Design Fires in Road Tunnels (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

157 APPENDIX E Fire Tests E-1 Full Scale Tests E-1.1 Ofenegg Tunnel Tests Ofenegg Tunnel (Switzerland 1965) (21) These tests were carried out in order to study the ventila- tion capacities in the case of a fire under the large Swiss tun- nel projects of the sixties. The total cross-sectional area of the Zwenberg and Ofenegg tunnels was approximately 24 m2 (258 ft2), which is much smaller than the cross-sectional area of normal road tunnels with two lanes, which is between 45 m2 (485 ft2) and 60 m2 (650 ft2). The facility was a railway tunnel with a dead end located 190 m (620 ft) from the portal. About 11 fires were performed using fuel pools from 6.6 m2 to 95 m2 (71 ft2 to 1,023 ft2). Gasoline was poured into a concrete tub and then ignited. The gasoline used was regular gasoline (86% carbon and 14% hydrogen) with a density of ρ = 730 kg/m3 (at 15°C) or 45.6 lb/ft3 (at 59°C) and a lower calorific value of approxi- mately 44 MJ/kg (18,917 Btu/lb). The rate of burning of gaso- line in free air is a function of the fire area. It first increases as the fire site increases in size and then remains constant when the fire site reaches an area of approximately 1 m2 (11 ft2). The Ofenegg report details a number of tests performed in an abandoned Swiss railway tunnel to investigate the CO con- centration, temperature distribution, visibility, response to ventilation, response to sprinklers, effect on tunnel systems and structures, and effect on vehicles and people of several fire sizes as a function of time. Several animal carcasses and vehicles were exposed at various distances to deliberately ignited pans of fuel. During the 500 L (132 gal) fuel tests, the semi-transverse supply had no mitigating effects, while the longitudinal venti- lation “drove the flames torch-like” downwind. During the 500 L (132 gal) sprinkler test, sprinkler droplets initially evap- orated into a high-temperature steam cloud, causing more damage than the unsprinklered fires. The open fire was appar- ently soon extinguished, accompanied by a strong odor of fuel at the portal, but the fire reignited after 17 minutes (status of sprinkler flow unstated) with significant but non-explosive wave-front propagation. During the 1000 L (264 gal) fuel tests, calculated burning rates were lower than those observed for similarly sized fires in the open. Started immediately after ignition, the sprinklers reduced the maximum ceiling arch temperatures from, but the steam apparently pushed burning gases and gasoline vapors into adjacent tunnel sections, where they continued to burn. The fire was apparently extinguished for 10 minutes, but the tunnel filled with fuel vapors, which exploded in the l9th minute. This caused extensive damage to the test facility injuring three technicians. All three incidents caused doubt on the effectiveness of sprinklers in containing a fire or in limiting the range and severity of damage E-1.2 Zwenberg Tunnel Tests Zwenberg Tunnel (Austria, 1975) (21). The ignited fuel areas were 6.8 m2 (73.2 ft2) and 13.6 m2 (146.4 ft2). The performed measurements were: temperature, gas concentration (CO, CO2, NOx, O2), opacity, and combustion rate. Tests were commissioned by the Australian Ministry for Construction and Technical Affairs. They were carried out in an abandoned rail tunnel equipped with a fully transverse ventilation system. The investigators attempted to answer the following questions: • How do conditions in the traffic space differ when apply- ing different patterns of ventilation? • What improvements can be expected from selected changes to the design, construction, and operation of exhaust air openings? The test program consisted of 23 tests of a “standard” fire using 200 liters (52.8 gal) of gasoline with a fire area of 6.8 m2 (73 ft2), three tests using 400 liters (106 gal) of gasoline with a fire area of 13.6 m2 (146 ft2), and four other tests using other fuels. These tests investigated the effect of varying five parameters: • Location of fresh air injection (high or low). • Quantity of smoke and fumes exhausted. • Quantity of fresh air injected. • Forced longitudinal ventilation in the traffic space. • Conditions in the traffic space (open or obstructed). The investigators believe the size of the area affected by the fire and thus the possibilities of escape and rescue depend to a great extent on the pattern of ventilation, more so than on any other parameter. With longitudinal flows of at least 6.5 ft/s (4.4 mph or 7.1 km/h), a “burner effect” was created on the exhaust air side of a fire. The smoke spread at approximately the same rate as the longitudinal flow (for the 200 L fires or

158 52.8 gal), but even small fires filled long sections of the tunnel on the exhaust side of the fire point with smoke. They suggest that it is not possible to rescue people on the exhaust air side from the fresh air side. Contrary to the condi- tions on the exhaust air side, however, a longitudinal flow cre- ates very favorable conditions on the fresh air side of the fire. If the longitudinal flow can be stopped or if none exists from the start, the danger area and the smoke area will be symmet- ric to the fire point. The tests confirmed that full extraction in connection with throttled fresh air reduces the danger area as well as the smoke area. Maximum exhaust air temperature reached during the full extraction tests was only 85°C (185°F) and decreased as the fire point approached the fan location. With this dilution, the investigators believe 250°C (482°F) is a sufficiently high tem- perature criteria for exhaust fans installed in a fully transverse system. This does not agree with actual conditions experienced in the Holland Tunnel and Caldecott fires. It was concluded that: • The fans allow for command from the control center to be executed within a very short period of time. • A fire alarm program for each tunnel specifies in detail the operating pattern of the ventilation system in relation to the location of the fire and other marginal conditions. • In cases where the control center is equipped with a com- puter, the individual programs are stored and available to be called off at any time. Regarding the location of fresh air injection and exhaust openings: • The overriding recommendation derived from the tests requires throttling of the fresh air supply (or change-over to extraction in case of a reversible semi-transverse sys- tem) in case of a fire. • When the fresh air supply is throttled, the injection “from below” shows no decisive advantage compared with the injection from “above.” • The only conclusion gained during the tests is that the enlargement of the exhaust openings near the fire point has no effect as long as a considerable (6.5 ft/s, 4.4 mile/h, or 7.1 km/h) longitudinal flow passes over the fire point. • In fully transverse systems, the immediate action must be to get longitudinal flow under control before trying to make further improvements by enlarged exhaust openings. E-1.3 PWRI Experiments The Japanese full-scale test program (Japan, 1980) used a 700 m (2,300 ft) long gallery built by the Public Works Research Institute (PWRI) and a 3300 m (10,830 ft) long road tunnel. Sixteen (16) experiments were performed in the gallery and 8 in the tunnel. The fire sources were fuel pools (10 tests with 4 m2 or 43 ft2, 2 tests with 6 m2 or 64.6 ft2), passenger cars (6 tests), and buses (6 tests). The physical conditions measured in the tunnel during the fires were based on the emergency capabilities. The influence of the longitudinal airflow velocity was tested. Other tests included oversized exhaust ports for smoke removal. The important results of this investigation were reported as follows: • Best smoke removal was achieved by operating both east and west fans for extraction regardless of the fire location, with the bulkhead damper fully open. • Under these conditions, air flowed toward the open dampers by as much as 5 meters per second (11 mph or 17.7 km/h). • The space between the fire point and the open damper or dampers is filled with smoke. • The inertial effect of longitudinal air flow is lost within three minutes after fire mode is activated. E-1.4 Repparfjord Tunnel Tests Near Hammerfest (Norway, 1990–1992) (21) These experiments were performed in an abandoned 2.3 km (1.4 mile) long mining gallery (rough wall surfaces and cross section varying from 30 to 40 m2 or 323 to 430.6 ft2). They gathered nine European countries (these experiments were the base of the EUREKA 499 “Firetun” project). A total of 21 tests were performed using rail and metro vehicles, passenger cars, heavy goods vehicles, and calibrated fires (heptane pools and wood cribs). About 400 sensors were installed along the tun- nel and inside the fire loads. The measurements dealt with air and wall temperature, velocity, opacity, gases concentration, smoke motion (video network), and so forth. In these tests performed in Norway, special attention was paid to the smoke development and the smoke dispersal result- ing from the combustion of vehicles (cars and trucks). The fire load was between 5,000 MJ (4.7 MBtu) for cars and 90,000 MJ (85.3 MBtu) for heavy goods vehicles. One fire test was performed with n-heptane C7H16 (84% C and 16% H). The density of n-heptane is about 680 kg/m3 (at 15°C) or 42.5 lb/ft3 (at 59°F), the calorific value is approxi- mately 44.4 MJ/kg (19,089 Btu/lb). Therefore, this fuel is very similar to gasoline or diesel oil. The mean value of the tunnel cross section was approximately 30 m2 to 35 m2 (323 to 377 ft2). As compared to fire tests performed with gasoline, diesel oil, and n-heptane, special attention must be paid to two factors that heavily influence the smoke development and the dispersal of smoke in fires involving real road vehicles: • The materials used for the vehicle construction (without load) are flame-retardant and hardly combustible.

159 • The natural initial temperatures at the tunnel wall in the test tunnel were relatively low. In addition, the tunnel wall was roughly excavated and very rough, so that the heat released was rapidly conveyed to the rock. Both factors retard the heat release and thus the smoke development, and they reduce the fire temperatures compared to fuel fires. On the other hand, these fires last much longer than fuel fires. In addition, the smoke temperatures decreased rapidly with increasing distance from the fire site. This allowed the smoke to become more quickly cooled down and then sink to the ground. The total tunnel cross section was filled with smoke. In contrast to other fire tests, where there is normally a ground zone without smoke, at least for a period of time, there was no such free zone during these fires (except in the case of a wood fire). Therefore, the conditions in this test were signifi- cantly worse than in the case of fuel fires. E-1.5 Memorial Tunnel Tests: Memorial Tunnel (United States, 1993–1995) (21, 25, 26) Description of Facility • Length: 2,800 ft (853.4 m) • Cross section: Former two-lane road alignment This facility is an abandoned two-lane tunnel near Stan- dard, West Virginia. The tunnel was converted to a fire venti- lation laboratory in 1993 to study the behavior of smoke and heat under various ventilation systems (see Figure E1). Instru- mentation includes temperature sensors, video cameras, and velocity probes. In contrast with the Zwenberg Tunnel and the Ofenegg Tunnel, the cross section in this tunnel was represen- tative of usual road tunnels (approx. 60.5 m2 or 651 ft2 without intermediate ceiling). Diesel oil was used as a fire source. The density of diesel oil is between 815 kg/m3 (50.9 lb/ft3) and 855 kg/m3 (53.4 lb/ft3) at 15°C (59°F). The lower calorific value is 42.5 MJ/kg (18,284 Btu/lb). In terms of weight per- centage, diesel oil mainly consists of carbon (86%) and hydro- gen (14%). The stoichiometric air consumption is 14.5 kg (32 lb) of air per kilogram of diesel oil. Except for the fact that diesel oil ignition qualities are not as good as those of gasoline, there are no major differences between diesel oil and gasoline in terms of smoke development and in terms of smoke dispersal. E-1.6 Runehamar Tunnel Tests Runehamar Tunnel Tests (27) In total, four tests were performed using a simulated HGV. In three tests, mixtures of different cellulose and plas- tic materials were used. In one test, a “real” commodity, consisting of furniture and fixtures, was used. In all tests, the mass ratio was approximately 80% cellulose and 20% plas- tic. A polyester tarpaulin covered the cargo. The reason for using furniture in one of the tests was to provide a compari- son to a past test (EUREKA 499), which was carried out with similar materials and a very high ventilation rate of 6 m/s (1,180 fpm) at the start of the test. This provided a good point of reference between the data from Runehamar and the EUREKA tests. In the first two fire tests, Test 1 and Test 2, a pulsation of the fire was experienced during a time period when the fire was over 130 MW (444 MBtu/hr). This created a pulsating flow sit- uation at the measuring station. The measurements showed that the maximum velocity was pulsating in the range of 3 to 4 m/s (591 to 787 fpm) down to a minimum in the range of 1 to 1.5 m/s (197 to 295 fpm). The frequency of the maximum velocities was about 45 seconds during this period. Since the air mass flow rate is dependent on the air velocity the HRR also pulsate during this period. E-1.7 UPTUN Project Tunnel Tests (28) This project was discussed earlier. The WP2 was devoted to the analysis of fire development in tunnels and potential mit- igation measures. Design fire scenarios and associated design fire curves were proposed by UPTUN WP2, and used as input to other work packages within UPTUN. These design fires can also be used in more general terms since they are based on current knowledge about fire scenarios as well as information created within the UPTUN project. All of the large vehicles have been burned in a tunnel, whereas passen- ger cars have either been burned under a calorimeter or in a tunnel. Small pool fires and small idle pallet fires, with a potential heat release rate of 10–20 MW (34–68 MBtu/hr), were also tested. A characteristic of the UPTUN experiments is the use of real road and rail vehicles as fire loads. The heat release rate FIGURE E1 Measuring equipment in the Memorial Tunnel; velocity cabinet, data acquisition unit, and instrument tree (26).

160 of such fires was one of the unanswered fundamental ques- tions for fire life safety systems design. This project was discussed earlier. The WP2 was devoted to the analysis of fire development in tunnels and potential mit- igation measures. Design fire scenarios and associated design fire curves were proposed by UPTUN WP2, and used as input to other work packages within UPTUN. These design fires can also be used in more general terms since they are based on cur- rent knowledge about fire scenarios as well as information cre- ated within the UPTUN project. All of the large vehicles have been burned in a tunnel, whereas passenger cars have either been burned under a calorimeter or in a tunnel. Small pool fires and small idle pallet fires, with a potential heat release rate of 10–20 MW (34–68 MBtu/hr), were also tested. A characteristic of the UPTUN experiments is the use of real road and rail vehicles as fire loads. The heat release rate of such fires was one of the unanswered fundamental ques- tions for fire life safety systems design.