. Tests before the design to develop design methodology. Tests during the design to verify assumptions and com- There is a great need for more large-scale testing to be able puter models. to better understand the fire and smoke dynamics. However, Tests during commissioning to verify the design and the tests must be carefully designed and equipped with appro- equipment operation. priate instrumentation. Tests for training purposes. Other tests as needed. Research programs using full-scale facilities generally deal with numerous but very specific aspects of safety character- Important work has been conducted at full-scale (large- ized by high human and economic stakes. They require signif- scale) tunnels, including: icant financial support. The main large-scale test programs that have been performed follow. EUREKA tests Memorial Tunnel Fire Ventilation Test Program Runehamar Tunnel fire tests Ofenegg Tunnel Tests Full-scale tests in Norway To gain at least a general impression of the temperature Tests in Japan. conditions and the amount of smoke to be expected from a gasoline fire, evaluations of the performance of the fixed Experimental tests and especially their replications are fire suppression system tests were performed in the Swiss expensive and there is a lack of willingness to carry them Ofenegg Tunnel and in the Zwenberg Tunnel in Austria. out. However, it is very important to perform tunnel fire Both test facilities were abandoned railroad tunnels. tests and there is a need for multi-agency and international collaboration. Two types of ventilation systems, longitudinal and semi- transverse supply, were evaluated. The tunnels had no exhaust FULL-SCALE TESTS provisions. Sprinklers were mounted over the fuel basin and their effectiveness was evaluated. Eight tests were scheduled Full-scale tests are often expensive to carry out. They require with different ventilation systems: access to a tunnel or to a full-scale mock-up with some basic installations. Large-scale and full-scale fire tests with HRRs of Natural 100 MW (341 MBtu/hr) or more require normal modifications Longitudinal and protection of the lining and installations. Measuring the Semi-transversely fire size (in terms of the HRR) needs advanced instrumentation With sprinklers for 500 L (132 gal.) fuel fire and data analysis. Some lessons learned during the previous With sprinklers for 1,000 L (264 gal.) fuel fire. large-scale tests included: Test results raised doubt of the effectiveness of sprinklers Lack of control of the conditions of the experiments in containing a fire or in limiting the range and severity of (e.g., humidity). damage. A delay in activation may produce a significant vol- Lack of careful design of the experiment (location of ume of high temperature steam as dangerous as the combus- thermocouples and other instrumentation). tion products. If all ignition sources cannot be extinguished

OCR for page 28
30 and the site uniformly cooled below a safe temperature the from vehicles and extending the visible distance by tak- fire may reignite, perhaps explosively, when the sprinklers ing into consideration estimated traffic volume, tunnel are shut off. Meanwhile, unburned vapors spread throughout length, natural ventilation, ventilation by movement of the tunnel and ventilation ducts are at great hazard far from vehicles, and so forth. Depending on these design con- the fire, even if the fire is extinguished. (Additional test ditions, there may be a small number of cases in which descriptions can be found in web-only Appendix E.) smoke can be reasonably extracted by existing ventila- tion systems. Stratification of smoke was partially destroyed by longi- Zwenberg Tunnel Tests tudinal ventilation at 1 m/s (197 fpm) and totally des- troyed by longitudinal ventilation at 2 m/s (394 fpm). This program was initiated in 1975 in connection with two For determining the capacity of ventilating fans in the major motorways projects in Austria (21). Longitudinal and future, the fire smoke exhaust capacity of the fans shall semi-transverse ventilation systems were tested. The tests be designed to meet the scale of a real vehicle fire. included a total of 30 pool fires. (See web-only Appendix E The sprinklers had an adverse effect on the tunnel envi- for additional information on the tests.) ronment by causing a reduction in smoke density near the ceiling and an increase in smoke density in the lower Because the test results so strongly supported the benefits part of the tunnel. of a fully transverse system running in a full extraction mode None of the car, bus, or pool fires was totally extin- during a fire, the investigators made the following recom- guished by the sprinklers; however, the heat generation mendations for the design and operation of a tunnel ventila- speed was reduced in each case. tion system: The very rapid development of a fire requires a suitable Repparfjord Tunnel Tests pattern of ventilation for creating the best possible con- ditions for rescue. Tests were undertaken at the Repparfjord Tunnel near Ham- To fulfill this requirement it is necessary merfest, Norway from 1990 to 1992 (21). That test report that the fire is quickly detected and the alarm trans- concluded that: mitted to a tunnel control center where the operating pattern can be selected, and The influence of damage both to the vehicles and tunnel that the appropriate technical and organizational lining, especially in the crow area, depends on the type measures be prepared, securing a fast and correct of vehicle. The roofs of those vehicles constructed of selection of the operating pattern of the ventilation steel resisted the heat, whereas the roofs of the vehicles system in the case of a fire. made of aluminum were completely destroyed during The tunnel must be equipped with a quickly responding an early stage of the fire. fire warning system. Signals are to be transmitted with The temperatures during most of the vehicle fires reached minimal possible delay to the control center. maximum values of 800C to 900C (1472F to 1652F). The primary goal must be to prevent the spread of hot The temperatures during the HGV test reached 1300C fumes and smoke in the traffic space. (2372F). Temperatures decreased substantially within a This recommendation must be implemented without short distance from each fire location and were greater any restriction in all tunnels with two-way traffic. downwind than upwind. The HGV burned at an HRR of more than 100 MW (341 MBtu/hr). Public Works Research Institute Experiments Fast fire development registered in the first 10 to 15 min. Growth rates of vehicle fires vary from medium to PWRI experiments (21) (Japan 1980) are described in web- ultrafast. only Appendix E. The PWRI test report concluded that: Longitudinal ventilation destroyed stratification down- wind of the HGV fire. Smoke can be kept within the minimum space and be extracted quickly if the kinetic energy of the smoke flow produced by the thermal energy of fire is less than Benelux Tunnel Tests the energy of ventilating air blowing along the tunnel toward the smoke ventilation dampers when the fans In the Benelux Tunnel, 14 fire tests were used to determine the are run in reverse direction. This is achieved by the rela- benefits of fitting large drop sprinklers. These sprinklers were tionship between the scale of the fire and the capacity selected so that the large droplets would penetrate the power- of the fans (i.e., if the fire is too big, the fans will not ful fire plumes and not be swept away by the tunnel ventilation. extract all of the smoke). In the tests, with ventilation at up to 5 m/s (984 fpm), sprin- Ventilation fans are generally designed for the pur- klers reduced temperatures to safe levels upstream and down- pose of reducing the concentration of exhaust gases stream of the fire. They also reduced the probability of fire

OCR for page 28
31 spread between vehicles. Results of these tests are discussed in Memorial Tunnel for fire tests ranging in size from 10 MW the chapter thirteen. to 100 MW. Memorial Tunnel Tests Single-Zone Transverse Ventilation Systems The Memorial Tunnel tests (United States, 19931995) (21, Single-zone, balanced, full-transverse ventilation systems 25, 26) were financed by the FHWA and the Commonwealth that were operated at 0.155 m3/s/lane-meter (100 ft3/min/ of Massachusetts for the Boston Central Artery Tunnel proj- lane-foot) were ineffective in the management of smoke and ect. The experiments were performed in an abandoned 854-m heated gases for fires of 20 MW (68 MBtu/hr) and larger. (2,800-ft)-long road tunnel located in West Virginia. Approx- Single-zone, unbalanced, full-transverse ventilation systems imately 90 tests were done with diesel oil pool fires. The generated some longitudinal airflow in the roadway. The result obtained HRRs varied from 10 MW (34 MBtu/hr) for a 4.5 m2 of this longitudinal airflow was to offset some of the effects of (48.4 ft2) area to 100 MW (341 MBtu/hr) for a 44.4 m2 buoyancy for a 20 MW fire (68 MBtu/hr). The effectiveness of (478 ft2) area. There were 1,450 devices installed in the tun- unbalanced, full-transverse ventilation systems is sensitive to nel, providing about 4 millions points of data per experiment. the fire location, because there is no control over the airflow (See web-only Appendix E for test facility description.) direction. The Memorial Tunnel program performed tests with fire sizes of 10, 20, 50, and 100 MW (34, 68, 172, and Multiple-Zone Transverse Ventilation Systems 341 MBtu/hr). The tests were done with various ventilation systems including: The two-zone (multi-zone) transverse ventilation system that was tested in the Memorial Tunnel Fire Ventilation Test Full-transverse Ventilation--Air is uniformly supplied Program provided control over the direction and magnitude and exhausted throughout the entire length of a tunnel of the longitudinal airflow. Airflow rates of 0.155 m3/s/lane- or tunnel section. meter (100 ft3/min/lane-foot) contained high temperatures Partial Transverse Ventilation--Either supply air or from a 20 MW (68 MBtu/hr) fire within 30 m (100 ft) of the exhaust air, but not both, is uniformly delivered or fire in the lower elevations of the roadway and smoke within extracted throughout the entire length of a tunnel. 60 m (200 ft). Partial Transverse with Single-Point Extraction--A series of large, normally closed exhaust ports distrib- The spread of hot gases and smoke was significantly uted over the length of the tunnel to extract smoke at a greater with a longer fan response time. Hot smoke layers point closest to the fire. were observed to spread very quickly, from 490 m to 580 m Partial Transverse with Oversized Exhaust Ports-- (1,600 ft to 1,900 ft) during the initial 2 min of a fire. Natural Normally closed exhaust ports that automatically open ventilation resulted in the extensive spread of smoke and in a fire emergency. heated gases upgrade of the fire, but relatively clear condi- Natural ventilation. tions existed downgrade of the fire. The spread of smoke and Longitudinal ventilation with jet fans. heated gases during a 50 MW (171 MBtu/hr) fire was con- siderably greater than for a 20 MW (68 MBtu/hr) fire. The depth of the smoke layer increased with fire size. Longitudinal Tunnel Ventilation Systems A longitudinal ventilation system employing jet fans is highly A significant difference was observed between smoke effective in managing the direction of the spread of smoke for spread with the ceiling removed (arched tunnel roof) and fire sizes of up to 100 MW in a 3.2% grade tunnel. The throt- with the ceiling in place. The smoke and hot gas layer migrat- tling effect of the fire needs to be taken into account in the ing along the arched tunnel roof did not descend into the design of a jet fan longitudinal ventilation system. roadways as quickly as in the tests that were conducted with the ceiling in place. Therefore, the time for the smoke layer Jet fans that were located 51.8 m (170 ft) downstream of to descend to a point where it poses an immediate life safety the fire were subjected to the following temperatures for the threat is dependent on the fire size and tunnel geometry, tested fire sizes: specifically tunnel height. In the Memorial Tunnel, smoke traveled between 290 m and 365 m (950 ft and 1,200 ft) along 204C (400F)--20 MW fire the arched tunnel roof before cooling and descending toward 332C (630F)--50 MW fire the roadway. The restriction to visibility caused by the move- 677C (1250F)--100 MW fire. ment of smoke occurs more quickly than does a temperature that is high enough to be debilitating. In all tests, exposure Air velocities of 2.54 m/s to 2.95 m/s (500 fpm to 580 fpm) to high levels of CO was never more critical than smoke or were sufficient to preclude the backlayering of smoke in the temperature.

OCR for page 28
32 The effectiveness of the foam suppression system Aqueous 5 m (16 ft) downstream. A mock-up of a partially filled lique- Film-Forming Foam (AFFF) that was tested was not dimin- fied petroleum gas (LPG) tank was tested for exposure and ished by high-velocity longitudinal airflow [4 m/s (787 fpm)]. boiling liquid expanding vapor explosion (BLEVE) risk. The time taken for the suppression system to extinguish the fire, with the nozzles located at the ceiling, ranged from The water mist system prevented a risk of a BLEVE for the 5 s to 75 s. diesel pool fire and for the solid fire if the water mist system was activated before the HRR exceeded 50 MW (171 MBtu/hr). The maximum temperatures experienced at the inlet to However, if the water mist system activation was delayed until the central fans that were located closest to the fire [approx- the HRR reached 200 MW (682 MBtu/hr) there was a serious imately 213 m (700 ft) from the fire] were as follows: risk of a BLEVE. Measurements were taken of the tempera- ture, CO concentration, and visibility downstream of the fires. 1. 107C (225F)--20 MW (68 MBtu/hr) fire It was concluded that tenability was regained within a few 2. 124C (255F)--50 MW (171 MBtu/hr) fire minutes of activation of the water mist system. 3. 163C (325F)--100 MW (341 MBtu/hr) fire. There have been numerous papers discussing and analyz- In a road tunnel, smoke management necessitates either ing the test results and what allowed the fire to grow to that direct extraction at the fire location or the generation of a lon- size. Some questions included: gitudinal velocity in the tunnel that is capable of transporting The type of truck burning (open trucks are not used in the smoke and heated gases in the desired direction to a point the United States). of extraction or discharge from the tunnel. Without a smoke The tunnel size, which was smaller (narrower) than a management system, the direction and rate of movement of typical road tunnel. the smoke and heated gases are determined by fire size, tunnel Protection of tunnel walls with heat protection material, grade (if any), pre-fire conditions, and external meteorological which does not allow for heat dissipation through the conditions. walls, but rather reflects heat from the walls back to the tunnel environment with superimposed heat waves. The program report showed that balanced full-transverse ventilation is ineffective in controlling smoke and tempera- Results of the tests have been published in the Annex tures when fires are above 20 MW (68 MBtu/hr). Being able materials of NFPA 502, in ASHRAE, and in other docu- to effectively control the temperature when fires are below ments impacting mechanical and structural tunnel design in 20 MW (68 MBtu/hr) depends on their locations. However, many countries of the world. if the transverse ventilation system is modified to be a two zone system, it can have the capability to control temperature and smoke for a 20 MW (68 MBtu/hr) fire positioned at dif- UPTUN Project Tunnel Tests ferent locations along the length of the tunnel. The HRRs for single passenger cars (small and large) vary from 1.5 to 9 MW (5.1 to 31 MBtu/hr); however, the majority of the Runehamar Tunnel Tests tests show HRR values of less than 5 MW (17 MBtu/hr). When two cars are involved, the peak HRR varies between 3.5 The Runehamar Tunnel fire tests were initiated, planned, and and 10 MW (12 and 34 MBtu/hr). There is a substantial variety performed by the Swedish National Testing and Research in the time to reach peak HRR; that is, between 10 and Institute from 2001 to 2003 as a part of the Swedish National 55 min. It has been shown that the peak HRR increases linearly Research program and in collaboration with the European with the total calorific value of the passenger cars involved in UPTUN project led by TNO (The Netherlands) (27). (See the fire. An analysis of all data available shows that the average web-only Appendix E for tests description.) increase is about 0.70.9 MW/GJ (2.43.1 MBtu/hr/GJ). Free-burn fire tests in the Runehamar Tunnel in Norway There have only been a few bus fire tests performed. The alarmed the industry with a 200 MW (682 MBtu/hr) HGV two tests shown in the Table 6 indicate that the peak HRR is fire size and its fast growth, because in the past no one on the order of 30 MW (102 MBtu/hr) and the time to reach believed in such high values. This led to a change in design peak HRR is less than ten minutes. HRRs for tunnel fires. In 2008, a third series of tests were run in the Runehamar Tunnel to evaluate the performance of The highest peak HRRs were obtained for the HGV trailers water mist. With ventilation of up to 5 m/s (984 fpm), the (single), which were found to be in the range of 13 to 202 MW water mist system was applied to a 100 m2 (1,076 ft2) diesel (44 to 689 MBtu/hr), depending on the fire load. The time to pool fire and a 200 MW (682 MBtu/hr) HGV fire. Within a reach peak HRR was in the range of 10 to 20 min. The fire minute, the diesel fire was extinguished. After a minute for duration was less than one hour for all the HGV trailer tests the HGV fire, the temperature had dropped below 50C presented in Table 6. The fire growth rate after reaching 5 MW (122F), 20 m (66 ft) upstream, and below 280C (536F), (17 MBtu/hr) was nearly linear during all the tests carried out

OCR for page 28
TABLE 6 LARGE-SCALE EXPERIMENTAL DATA RESULTS FROM UPTUN TESTS Peak Peak Type of vehicle, model year, test Calorific HRR Time to Temperatures in Reference nr. Value (Qmax) Peak HRR Tunnel Ceiling [see Ingason u = longitudinal ventilation m/s (GJ) MW (min) (C) (28)] Passenger cars Ford Taurus 1.6, late 70s, Test 1 4 1.5 12 N/A Mangs and Datsun 160 J Sedan, Late 70s, 4 1.8 10 N/A Keski- Test 2 Rahkonen Datsun 180 B Sedan, Late 70s, 4 2 14 N/A Test 3 Fiat 127, Late 70s, 0.1 m/s N/A 3.6 12 N/A Ingason et al. Renault Espace J11-II, 1988, Test 7 6 8 480 Steinert 20, u = 0.5 m/s Citron BX, 1986 5 4.3 15 N/A Ship and Austin Maestro, 1982 4 8.5 16 N/A Spearpoint Opel Kadett, 1990, Test 6, N/A 4.9 11 210 u = 1.5 m/s Lemaire et al. Opel Kadett, 1990, Test 7, N/A 4.8 38 110 u = 6 m/s Renault 5, 80s, Test 3 2.1 3.5 10 N/A Renault 18, 80s, Test 4 3.1 2.1 29 N/A Joyeux Small Car, 1995, Test 8 4.1 4.1 26 N/A Large Car, 1995, Test 7 6.7 8.3 25 N/A Trabant, Test 1 3 .1 3 .7 11 N/ A Austin, Test 2 3.2 1.7 27 N/ A Steinert Citroen, Test 3 8 4.6 17 N/ A Renault Laguna, 1999 13.7 8.9 10 N/A Marlair and Lemaire Two passenger cars Citroen BX + Peugeot 305, 8.5 1.7 N/A N/A 80s, Test 6 Small Car + Large Car, Test 9 7.9 7.5 13 N/A Joyeux Large Car + Small Car, Test 10 8.4 8.3 N/A N/A BMW + Renault 5, 80s, Test 5 N/A 10 N/A N/A Polo + Trabant, Test 6 5.4 5.6 29 N/A Peugeot + Trabant, Test 5 5.6 6.2 40 N/A Steinert Citroen + Trabant, Test 7 7.7 7.1 20 N/A Jetta + Ascona, Test 8 10 8.4 55 N/A Three passenger cars Gold + Trabant + Fiesta, Test 4 N/A 8.9 33 N/A Buses A 2535-year-old, 12-m long 41 29 8 800 Volvo School Bus with 40 Ingason Seats, EUREKA 499, u = 0.3 m/s A Bus Test in the Shimizu N/A 30 7 303 Kunikane Tunnel, u = 34 m/s et al. HGV A Trailer Load with Total 10.9 240 202 18 1365 Ton Wood (82%) and Plastic Ingason and Pallets (18%). Runehamar Test Lonnermark Series, Test 1, u = 3 m/s A Trailer Load with Total 6.8 Ton 129 157 14 1282 Wood Pallets (82%) and PUR Ingason and Mattresses (18%). Runehamar Lonnermark Test Series, Test 2, u = 3 m/s A Leyland DAF 310ATi: HGV 87 128 18 970 Grant and Trailer with 2 Tons of Furniture, Drysdale EUREKA 499, u = 36 m/s A Trailer with 8.5 Ton Furniture, 152 119 10 1281 Fixtures, and Rubber Tires. Ingason and Runehamar Test Series, Test 3, Lonnermark u = 3 m/s A Trailer Mock-up with 3.1 Ton 67 67 14 1305 Corrugated Paper Cartons Filled Ingason and with Plastic Cups (19%), Lonnermark Runehamar Test Series, Test 4, u = 3 m/s (continued on next page)

OCR for page 28
34 TABLE 6 (continued) Peak Peak Type of vehicle, model year, test Calorific HRR Time to Temperatures in Reference nr. Value (Qmax) Peak HRR Tunnel Ceiling [see Ingason u = longitudinal ventilation m/s (GJ) MW (min) (C) (28)] HGV A Trailer Load with 72 Wood 19 26 12 600 Pallets. Second Benelux Tests, Lemaire et al. Test 14, u = 12 m/s A Trailer Load with 36 Wood 10 13, 19 16, 8, and 8 400, 290, 300 Pallets. Second Benelux Tests, and 16 Lemaire et. al. Tests 8, 9 and 10, u = 1.5, 5.3, and 5 m/s A Simulated Truck Load (STL), 63 17 15 400 Ingason EUREKA 499 Source: Ingason (28). N/A = not available. in the Runehamar Tunnel and it varied between 16.4 and 26.3 In these conditions, a large amount of recorded data would MW/min (55.9 and 89.7 MBtu/hr/min). be helpful to build interpretations concerning the phenomena developed during the fire. The type of measurement instru- The measured ceiling temperatures varied from 110C to mentation and its location on three-dimensional (3D) mesh 1365C (230F to 2489F). These temperatures can be com- appears fundamental for the analysis of tests results. pared with standardized timetemperature curves for load- bearing design in buildings and underground construction. The goal of most of the experiments was not to research After one hour of exposure, the temperature exceeded 925C the physical relations of the phenomena, but to check specific (1697F). equipment or materials being sponsored by the vendors. It is difficult to obtain general laws from the full-scale experi- The results in Table 6 indicate that there is a correlation ments; however, general observations under specific condi- between high HRR and high temperatures. Ingason has tions can be made. This is the result of the relatively low shown that the highest temperatures (>1300C or 2372F) number of experiments performed in each program. For exam- are obtained with HRRs larger than 20 MW (68 MBtu/hr) ple, the Japanese tests were partly planned to provide qualita- and low ceiling heights (approximately 4 m to 5 m) in com- tive information about the escape routes in different air bination with intermediate ventilation rates. For high HRR, velocity control conditions. This target does not appear to the flames reach the ceiling and the combustion zone where the be compatible with the use of the results in scientific models. highest temperatures are usually found. It is located close to the ceiling, even when the longitudinal ventilation deflects the Because of the uncertainties on the measurement results, flames. When the longitudinal ventilation rate increases fur- the interpretations generally concluded that the calculated ther, the cooling effects predominate and the temperature HRR is linked to the method used for its evaluation. drops again. The geometrical shape and size of the fire, the tun- nel cross section (especially the height), and the ventilation rate are thought to be the principal parameters that determine The full-scale experiments generally provide interesting the temperature level at the ceiling. (See web-only Appendix qualitative observations. For example, some opacity situa- E for additional information.) tions appear clearly as a combination of the HRR, the nature of the burning object (smoke density), and the longitudinal air velocity. The relatively low number of experiments does General Observations on Large-scale Tests not lead to general laws or conclusions. (An exception would Based on Reported Results be the Memorial Tunnel program because of the large num- The recent research programs are based on complete mea- ber of tests.) These observations might be used as a reference surement systems. They use numerous instrumentations and for more specific research using appropriate tools (small- are organized into networks quite similar to the mesh used in scale or numerical models). CFD models. In general, the measurements made during the experi- One of the characteristics of these experiments is that no ments can be used as a basis for simulations and particularly access is possible in the fire area. No visual observation is for CFD. The qualification of a simulation tool must follow then possible, except when a video camera is installed in that several rules: zone. In some cases, operators could be present in the sec- tions located upstream from the fire. This situation cannot Thematic: a reference experiment must deal with fires in provide an overview of the experiment. tunnels. Cold smoke tests cannot represent fire behavior.

OCR for page 28
35 Reliability: the quality of the results must be correct. These tests are generally performed in tunnels before they Appropriate instrumentation shall be used. are put into operation to demonstrate if the smoke extraction Representatively: the measurements have to describe as system will work correctly if an accidental fire occurs. The completely as possible the phenomena that have to be recent developments of such tests show that the efficiency of characterized by the numerical simulation. the ventilation is linked both to its quantitative capacity and Adaptability: even if the previous characteristics are to the way it is operated. As this second point is never treated satisfied, the reference experiment must be adapted to by recommendations or regulations, specific developments a comparison with simulation. For example, chaotic are necessary to determine optimal reactions adapted to the behaviors linked to uncontrolled fires such as vehicle fires fire (location, HRR, natural ventilation, and so forth). are not easy to understand and to integrate as boundary conditions. The second goal of these tests is to show the operators how to react in case of a fire. The tests may be completed None of the large-scale tests completely meet those require- with fire department exercises and intervention evaluations. ments because of the relatively small number of tests with real vehicles. PIARC (21) suggests performing tests before opening the tunnel to establish instructions for fire situations. The sec- The number of experiments is limited because of the huge ond kind of test, suggested during operation, is used to train costs involved in such programs (about $40 million USD for operators and fire departments. The tunnel must be closed the Memorial Tunnel program). These costs lead to limiting specifically for these tests. One of the PIARC report recom- the duration of the program and, as a consequence, the num- mendations is to conduct such tests regularly. ber of affordable experiments. Because the HRR is limited, it is possible to observe the Most of these tests were performed in abandoned tunnels. phenomena in different zones of the tunnel, even near the For a road application, extrapolations are often necessary fire. These observations may be correlated with the measure- because of the reduced cross section and its different shape ments (smoke motions compared with temperature fields, (e.g., horse shoe instead of rectangular or other shape). backlayering evolution, and stratification downstream of the fire, and so forth). Tests in Tunnels Before or Under Operation Many tests can be performed in a rather short time. It is estimated that about 20 fires can be studied in one week, con- There is a requirement and a standard practice in most coun- sidering safety precautions. tries for performing tests before a tunnel is opened. In the United States, the typical requirement is to test all the sys- Instrumentation is limited, but the evolution of these tests tems and perform a cold smoke test for witnessing the smoke tends to increase the number of sensors. Also, the total amount movement. Typically, there are no requirements for hot smoke will be limited because this kind of experiment is distinct tests or tests of burning vehicles before commissioning in new from research programs; in particular, it will be difficult to U.S. tunnels. characterize the phenomena occurring at large distances from the fire zone. Many European countries perform small-size (35 MW or 1017 MBtu/hr fire) hot smoke tests, burning a pan with The size of the fire must also be limited because these tests fuel, while activating the fire life safety systems and simulat- must be nondestructive. Actually, it is necessary to limit the ing emergency response procedures. Tests in tunnels before product "Heat release rate Duration." Tests involving 20 MW they are put into operation are generally done with calibrated (68 MBtu/hr) sources were performed, but this value is consid- fires such as fuel pools or wood cribs. Pool fires can be used ered an exception. Generally, the test fires do not exceed 5 MW to obtain steady states, which are needed to measure the (17 MBtu/hr). During passenger cars tests, peaks of 7 to 8 MW combustion rate to evaluate the HRR. There is a substantial (24 to 27 MBtu/hr) were observed, but they did not last long. amount of information on heptane pool fires. Diesel oil can be used to avoid explosions or to produce more smoke. The Puymorens and Chamoise Tunnel tests have been based on heptane pool fires (21). Many different steady states In France, to be more demonstrative, they usually burn have been characterized and these results have been used cars in new tunnels before commissioning a tunnel system. to determine ventilation requirements. They have also been Although it is more expensive, it provides a better simulation analyzed from a scientific point of view to determine the gen- of an actual fire event, because the HRR is very chaotic and eral laws governing smoke motion and other thermodynamic unpredictable. The tunnel ventilation system effect is better behaviors. For example, during the Chamoise Tunnel tests, characterized when the thermal situation is stabilized in the it was possible to measure the backlayering distance in each pool fire tests; therefore, the use of cars as fire loads is rec- case (Figures 9 and 10). The complete analysis of the various ommended after the fire pool tests are done. parameters shows that the backlayering distance may be