Key Findings of Liquefied Natural Gas Pool Fire Outdoor Tests with

ARTICLE pubs.acs.org/IECR

Key Findings of Liquefied Natural Gas Pool Fire Outdoor Tests with Expansion Foam Application Geunwoong Yun, Dedy Ng, and M. Sam Mannan* Mary Kay O’Connor Process Safety Center, Artie McFerrin Department of Chemical Engineering, Texas A&M University System, College Station, Texas 77843-3122, United States ABSTRACT: The unique properties of expansion foam in blanketing the surface of most hydrocarbon fuels have made it possible to be used as a mitigation measure against a boiling and evaporating pool of flammable gases and subsequent pool fires. Because of this fire suppression characteristic, the liquefied natural gas (LNG) industry has identified expansion foam as one of its safety provisions for pool fire suppression. However, the effectiveness and key parameters of foam in controlling LNG fires have not been thoroughly investigated from previous field tests. In this paper, we investigated the effects of foam application on LNG pool fires through outdoor spill experiments at the Brayton Fire Training Field. The primary objectives of this study are to identify the foam effectiveness in suppressing LNG pool fires and to determine the thermal exclusion zone, by investigating temperature changes of foam and fire, profiles of radiant heat flux, and fire height changes with foam. Additionally, a schematic model of a LNG-foam system with fire for theoretical modeling was also developed. Results showed that expansion foam has positive effects on reducing flame height and radiant heat flux by decreasing heat release and radiant heat feedback from the LNG pool fire, ultimately reducing the safe separation distance. Through extensive data analysis, we also identified several key parameters, such as the minimum effective foam depth and the mass-burning rate of LNG with applied foam. Results from this study can be used to design an effective expansion foam system as well as to develop defensive measures and emergency response plans for mitigating the consequences of LNG releases.

1. INTRODUCTION Liquefied natural gas (LNG) primarily consists of methane that has been supercooled to its liquid state at atmospheric pressure. Although LNG has gained wide acceptance as an alternative fuel with much lower air emissions than other fossil fuels, the safe handling of LNG has become the subject of widespread controversy.1-5 LNG is flammable when the methane vapor is mixed with air in volumetric concentrations between 5% and 15%.1,2 If LNG is spilled on the ground, a liquid pool will be formed that can result in a pool fire in the presence of an immediate ignition source. If delayed ignition takes place in the vapor cloud during the release of LNG, then a flash fire may occur and subsequently result in a pool fire caused by burning back from the vapor cloud to the LNG pool. The primary hazard of pool fires arise from the radiant heat flux emitted by the fire to the surrounding plant. Therefore, NFPA 59A requires a thorough risk assessment of potential release consequences when designing LNG facilities.6,7 Most notably, the standard requires that LNG storage tanks should have dikes or secondary impounding walls for holding accidental spill from tanks. The standard also requires LNG facilities to design a “thermal exclusion zone” so that thermal radiant flux from an LNG fire will not create damage beyond the property line, as measured from the distance to a thermal radiation intensity level of 5 kW/m2.6,8,9 Expansion foam has been used as mitigation measures for suppressing most nonliquefied hydrocarbon fires. Since the foam has lower density than most flammable liquids, it flows freely over the fuel surface and blankets the spilled area to provide insulation to the fuel surface and prevent fuel from reaching the fire. Subsequently, the water content in foam cools the fire upon contact and generates adequate steam to reduce the oxygen conr 2011 American Chemical Society

tent of the surrounding area, resulting in suppression of combustion and extinguishment of a fire by smothering action. Since LNG fires pose greater hazards than most nonliquefied hydrocarbon fuels, there is an urgent need to investigate the effectiveness of foam for mitigating LNG fires. Little understanding of the physics on the application of expansion foam on LNG pool fires came from a literature review. Most of these experiments were performed to understand the applicability of the foam onto different medium. University Engineers, Inc. investigated the effectiveness of expansion foam on LNG spills on land in pool fire control by measuring the LNG burning rate and the radiant heat flux emitted by a fire.10 Their findings identified that a 0.91 m (3 ft) foam depth can reduce the radiant heat flux; however, no information was given on the distance to average heat flux of 5 kW/m2 for determining the thermal exclusion zone. In addition, the study reported that small-bubble foams tend to reduce the mass-burning rate while large-bubble foams increase the burning rate during fire control. This study presented good understandings of foam effectiveness in terms of bubble size and expansion ratio; however, it did not provide information on the variation of foam temperature, required foam thickness for effective fire control, and the reduction of radiant heat fluxes in different wind directions. In the report of the FOAMSPEX project, Persson et al. reported simulation studies of foam spread and extinguishment in largescale foam application.11 Their studies were based on foam Received: June 26, 2010 Accepted: December 21, 2010 Revised: December 14, 2010 Published: January 19, 2011 2359

dx.doi.org/10.1021/ie101365a | Ind. Eng. Chem. Res. 2011, 50, 2359–2372

Industrial & Engineering Chemistry Research

ARTICLE

spread experiments over water, fuel oil, and lubrication oil with or without fire in a channel and a pool configuration. The experiments revealed that the foam spread is influenced by gravitational forces and is caused by the hydrostatic pressure differences within the foam. On the contrary, the resisting force was identified as viscous friction between the foam blanket and the fuel beneath it in a pool fire. Attempts were made to study the foam coverage time for estimating foam application rates; however, the study was limited only to fuel oils. The reduction of radiated heat flux by foam was not investigated. Recently, Suardin et al. at the Mary Kay O’Connor Process Safety Center (MKOPSC) conducted a few field tests to investigate the application of expansion foam on an LNG pool fire at the Brayton Fire Training Field (BFTF), which is affiliated with the Texas A&M University System.12,13 These tests identified that a foam application rate of 10 L/min 3 m2 is effective for LNG pool fire control on concrete containment pits. The pool fire control time (the time required for achieving 90% radiant heat reduction, which is recommended by NFPA 11A14) was reduced by increasing the foam application rate. These tests revealed that the design of the containment pit could result in different movement patterns and shapes of LNG pool fires. Currently, the most widely used model for predicting thermal hazard distance is the solid flame model. This model assumes that fire has a simple geometrical shape, typically a cylinder.1 The model represents fire as a surface emitter of radiant heat flux represented by eq 1 below15 : q00 ¼ EFτ ð1Þ where q_ 00 is the radiant heat flux received by an object, E is the surface emissive power, F is a geometric view factor between the radiation receiving object and the radiation emitting parts of the fire, and τ is the transmissivity of the atmosphere to thermal radiation from fire. This equation assumes that emissive power is constant over the entire surface of the fire and the base diameter of the fire is equal to that of the burning liquid pool.2,15 The surface emissive power (E) and the geometric view factor (F) are functions of flame height; thus, it is important to identify the flame height for modeling the radiant heat flux. Several correlations have been developed to estimate the flame height, and the most widely used is the Thomas correlation, as shown in eq 2 below2 0 1q L m_ 00 ¼ AFcp ðU Þq ¼ A pffiffiffiffiffiffi D Fa gD

C !p B B C B Uwind C B C !1=3 C B B m_ 00 C @ A gD Fa

ð2Þ

where L is the height of fire, D is the fire diameter, Fa is the density of air, g is the acceleration due to gravity, and A, p, and q are the correlation constants. Fc is the combustion Froude number, which is also called the dimensionless burning rate, and is given in eq 3. U* is the dimensionless wind speed and is depicted in eq 4. Fc ¼

m_ 00 pffiffiffiffiffiffi Fa gD

ð3Þ

U ¼

Uwind !1=3 m_ 00 gD Fa

ð4Þ

Since expansion foam is used to control LNG pool fires and reduce the radiant heat prior to fire extinguishment, it may be useful to modify eq 2 for estimating the flame height controlled by the expansion foam to provide identification of the mitigated fire hazard. Although previous research have identified that expansion foam has a positive contribution on LNG pool fire control,10-13 further research is required to obtain more concrete evidence on the effects of foam and the properties generated through continued experiments and modeling efforts. Driven by this motivation, the objectives of this paper are summarized as follows: (1) To investigate the effectiveness of expansion foam on LNG pool fire control by identifying the extent of reduction in the radiant heat flux caused by expansion foam in different wind directions (upwind, downwind, and two crosswinds) and determine heat flux contours emitted by the fire. These findings were used to identify thermal hazard distances at the radiant heat flux level of 5 kW/m2. In addition, the LNG mass-burning rate after foam application and temperature profiles of foam and fire were measured to support the findings from foam experiments. (2) To observe changes in fire sizes (length) due to foam application and correlate the data to model the relationship between the flame height and the diameter of the fire base. The proposed equation can be used to estimate the foam-controlled flame length as a function of LNG massburning rates and pool fire diameters. (3) To develop theoretical models for explaining the underlying mechanisms of foam on an LNG pool fire. Utilizing the combination of proposed model and experimental data allows the estimation of effective foam depth for both pool fire suppression and vapor dispersion control. It is expected that results from this study would allow LNG industries to develop firefighting strategies and emergency response plans for an existing design and/or new installation of expansion foam systems.

2. OVERVIEW OF RESEARCH 2.1. Methodology. This study was carried out in two parts: (1) medium-scale field tests at the BFTF to gather experimental data of radiant heat fluxes, fire sizes, and temperature profiles and (2) theoretical study of the foam-controlled fire lengthdiameter correlations with obtained experimental data and construction of a schematic model to depict the interaction of expansion foam with an LNG pool fire. 2.2. Experimental Setup. Outdoor medium-scale experiments were designed to spill LNG on a confined concrete pit to simulate LNG release scenarios on land. Two medium-scale field tests were conducted in a large pit (6.40 m 10.06 m 1.22 m) at the BFTF in March and December 2009 to identify key parameters of foam effectiveness on LNG fire control. Figure 1 shows the schematics of the test setup. A total of 166 thermocouples were mounted vertically on steel frames inside the pit in 2360

dx.doi.org/10.1021/ie101365a |Ind. Eng. Chem. Res. 2011, 50, 2359–2372


ARTICLE

Figure 1. Setup for medium-scale field test.

Figure 2. Setup for radiometers.

16 locations to measure the temperatures of expansion foam (or LNG vapor) and fire. Several thermocouples were installed up to 20 cm (8 in.) elevation from the bottom of the pit to measure the changes of LNG level. One wide-angle radiometer was placed in the center of the pit. Figure 2 depicts the setup for the wide-angle radiometers outside the pit during March and December 2009 tests. During the tests, prevailing wind directions were opposite of each other; thus, foam generators and the data acquisition system were moved accordingly in the crosswind direction. In the March test,

several radiometers were placed in the crosswind direction, whereas these radiometers were positioned in the downwind direction in the December test. During both tests, one portable radiometer, which gives the radiative heat flux readings in real-time, was carried to decide the final locations of R5 and R7 as depicted in Figure 2a, for identifying the safety distance at an intensity level of 5 kW/m2. However, during the December test, the locations of R6 and R7 in Figure 2b could not be changed because these radiometers were located downwind of the fire. 2361



ARTICLE

Figure 3. Experimental procedure.

Table 1. Summary of Experimental Facts for 2009 Tests conditions

variables

March 2009

December 2009

note

atmospheric

temperature

22.42 ( 0.03 °C

14.31 ( 0.16 °C

the two standard deviations have 95% level of confidence

wind speed

3.26 ( 0.42 m/s

1.25 ( 0.39 m/s

maximum wind speed: March test, 6.70 m/s; December

wind direction

S and SSE and W

N and ENE

relative humidity solar radiation

51.90 ( 1.73% 0

44.17 ( 0.51% 64.00 ( 8.15 W/m2

during March test, it was night from 7:50 to 7:59 pm.

methane

99.85%

99.87%

other components: nitrogen and ethane

classified as “high expansion foam”

conditions test, 1.8 m/s

LNG conditions

composition foam conditions

expansion ratio

500:1

500:1

foam application rate

6.5 L/m2 3 min

6.5 L/m2 3 min

foam concentrate

Jet-X

Jet-X

Figure 3 illustrates the schematic drawings of the experimental procedure. Initially, LNG was spilled into the pit through a 0.10 m stainless steel pipeline up to a 0.20 m (8 in.) depth. As shown in test 1 of Figure 3, high expansion foam was applied simultaneously by two foam generators and filled the pit up to 2.44 m (8 ft) depth. The pit was enclosed with a 1.22 m vapor fence of wooden walls. For test 2, the wooden walls surround the pit were removed and the LNG was ignited using a torch on a 3 m long pole. The foam application was suspended until the previously applied foam from test 1 had consumed completely. When the LNG fire began to develop at a certain time, the foam was reapplied to suppress the fire. The temperature and the radiant heat flux were measured during the whole test period. Since the results of test 1 in Figure 3 have been published elsewhere,16 they will not be reported in this paper.

3. TEST RESULTS 3.1. Summary of Experimental Facts. Table 1 summarizes the test variables and their corresponding measurements during March and December 2009 tests. 3.2. Measurement of Flame Height. The first step in quantifying the foam effectiveness on LNG fire control is to mea-

components: HMIS surface active agents, ethyl alcohol, lauryl alcohol, glycol, inorganic salts

sure the flame height. Figure 4 shows actual snapshots of the pool fire before and after foam application during the March test. The still photos were taken with a video camera which was placed at a fixed position in a crosswind direction. The reduction in fire size was evaluated by comparing its flame height. Right after the foam application the fire grew larger rapidly and then decreased significantly after a certain period of time, indicating fire control at steady-state. However, this analysis was not able to provide quantitative values for the reduction of the flame length due to foam application. Using an alternative approach, we attempted to obtain the flame height over time from video clips using ImageJ software (National Institutes of Health).17 As seen in Figure 5, the flame height increased steadily up to 13.13 m during free burn (36 s period); subsequently, foam was applied at 37 s. Right after foam application, the flame height grew rapidly up to 17.17 m. The rise in fire size might be attributed to the discharge of water on the pool surface, which passed the foam-forming screen without being converted into foam at the beginning of foam generation. Moreover, the water content of the foam could also contribute to a rapid increase in evaporation rate, resulting in an exaggeration of the fire’s intensity. Five seconds after the foam was applied with a rate of 6.5 L/m2 3 min in the 64.83 m2 pit, the flame 2362



ARTICLE

Figure 4. Pictures of fire before and after foam application.

Figure 5. Flame height over time.

height began to decrease, indicating that the foam was effective at reducing the fire heat feedback to the LNG pool. Approximately 55 s after foam application, it was observed that the flame height reached a stable limit at 5.13 ( 0.38 m (i.e., 61% reduction in the flame height as compared to the initial height of 13.13 m prior to foam application). On the basis of this finding, it can be concluded that the expansion foam was able to reach a steady state within 55 s. As seen in Figure 5, the fluctuations of fire length in the range of 200-210 s were caused by an intermittent reapplication of foam, in order to maintain a 1.22 m foam thickness. 3.3. Mass-Burning Rate. The mass-burning rate of LNG is one of the key parameters in predicting how fast the fire is burning and to calculate the flame height and the radiative heat flux. To date, little work has been done on investigating the mass-burning rate (or liquid regression rate) of LNG on concrete or land in the literature.1,2 A mathematical model to describe the pool burning rate of flammable liquid has been developed on the basis of heat transfer principles.18 However, there has not yet been any model available to describe the LNG burning rate applied with the use of expansion foam. This information will be useful to understand fire characteristics after foam application.

In this study, the mass-burning rate of LNG after foam application was determined from the thermocouple readings, which measure a temperature range from -270 to 1372 °C within the LNG levels. The change in LNG levels was observed by checking the position of thermocouples and the time in which each thermocouple gives a reading of LNG boiling temperature, -162 °C. Subsequently, the level changes were converted into the massburning rate using the LNG density (450 kg/m3 18,19) and the dimensions of the pit. Figure 6 summarizes the reported mass-burning (or evaporation) rates of LNG from the literature and their corresponding test conditions. Figure 6 also shows the burning rate data at steady state from this study. During the March test in 2009, the average burning rate with 1.22 m thick foam was measured as 0.082 ( 0.002 kg/m2 3 s, while a lower value, 0.062 ( 0.005 kg/m2 3 s, was observed in the December test. The differences in this data might be caused by several factors. For instance, the wind speed during the March test was higher than that of the December test. One possible reason is that the higher wind speed increased the tilt of the flame, thereby increasing the heat in the foam layer. Moreover, previous studies have shown that the wind speed can increase the burning rate of a hydrocarbon pool (e.g., hexane, gasoline, and diesel),18,20 2363



ARTICLE

Figure 6. Summary of mass-burning (or evaporation) rate.1,2,17

which is in agreement with our findings. Another parameter to be considered is time. During the December test, the duration of the foam controlled fire test was 17 min, while it was only 9 min in the March test. The longer the pool of LNG is in the pit, the lower the heat input is from the concrete. The reported burning rates for LNG are smaller than the rates from the LNG pool fire on land and on water, without applied foam. This is a reasonable result because 1.22 m foam depth is capable of reducing the burning rate by decreasing the heat feedback of a fire to the LNG pool. Also, the evaporation rate of the LNG release on concrete without fire (0.012 ( 0.001 kg/m 2 3 s) at steady state when the concrete is cooled should be smaller than that of the release with fire and foam. 3.4. Flame Temperature Measurements. Whether expansion foam has an effect on flame temperature has not been confirmed. To confirm this effect, several thermocouples were installed at different elevations above the foam and LNG level. In order to withstand the intense heat of LNG fire, the thermocouples used in this work were protected with abrasion-resistant Inconel over braid with high-temperature ceramic fiber insulation. Figures 7 and 8 show the temperature profiles of the LNG flame over time at different elevations above the foam level. At the beginning, the temperatures increased rapidly due to the development of fire. When the foam was applied, both figures show that flame temperatures increased for some period of time and then fluctuated continuously. In the March 2009 test, the LNG fire was able to reach up to a temperature of 951 °C at the B0 location after the utilization of foam (referring to Figure 7). We believe that the fluctuation and indefinite pattern in the flame temperature are caused by the pulsating nature of fire. Because of this random fluctuation, it was not feasible to conclude that the foam can affect the flame temperature. However, it was clear that the flame temperature of an LNG pool fire decreases at higher

elevations, as shown in Figures 7 and 8, which is in agreement with the previous study on a hexane pool fire.21 These figures also indicate that the magnitude of flame temperature reduction is different for both locations, i.e., the temperature differences at B1 location are greater than the one at B0 location). The temperature distribution is affected by the action of induced wind. As shown in Figure 9, the flame temperature distributions before and after foam applications are tilted and not symmetrical. These phenomena might be caused by the prevailing wind rather than by the introduction of foam and its flow direction. Placing the foam generators in an upwind direction has a higher efficiency than the downwind one, because the wind force can assist in directing the foam discharge toward the LNG pool. 3.5. Temperature Profile of Foam. Minimum effective foam depth is one of the crucial parameters to determine foam application rate and the number of foam generators required for mitigating LNG fire hazards. Here, we attempted to determine the effective foam depth for pool fire scenario. Foam effectiveness on fire suppression is mainly dependent on foam bubble stability (size, uniformity, and ability to retain water). The mechanisms of foam in extinguishing the LNG fire can be explained as follows: (1) the water content of the bubble will evaporate when it is heated up to its boiling point (100 °C), thereby absorbing some portion of heat from fire, and (2) the aggregate of foam bubbles floats on the LNG pool surface and prevents air from entering the foam layer, thus decreasing fire. Therefore, it is imperative to determine the effective foam depth from the top surface of expansion foam in which the temperature is less than 100 °C. Figure 10 illustrates temperature profiles of six thermocouples inside the pit during the March test. Figure 10a shows several thermocouple readings at a 0.84 m elevation in which temperatures greater than 100 °C are observed during a period of stable fire control. On the other hand, most thermocouples at 0.61 m elevation give temperature 2364



ARTICLE

Figure 7. Temperature profile of fire at the B0 location in the pit.

Figure 8. Temperature profile of fire at the B1 location in the pit.

readings below 100 °C, as shown by Figure 10b. This implies that a 0.61 m (1.22 m foam level minus 0.61 m of thermocouple height) foam depth can provide insulation against the fire, thereby reducing the flame height and the flame heat flux. It can be concluded that by considering pool fire scenario, the foam depth should be at least 0.61 m. When determining the effective foam depth for actual facilities, some safety factors should be established to account for the unexpected loss and collapse/shrinkage of foam bubbles during fire extinguishment. It is not known if during a fire test expansion foam could form a frozen layer (and/or ice plates). Here we attempted to answer this question with the temperature profiles of the foam. Figure 11 depicts temperature profiles of three thermocouples installed at the LNG levels during the March fire tests. These thermocouples were initially placed in the liquid level, thus reading the temperature of LNG at its boiling point (-162 °C). When the LNG

was vaporized due to the influence of the fire and the atmosphere, the pool surface were covered with foam layers. As shown in Figure 11, after foam was activated, all thermocouples gave temperature readings below 0 °C for some period of time despite the flame still burning; these regions are designated the “frozen layer” of foam in this paper. The time periods when frozen layers were present were different depending on the height of each thermocouple, and it was observed that lower thermocouples have longer frozen periods, as was expected. In the December 2009 test, ice plates were observed when the test was completed, as shown in Figure 12, despite the fact that the ambient temperature was 14.31 ( 0.16 °C. We believe that the ice layer was methane hydrate, which contained vaporized methane that was captured in the ice structure when it passed through the frozen layer of foam. The formation of ice plates is affected by two factors: the duration of fire tests with foam and ambient temperature. During 2365



ARTICLE

Figure 9. Temperature contours of fire at 3.20 m height inside a pit.

Figure 10. Temperature profiles of foam at 0.84 and 0.61 m elevations from the pit bottom during the fire test.

the December test, the duration of the foam-controlled fire test was 17 min, while it was only 9 min in March test. Because of the longer test duration, we were able to observe ice plates only during the December test. The thickness of the ice plates was found to vary on the basis of the location within the pit. It was observed that the thickest part of the ice at the northeastern corner of the pit was around 5 cm (2 in.). But the eastern edge of the pit had a thicker ice layer, greater than 18 cm (7 in.). The expansion foam at the eastern edge had relatively longer contact time with LNG than any of the other locations, since the two foam generators were placed near the eastern edge of the pit, thus resulting in the formation of thicker ice plates. An article by Vesovic22 states that

the formation of ice layer reduces the evaporation rate because it decreases the temperature difference between LNG and the heating source (fire). Therefore, it has become clear that this ice layer formed by foam can help to reduce the mass-burning rate, thus creating a reduction in flame height and fire radiant heat. For the theoretical study of the foam-LNG system, these frozen layers (including ice plates) should be included to account for its effects on the vaporization rate. 3.6. Radiative Heat Flux. 3.6.1. Heat Flux Profiles and Contours of Fire. The LNG pool fire emits a radiative heat flux to its surroundings, and it can have an adverse impact on personnel and properties on site. Therefore, federal regulation 49 CFR Part 2366



ARTICLE

Figure 11. Temperature of foam at 13.97 cm or less height during fire test.

Figure 12. Observation of ice plate after fire test.

193 and standard NFPA 59A have required that exclusion zones around LNG facilities are evaluated on the basis of a 5 kW/m2 heat flux.6,15 In this study, we attempted to identify how much expansion foam can reduce the thermal hazard zone by measuring flame radiant heat flux from wide-angle radiometers. Figure 13 shows radiative heat flux profiles of the LNG pool fire measured by two radiometers (R8 and R9) positioned in the crosswind boundary. It is clearly shown that after foam application, heat fluxes decreased up to 90% in 55 s. The two heat flux curves show exactly the same profiles and the radiometer (R8) closer to the fire reads higher values than that of the radiometer (R9). Similar patterns of heat flux were also observed in the December 2009 test. Figure 13 also includes the flame height curve, which is blue colored and identical to the curve in Figure 5. It is evident that heat flux profiles have a close similarity to the curve of plume height over time. This demonstration implies that the radiative heat flux of fire has a strong correlation with the flame height. This relationship can also be explained through the solid flame model, as shown in eq 1. According to the model, radiant heat flux is a function of mean surface emissive power (E), geometric view factor (F), and atmospheric absorption (τ). The value of surface emissive power is dependent on the geometry of the fire, especially plume height.1,15,23,24 Geometric view factor is the fraction of energy emitted by the fire to the object on the basis of the view of geometric shapes from the emitter (fire) and receiver. Previous researchers have developed mathematical models for view factor calculations, and it was discovered that the view factor

is proportional to the flame height.25,26 Therefore, it became apparent that the fire radiant heat flux is strongly dependent on the height of flame, as shown in Figure 13. 3.6.2. Estimation of Thermal Exclusion Zone (5 kW/m2). Prior to estimating the thermal exclusion zone, radiant heat fluxes in four directions were measured. During the December test, 10 radiometers were positioned around the pit (64.83 m2 area) and one radiometer was placed at the center of the pit to measure heat fluxes of fire. Figure 14 compares three contour plots of flame radiant heat flux (before and after foam application and after reaching steady fire by foam) in the same scale. At this time the expansion foam was able to control the LNG fire in 48 s in a steady-state manner. The heat flux data were averaged in 5 s to give a clear trend of observed heat flux. The red line in the contour plots represents the thermal threshold distance of 5 kW/m2 flux. As shown in Figure 14b, when foam was applied to an LNG pool fire, the thermal heat flux increased slightly, as shown by the increasing perimeter around the pit center. Subsequently, when fire was controlled by the foam at a steady-state burning, the threshold distance decreased significantly, as illustrated in Figure 14c. These findings agreed with the aforementioned flame height and radiative heat flux observations in Figures 5 and 13. It was noted that the wind speed was relatively low during the December 2009 test, 1.25 ( 0.39 m/s. Thus, the shapes of the contours appear as uniform and circular-shaped, without any significant alterations. Figure 15 shows the thermal hazard distances, which are estimated on the basis of contour plots of heat flux and the corresponding 2367



ARTICLE

Figure 13. Heat flux and fire length.

Figure 14. Changes of heat flux contours with foam.

wind direction. As seen in Figure 15, the thermal hazard distance was reduced with the application of 1.22 m thick foam in the range from 47% to 52%, as measured from different wind directions. One question that may arise is why the distance was reduced only by 52% when radiative heat flux can be decreased up to 90% in the same time span. It can be explained that flame

radiant heat is not linearly dependent on the distance from the fire source to the receptor due to the influence of the geometric view factor and atmospheric transmissivity. 3.7. Simplified Model of LNG-Foam System with Fire. On the basis of experimental findings in the March and December tests, we had previously proposed a simplified model of LNG and 2368



ARTICLE

Figure 15. Reduction of thermal hazard distance with foam.

Figure 16. Schematic model of LNG and expansion foam system with fire.

expansion foam for the vapor dispersion control.16 For the LNG pool fire scenario, it was found that expansion foam may form three layers, as illustrated in Figure 16: a foam breaking layer, a nonfrozen layer, and a frozen layer. The foam breaking layer represents the zone within which foam bubbles are collapsed/ evaporated due to rapid boiling (greater than or equal to 100 °C) of water content within bubbles and is attributed to the heat of

the fire. Although this layer may not reduce the fire size significantly, the cooling effects of foam may be able to reduce the flame temperature to some extent, by absorbing some portion of fire heat. On the other hand, the nonfrozen layer is composed of intact bubbles of foam, at temperatures ranging from 0 to 100 °C. This layer is effective in reducing flame height and vaporization rate by depriving the fire of oxygen and insulating the heat feedback 2369



ARTICLE

Figure 17. Experimental correlation of fire height and diameter with 1.22 m thick expansion foam (adapted and modified from refs 2, 17).

of the fire to the LNG pool. Therefore, the depth from the top surface of foam to the interface between the foam breaking layer and the nonfrozen layer can be identified as a minimum effective foam depth for pool fire control. Subsequently, the frozen layer is composed of ice tubes of vaporized LNG, ice plates, and a bulk of frozen foam, with temperatures below 0 °C. The formation of these layers can be explained as follows: (1) due to heat of a fire, foam may collapse, and then the water content of foam will drain through the foam layers up to the LNG pool surface by gravity and (2) water loses its heat when contacting with the LNG and becomes ice plates (or frozen layers). This layer may be beneficial in reducing flame height and mass-burning rate, as it decreases the temperature differences in LNG and foam and blocks the fire’s heat feedback to the LNG. When modeling the interfaces of a nonfrozen layer and a frozen layer, one should consider the release of latent heat of water in the foam due to freezing for heat balance estimation. In order to model the heat balance of the LNG-foam system for the pool fire scenario, all contributing heat sources should be considered to give accurate explanations of this phenomena. If it is assumed that LNG is the system of interest, LNG may have multiple heat sources: radiative (qf,c) and convective heat input (qf,r) from fire, convective heat input from the bottom side of the expansion foam (qfoam,c), conductive heat from the concrete ground and walls (qg; assuming the concrete pit bottom is a semiinfinite slab), solar radiation (qr), convective heat transfer from the atmosphere (qa), and the latent heat of water in the foam due to freezing (qL,f). Due to the insulation characteristics of foam on the LNG pool surface, the contributions of solar radiation (qr) and the convective heat transfer from atmosphere (qa) may be neglected, since their effects are a lot smaller than other parameters. For simplicity, it is assumed that conductive heat input from the wall may be negligible; this is because the conductive heat input from the ground is much larger than that from the wall, especially

in the case of large areas of confinement (e.g., dike area of the LNG storage tank), thus resulting in a one-dimensional conduction equation, as shown at the bottom of Figure 16. Additionally, there could be negative heat sources accompanying the LNG system, such as latent heat of evaporation (qL,LNG) and latent heat of water in the foam due to boiling (qL,b). Here, the role of latent heat of water in the foam due to boiling (qL,b) may be neglected because it is not strongly significant to the LNG pool due to the existence of a nonfrozen layer. Therefore, the total heat balance can be expressed as follows: qf , c þ qf , r þ qf oam, c þ qg þ qr þ qa þ qL, f ¼ qL, LNG þ qL, b

ð5Þ

If we consider several negligible factors mentioned above, then eq 5 can be reduced to the following expression. qf , c þ qf , r þ qf oam, c þ qg þ qL, f ¼ qL, LNG

ð6Þ

This heat balance equation can be used to solve heat balance problems that arise from the incident scenario of LNG pool fire control by foam and to calculate the mass-burning rate with expansion foam. However, caution must be used when employing this model, particularly when the system of interest is different than the one described here (for instance, the interested system is expansion foam, not LNG). In that case, the heat balance should be modified accordingly. 3.8. Correlation of the flame plume height and flame diameter. As aforementioned in section 1, it is imperative to know the relationship between flame diameter and height in order to determine the experimental values of surface emissive power of LNG pool fires. To achieve this purpose, the Thomas’s correlation of flame height for the diameter of LNG pools, as shown in eq 2, has been widely used in fire modeling. However, this correlation only works when the diameter of the pools is less 2370



ARTICLE

than or equal to 22.9 m (D e 22.9 m).2,15,27 To the best of the authors’s knowledge, no correlation has been developed for flame height-diameter with foam application. This correlation may be useful to identify the reduction of thermal heat flux by foam and to evaluate safe separation distances in LNG industries, wherein the foam system is typically used as an independent layer of protection. Herein, we propose a new correlation with an application of 1.22 m thick foam with the following parameters: the mass-burning rate at steady state (m_ 00 ), plume length (L), and equivalent pool diameter (D = 9.06 m, which is obtained by calculating the area of rectangular pit and converting the area into the diameter of circle), as shown in eq 7 and plotted in Figure 17. In this equation, it is assumed that the effect of pool shape (rectangle or circle) is not significant in flame height to pool diameter correlation and that the foam system worked appropriately. L ¼ 17:40F 2=3 D

ð7Þ

The ratio of L to D has a mean value of 0.62, and its standard deviation is 0.07. This equation assumes that the Thomas’s correlation may also be valid when D e 22.9 m for fire with foam suppression; thus, the same trend in the slope of Thomas’s correlation is also observed in the correlation of this work, as shown in Figure 17. This may be a reasonable assumption, because fire suppression by expansion foam predominates over the physical mechanisms, rather than the chemical reactions (or combustion mechanics). Although this correlation provides a useful tool to determine the relationship of flame diameter and height in the foam system, more experimental work is needed to validate the present correlation thoroughly.

4. CONCLUSIONS The paper summarizes the outdoor LNG field tests for evaluating the foam effectiveness in the LNG pool fire scenario. Additionally, a wide range of data analysis was reported to identify important parameters such as (1) mass-burning rate, (2) effective foam depth, (3) the relationship between flame height and radiative heat flux, (4) thermal exclusion zone, and (5) the correlation of LNG plume height to the diameter with foam application. From this work, it was confirmed through image processing analyses of the fire that expansion foam can reduce the flame height by at least 61%, where the foam insulation provides a barrier to the fire’s heat feedback to the LNG pool. The massburning rate of LNG at steady state with 1.22 m thick foam was found to be smaller than the burning rate of a pool fire on water and on concrete without foam application reported by previous researchers. A minimum effective foam depth for pool fire control was found to be 0.61 m through the analysis of foam temperature data during the fire test. For the actual foam application in the LNG industries, additional safety factors should be added to the observed foam depth to account for unexpected loss and collapse/shrinkage of foam bubbles during fire extinguishment. It was observed that expansion foam is capable of reducing the thermal hazard distance up to 52%, as confirmed from heat flux and wind direction analyses. Additionally, it was found that the radiant heat flux has a strong correlation to the flame height due to the effects of the geometric view factor and the surface emissive power. Through the temperature analyses and observations after the pool fire tests, a schematic model of the LNG and

expansion foam with the fire was proposed. It was deduced that foam has three layers: a foam breaking layer, a nonfrozen layer, and a frozen layer. All these layers have different impacts on pool fire suppression. Despite extensive findings in this paper, further research and field tests are required to advance the current understanding of foam mitigation in LNG system. For instance, further tests with foam in different sizes of LNG pool are needed to validate the flame height correlation for the diameter of the fire base with foam application. This paper provides experimental data and initial work for theoretical modeling of LNG-foam systems in fire; thus, one of the future works can be mathematical modeling to determine the mass-burning rate and the effective foam depth. Finally, we recommend that these findings should be associated with the vapor dispersion findings for a comprehensive evaluation of foam effectiveness.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT Authors would like to express gratitude to BP Global Gas SPU for their financial support. Authors would like to thank the Texas Engineering Extension Service (TEEX) for providing supports and guidance on the field tests at the Brayton Fire Training Field. Support from the LNG team members during the test setup is greatly appreciated. ’ REFERENCES (1) Luketa-Hanlin, A. A review of large-scale LNG spills: Experiments and modeling. J. Hazard. Mater. 2006, 132, 119–140. (2) Raj, P. K. LNG fires: A review of experimental results, models and hazard prediction challenges. J. Hazard. Mater. 2007, 140, 444–464. (3) Koopman, R. P.; Ermak, D. L. Lessons learned from LNG safety research. J. Hazard. Mater. 2007, 140, 412–428. (4) Ohba, R.; Kouchi, A.; Hara, T.; Vieillard, V.; Nedelka, D. Validation of heavy and light gas dispersion models for the safety analysis of LNG tank. J. Loss. Prev. Process Ind. 2004, 17, 325–337. (5) Briscoe, F.; Shaw, P. Spread and evaporation of liquid. Prog. Energy Combust. Sci. 1980, 6, 127–140. (6) National Fire Protection Association. NFPA 59A: Standard for the production, storage, and handling of liquefied natural gas (LNG); 2001. (7) Myron, L.; Casada, D. C. N. The current status of LNG facility standards and regulations. Process Saf. Prog. 2005, 24, 152–157. (8) Raj, P. K. Field tests on human tolerance to (LNG) fire radiant heat exposure, and attenuation effects of clothing and other objects. J. Hazard. Mater. 2008, 157, 247–259. (9) Havens, J.; Spicer, T. LNG vapor cloud exclusion zones for spills into impoundments. Process Saf. Prog. 2005, 24, 181–186. (10) University Engineers. I. An experimental study on the mitigation of flammable vapor dispersion and fire hazards immediately following LNG spills on land; 1974. (11) Persson, B.; Lonnermark, A.; Persson, H. FOAMSPEX: Large scale foam application;Modelling of foam spread and extinguishment. Fire Technol. 2003, 39, 347–362. (12) Suardin, J. A.; Wang, Y.; Willson, M.; Mannan, M. S. Field experiments on high expansion (HEX) foam application for controlling LNG pool fire. J. Hazard. Mater. 2009, 165, 612–622. (13) Suardin, J. A. Ph.D. thesis, Texas A&M University, College Station, TX, 2008. 2371



ARTICLE

(14) National Fire Protection Association. NFPA 11A: Standard for low-, medium-, and high-expansion foam; 2002. (15) Raj, P. K. Large LNG fire thermal radiation: Modeling issues and hazard criteria revisited. Process Saf. Prog. 2005, 24, 192–202. (16) Yun, G.; Ng, D.; Mannan, M. S. Key observations of LNG vapor dispersion test with expansion foam application. Ind. Eng. Chem. Res. 2011, in press, dx.doi.org/10.1021/ie100822h. (17) Collins, T. J. ImageJ for microscopy. Biotechniques 2007, 43, 25–30. (18) Babrauskas, V. Estimating large pool fire burning rates. Fire Technol. 1983, 19, 251–261. (19) Rana, M. A.; Guo, Y.; Mannan, M. S. Use of water spray curtain to disperse LNG vapor clouds. J. Loss. Prev. Process Ind. 2009, 23, 77–88. (20) Chatris, J. M.; Quintela, J.; Folch, J.; Planas, E.; Arnaldos, J.; Casal, J. Experimental study of burning rate in hydrocarbon pool fires. Combust. Flame 2001, 126, 1373–1383. (21) Planas-Cuchi, E. i.; Casal, J. Flame temperature distribution in a pool-fire. J. Hazard. Mater. 1998, 62, 231–241. (22) Vesovic, V. The influence of ice formation on vaporization of LNG on water surfaces. J. Hazard. Mater. 2007, 140, 518–526. (23) Raj, P. K. A review of the criteria for people exposure to radiant heat flux from fires. J. Hazard. Mater. 2008, 159, 61–71. (24) Hugues, M.; Phani, K. R. Thermal emission and other characteristics of large liquefied natural gas fires. Process Saf. Prog. 2007, 26, 237–247. (25) Weber, R. O. Analytical models for fire spread due to radiation. Combust. Flame 1989, 78, 398–408. (26) Davis, B. C.; Bagster, D. F. The computation of view factors of fire models: 1. Differential targets. J. Loss. Prev. Process Ind. 1989, 2, 224–234. (27) Atallah, S.; Allan, D. S. Safe separation distances from liquid fuel fires. The Central States Section of the Combustion Institute Meeting, Houston, TX, 1970.

2372


Key Findings of Liquefied Natural Gas Pool Fire Outdoor Tests with

Recommend Documents