- Email: [email protected]
Precursor flame characteristics of flame spread over aviation fuel
Accepted Manuscript Precursor flame characteristics of flame spread over aviation fuel Manhou Li, Changjian Wang, Shenlin Yang, Jiaqing Zhang PII: DOI: Reference:
S1359-4311(16)33753-X http://dx.doi.org/10.1016/j.applthermaleng.2017.02.041 ATE 9922
To appear in:
Applied Thermal Engineering
Received Date: Revised Date: Accepted Date:
29 November 2016 20 January 2017 10 February 2017
Please cite this article as: M. Li, C. Wang, S. Yang, J. Zhang, Precursor flame characteristics of flame spread over aviation fuel, Applied Thermal Engineering (2017), doi: http://dx.doi.org/10.1016/j.applthermaleng.2017.02.041
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Precursor flame characteristics of flame spread over aviation fuel Manhou Li a, b,*, Changjian Wang a, *, Shenlin Yang a, Jiaqing Zhang c a
School of Civil Engineering, Hefei University of Technology, Hefei 230009, China
b
State Key Laboratory of Fire Science, University of Science and Technology of China, Hefei 230026,
China c
State Grid Anhui Electric Power Research Institute, Hefei 230601, China
ABSTRACT Research on flame spreading over liquids is indispensable to make a fire risk assessment of large-sized pool fires in their incipient stages. This precursor flame that is a unique behavior for flame spread over hydrocarbon oils is still not exhaustively understood. A series of tests on flame spread over aviation fuel of RP-5 is well designed and conducted by using a high-speed camera, an infrared camera and several high-sensitive thermocouples. Pulsating performance, spreading velocity, and temperature distribution of this flame are achieved and compared with previous hypotheses. The controlling mechanism of precursor flame is interpreted by coupling effects of gas- and liquid-phase flows in flame spread. The pulsation frequency is qualitatively predicted by Fick’s second law and Raoult partial pressure law. The precursor flame attribute is achieved according to spreading velocity. This precursor flame velocity synchronously illustrates that hydrocarbon fuel spilling is potentially more hazardous than alcohol fuel owing to significant large speed of precursor flame.
*Corresponding author. E-mail address: [email protected] (M. Li); [email protected] (C. Wang). Tel: (86) 551-62901443.
The precursor flame propagation is directly related to liquid surface temperature, hence, spatial temperature distribution near oil surface is revealed. The results of this study will help us understand hydrocarbon spilling fires and possess implications on fire protection design of such flames. Keywords: Precursor flame; flame spread; aviation fuel; initial fuel temperature; fuel pulsation 1. Introduction No. 5 aviation fuel, also called as RP-5 is primarily used as energy sources for turbine engine of carrier-based aircraft. In recent years, fuel leakage in aircraft crash, ship and other transportations frequently occur with the rapid development of Chinese Navy; as a result, fire accidents originated from spilling fuels continue to cause disastrous consequences [1]. Research on flame spreading over liquids seems indispensable to make a fire risk assessment of large-sized pool fires in their incipient stages [2]. Besides, this study possesses scientific significance to appreciate complicated combustion and thermal exchange involving in this issue. Therefore, flame spread over liquid fuel is an important topic in liquid fire dynamics and the related research results have been applied to the prevention and control of oil fires [3]. For alcohol fuels, merely a tall diffusion flame was observed during flame spreading [4-7]. By contrast, observation of flame spread over hydrocarbons verified that a low-height unsteady precursor flame periodically existed ahead of diffusion type flame. The precursor flame phenomenon, also called flash flame has been independently observed by several pioneering researchers [8-13]. Glassman and
co-workers [14] described oscillatory and cyclical characteristics of precursor flame for flame spreading over n-decane, and the authors supposed precursor flame speed as the order of laminar flame velocity. Zhou et al. [9] revealed that the liquid-phase flow preheating non-burning region performed an essential way in periodical appearance of precursor flame. Very recently, Guo et al. [15] discovered that the frequency of precursor flame over oil floating on water decreased with fuel depth increasing, whereas amplitude increased. Schiller and associates [16] supposed that the precursor flame was a lean premixed fuel and air flame in their numerical models. However, in the treatment of the combustion process of flame propagation, Remick [17] assumed the precursor flame as a diffusion burning flame. Among the majority of previous studies, some typical characteristics of precursor flame are still unclear. First, most of previous studies focused on phenomenal description of this flame and less quantitative conclusion i.e., flame speed, oscillation frequency is presented. Second, whether the precursor flame is a premixed flame or a diffusion type flame is still a controversial issue. Third, the temperature evolution near oil surface is very important for precursor flame propagation but still neglected subject. Traditionally, fire risk of spilling accident merely considers igniting possibility imposed by well-developed diffusion flame on nearby unburnt oils. In fact, the precursor flame is more likely to act as fire source due to it locating ahead of diffusion flame. The purposes of current research are to get a better understanding of important issues involved with precursor flame spreading, like pulsating performance, velocity, temperature distribution and controlling mechanisms. In order to find
answers for above-mentioned issues, a group of lab-scale tests and analyses on flame spread over sub-flashpoint RP-5 are conducted. 2. Experimental apparatus The experimental apparatus is depicted in Fig. 1 that has been specified in detailed in Ref. [18]. The environmental conditions as well as technical parameters of RP-5 are displayed in Table 1. The open steel tray size was 100 cm long by 4 cm wide by 10 cm deep. Akita [19] noted that the tray widths did not affect flame pulsating mechanism. Hence, in the present study, the tray of 4 cm wide was used, although the width would be insufficient for obtaining the results independent of the tray width at initial liquid temperatures below flashpoint. The tray walls were made up with two Pyrex widows whose coefficient of thermal conductivity was significantly smaller than steel, so the heat conduction through the walls was greatly reduced. The temperature of the test oils was calibrated by using an electric heating board combing with a thermocouple measurement. For all tests, there was a 1 cm free board height above the oil surface to the tray rim. A baffle barrier was set up at one extreme of the tray and it was made up of a 4 cm-long bottomless cube. In each test, 2 ml heptane was poured into the baffle barrier to establish an ignition pilot flame. The position of the baffle barrier was set to minimize heat flux from the ignition flame, so that the flame spreading should not be altered by the pilot flame. After the heptane fuel burned out, the baffle barrier was removed to allow the flame spread forward. When the flame had spread over the entire oil surface, it was quenched by using a thin fire-retardant plate that could completely isolate the oxygen supply.
Fig. 1. Experimental arrangement for flame spreading over RP-5. Table 1 Environmental testing conditions and technical parameters of RP-5.
Three measurements were available for recording flame spread behaviors. First, the spreading process was registered by a lateral Sony NEX-FS700RH type HD video camera at photographing frequency of 120 frames per second. Second, fuel surface nearby temperature measurements were obtained using six 0.1 mm S-type fine-thermocouples. The horizontal and vertical distances for adjacent measurement points were 10 cm and 5.0 mm respectively. Third, a SAT HY6850 infrared thermal imager with a spectral range of 8 μm to 14 μm was placed approximately 1 m above oil surface to achieve the 2D temperature field of oil surface. The spectral range of IR imager was 8 to 14 μm and the emissivity of RP-5 was about 0.95. 3. Experimental results and discussion 3.1. Precursor flame appearance Fig. 2 is an array of flame images with a full pulsating period of precursor flame for flame spreading over RP-5 at 20 °C. The onset of a pulsation period is defined as the time when no precursor flame is observed ahead of the diffusion flame. The precursor flame extends for 7.0 cm at 83 ms and reaches its maximum value of 15.7 cm ( l ) at 183 ms. Then, it retreats to the head of diffusion flame where it is starting point of precursor propagation. This phenomenon occurs at the end of pulsation period of the precursor flame. The process is repeated several times and the precursor flame
eventually heats the surface temperature to support a relatively stable flame, or diffusion flame [14]. The precursor flame extends more than ten centimeters in a short period. This finding indicates that the fire-protection distance of adjacent oil tanker of hydrocarbon fuel spilling should not be smaller this value because large extended distance may ignite fuels to spilling site.
Fig. 2. Periodic appearance of precursor flame. 3.2. Pulsating behaviors of precursor flame Fig. 3 displays the transient positions of diffusion and precursor flame positions which are obtained by a self-programming procedure with the MATLAB software. The starting points of the horizontal axis in Fig. 3 are selected as the time when the blue precursor flame could be viewed by the camera. The length of precursor flame is achieved to display in Fig. 4 by subtracting the leading distance of diffusion flame from the precursor flame front movement. The solid points in Fig. 4 represent the maximum values of the length of the precursor flame. It can be seen that the precursor flame length is periodic variation with the spreading time in a very fast frequency.
Fig. 3. Leading edge positions of diffusion and precursor type flames versus time.
Fig. 4. Precursor flame length as a function of spreading time. In this research, this maximum value of the precursor flame length ( l ) is labeled as the pulsation wavelength of the precursor flame, whereas the pulsation period of
precursor flame is labeled the spreading time when an entire propagation is completed. The average values of pulsation frequency and wavelength of precursor flame varying with initial temperature are achieved by utilizing an FFT (Fast Fourier Transformation) technique [20], and the experimental results are plotted in Fig. 5. In order to interpret precursor flame oscillations, the lengths of subsurface convection flow from Ref. [18] are added into this figure. Evidently, the pulsation frequency of precursor flame gradually increases with an increase in liquid temperature, whereas the pulsation wavelength decreases.
Fig. 5. Pulsation frequency and wavelength of precursor flame versus initial fuel temperature. The precursor flame wavelength is directly proportional to the liquid-phase flow length as well as the size of gas-phase recirculation cell established in front of precursor flame [21]. The data shown in Fig. 5 verify that the length of subsurface flow declines when the initial temperature of RP-5 increases. Additionally, the size of gas-phase recirculation cell was discovered smaller in a higher initial fuel temperature using a smoke tracking technique [22]. Therefore, both the liquid- and gas-phase measures predict that the precursor flame wavelength should be smaller at a higher initial fuel temperature. After imposing an external heat flux, the liquid temperature increases to vaporization temperature with a period of vaporization time, t . Then, increasing vap
amount of fuel vapor diffuses into air to produce a combustible gaseous mixture close
to oil surface. This period is named as the flammable mixture time (t ). As time mix
elapses, combustion reaction within the flammable mixture reinforces to overcome heat losses, so that the precursor propagation will arise. This period is called as the induction time (tin). The pulsating period of precursor flame time (t) is proposed by the sum of vaporization time, flammable mixture time and induction time, it yields, t tvap tmix tin
(1)
The induction time (t ) as well as flammable mixture time (t ) is significantly in
smaller than t
vap
mix
[12]. Thus, merely the fuel vaporization time is considered. Actually,
the precursor flame advances to a position where the saturated fuel vapor concentration out of quenching layer distance is assumed to be lean flammability limit [23]. The vaporization time (t ) is essentially equivalent to the diffusion time (t ) for vap
dif
fuel vapor concentration satisfying lean flammability limit. When the diffusion distance is very short, the vapor concentration is non-uniform from oil surface (or there is a concentration gradient), and the concentration gradient changes with time, the relationship between vapor concentration and time can be expressed by Fick’s second law [24]:
C f t
D
2C f z 2
(2)
where C f is fuel vapor concentration, D is diffusion coefficient, z is the distance above oil surface, t is diffusion time. Eq. (2) is integrated by one initial condition [25]: t 0,C f 0 , and two boundary conditions: z 0,C f C f 0 and z ,C f 0 . It yields,
C f ( z, t ) C f 0erfc(
z ) 2 Dt
(3)
Therefore, the diffusion time for the fuel vapor concentration meeting lean flammability limit, tq , is predicted by [26],
tq
q2 D(1 C f q / C f 0 )2
(4)
where q is quenching layer distance, about 3~4 mm for hydrocarbon fuels [13], C f q is fuel vapor concentration at quenching layer distance, and C f 0 is initial fuel
vapor concentration. The initial fuel vapor concentrations near oil surface C f 0 under various liquid temperature are predicted by Raoult partial pressure law and Clausius–Clapeyron equation [27],
Cf 0
h 1 1 Psat exp[ vap m ( )] P R Tb T0
(5)
where Psat is saturated vapor pressure in liquid temperature of T0 , Tb is boiling point temperature, P is ambient pressure, and vap hm is latent heat of evaporation which is a constant for a given liquid. Eq. (5) indicates that the saturated vapor pressure Psat should increase as the liquid temperature rises, causing a larger initial fuel vapor concentration. Consequently, more fuel vapors evaporate from oil surface and less time is needed for the fuel/air mixture meeting burning condition, so that the precursor flame oscillates more frequently in a higher initial pool temperature. Moreover, as presented in Fig. 5, the subsurface convection flow length is longer at lower temperature [18], thus larger heat transfer areas are available and more energy will be transferred from the subsurface convection flow to bulk of cold fuels. The
larger heat losses may reduce fuel evaporation rate and further result in a smaller pulsation frequency. 3.3. Mechanisms of precursor flame Fig. 6 demonstrates the schematic diagram of gas- and liquid-phase flows in flame spread. Pulsating spread of precursor flame is related to a gas-phase recirculation cell [28-30]. The formation of the gas cell is competitive result of liquid-phase flow induced vapor motion and gas-phase buoyancy induced opposed natural convection. In the process of flame propagation, the moving flame front produces temperature gradients along the pool surface. The oil temperature below the flame tip is higher, while the fuel surface temperature far away from the flame leading edge is lower. It is well believed that the surface tension coefficient decreases with liquid temperature [31], thus, the warm interfacial liquid under the spot is pulled by the cold, highly tensioned liquid on the surface toward the end of the pool.
Fig. 6. Schematic illustration of gas- and liquid-phase flows in flame spread. This forward moving liquid flow drives gas-phase fuel/air mixture near the oil surface to the same direction of the flame spread by no-slip condition. Further off the pool surface, an opposed buoyant flow moves to the flame front owing to the entrainment of flame to supply oxidizer for supporting fuel combustion and flame spread. This opposite stream combined with the no-slip condition-driven flow products produces a gas-phase recirculation cell outside the quenching distance ahead of flame front [28]. The fuel vapor concentration inside the gas cell meets the lean
flammability limit at the time when the oil surface temperature reaches the oil’s flashpoint. Then, a blue precursor flame close to the oil surface starts to rapidly proceed ahead of the well-developed yellow diffusion flame. The precursor flame jumps through the mixture at a high rate of speed and then it extinguishes due to both a lack of flammable mixtures induced by the hot gas expansion and quenching by the liquid surface. Meanwhile, the mixture inside the gas-phase recirculation cell is no longer in the flammability limit and it is destroyed by the hot gas expansion that is formed ahead of the flame tip [32]. Therefore, the hot gas expansion in front of flame tip plays a significant role in the precursor flame oscillation by periodically destroying the gas cell. 3.4. Velocity of precursor flame The velocities of precursor flame and diffusion flame spreading [18] are displayed in Fig. 7. Apparently, the diffusion flame speed increases nearly exponentially with temperature, whereas the velocity of the precursor flame remains stable at ~85 cm/s. The diffusion flame speed is significantly smaller than the counterpart of precursor spreading. This indicates that the hydrocarbon fuel spilling accidents are potentially more hazardous than those of alcohol fuels owing to the significant large spreading rate of precursor flame. The laminar burning velocity is measured for a range of hydrocarbon fuels and it is found as approximately 40 cm/s at the stoichiometric concentration [33]. Thus, the measured velocity of precursor flame is in contradiction with Glassman’s hypothesis who supposed the precursor flame as a premixed laminar combustion flame [14]. Moreover, the maximum value of diffusion combustion flame
velocity should not be greater than 10 cm/s [13], thus, the precursor flame is a premixed type flame in essence.
Fig. 7. Velocity of flame spread versus initial temperature of RP-5. According to our previous measures [18], the velocity of diffusion flame within lean flammability limit is approximately 7.53 cm/s. This suggests a distinct phenomenology between diffusion flame and precursor flame. Although we are not sure the intrinsic reasons for such occurrence, some probable interpretations are presented. First, the diffusion flame always exists during the flame spreading process, while the precursor flame is periodic existence and disappearance. More energy is transferred to support the continuous propagation of diffusion flame, causing a relative slow flame spread rate. Second, the diffusion flame possesses taller flame height and higher temperature than the precursor flame, thus the radiant heat losses are larger for the diffusing combustion flame. 3.5. Temperature profile of oil surface Fig. 8 shows two infrared images for flame spreading over RP-5 under pool temperature of 30 °C. Fig. 8 (a) is in the initial stage of a pulsation period and Fig. 8 (b) is 80 ms after the onset of this pulsation period. At beginning, the preheating area length is largest (about 16.4 cm). As time elapses, the oil surface temperature as well as the gas mixture concentration increases due to the coupling preheating effects by warm subsurface convection flow and flame radiation. As illustrated in Fig. 8 (b), when the fuel vapor concentration reaches the lean flammability limit of RP-5, the
precursor flame “jumps” 6.1 cm forward. This advancement is potentially equivalent to the length of precursor flame at this time.
Fig. 8. Infrared images of oil surface for flame spreading over RP-5.
Fig. 9. Surface temperatures for flame spread over oil surface. Fig. 9 is the oil surface oil surface temperature distributions, corresponding to the occurrence times of the IR images in Fig. 8. The whole oil surface can be separated into four regions at t + 80 ms: a diffusion flame area, a precursor flame area, a preheating area and the bulk of cold oil area. The oil surface temperatures below the precursor flame are not very accurate owing to the emissivity difference between the flame contour and the liquid fuel. The oil surface temperature gradually increases with the distance close to flame area due to the preheating effects by both liquid-phase thermal vortex and external energy import by flame radiation. This oil surface temperature reaches 66 °C at the precursor flame tip and 95 °C at the diffusion flame tip, which are supposed as the flashpoint and fire point temperature of the RP-5, respectively. The fire point is the temperature at which the production rate of vapor is enough to sustain a steady flame and it is typically higher than the flashpoint for hydrocarbon [34]. The oil surface temperatures both at diffusion flame and precursor flame fronts are measured as a function of initial temperature of RP-5. Fig. 10 identifies that the diffusion flame proceeds to a point where the liquid temperature is equivalent to fire point. By contrast, the precursor flame locates at a position where
the oil surface temperature corresponds to flashpoint temperature [14].
Fig. 10. Oil surface temperatures under flame front.
Fig. 11. Details of the thermocouple readings. A representative thermocouple temperature measurement is presented in Fig. 11. The x-coordinate indicates the horizontal distance from the first thermocouple, while the y-coordinate identifies the vertical distance from oil surface. The evolutions of TC-1 reflect a higher temperature as the convective flow front arriving (21 s). When the flame’s leading edge approaches the measuring point, the sudden peak in recorded temperatures by gas-phase thermocouples indicates this event. Then, the readings of oil surface thermocouples fluctuate seriously after the diffusion flame tip passes over the detection point of this thermocouple. The oil surface temperature is close to or above the fire point temperature, but lower than the boiling point of RP-5 (181 °C), indicating that the oils under the diffusion flame root are not heated to boil during the whole flame spread test. Moreover, the interval time between two gas-phase thermocouples i.e., TC-2 and TC-4 is 6 s, so that the flame speed equals to ~1.67 cm/s. As mentioned above, the velocity of precursor flame is ~85 cm/s, thus the gas-phase thermocouple measured flame spread rate should be corresponding to diffusion flame type, not precursor flame. The thermocouples placed in gas phase are unaffected by the precursor flames and the second sudden temperature rise corresponds to arrival of diffusion flame tip.
4. Conclusions In order to get a better understanding of the important issues involved with precursor flame spreading, a great group of lab-scale tests and analyses on flame spread over sub-flashpoint RP-5 are conducted. Main conclusions include: (1) The pulsation frequency of precursor flame steadily increases with fuel temperature, whereas the pulsation wavelength diminishes. (2) The oscillation of precursor flame is attributed to periodical existence and disappearance of gas-phase recirculation cell. (3) The velocity of precursor flame is ~85 cm/s under different initial fuel temperatures, which is significantly larger than the diffusion flame spread rate in low flammability limit. On the basis of the flame speed, the precursor flame is a premixed combustion flame in essence. (4) The diffusion flame proceeds to a point where the liquid temperature is equivalent to fire point. By contrast, the precursor flame locates at a position where the oil surface temperature corresponds to flashpoint temperature. Acknowledgement The authors would like to thank the Opening Fund of State Key Laboratory of Fire Science (No. HZ2016-KF09), and the Fundamental Research Funds for the Central Universities (No. JZ2015HGBZ0502) for their supports. Reference [1] C.G. Fan, J.Q. Zhang, K.J. Zhu, K.Y. Li, An experimental study of temperature and heat flux in a channel with an asymmetric thermal plume, Applied Thermal
Engineering,
113
(2017)
1128-1136.
http://dx.doi.org/10.1016/j.applthermaleng.2016.11.017. [2] T. Hirano, Flame Spread Across Condensed Combustibles, in: W.A. Sirignano, A.G. Merzhanov, L.D. Luca (eds.) Advances in Combustion Science: In Honor of Ya. B. Zel'dovich, American Institute of Aeronautics and Astronautics, 1997, pp. 131-144. http://dx.doi.org/10.2514/4.866456. [3] F. Tang, L. Hu, Z. Qiu, Q. Wang, A global model of plume axial temperature profile transition from axisymmetric to line-source pool fires in normal and reduced
pressures,
Fuel,
130
(2014)
211-214.
http://dx.doi.org/10.1016/j.fuel.2014.04.053. [4] H.D. Ross, F.J. Miller, Understanding flame spread across alcohol pools, in: Proceedings of the sixth international symposium, International Association for Fire
Safety
Science,
Poitiers,
France,
2000,
pp.
77-94.
http://dx.doi.org/10.3801/IAFSS.FSS.6-77. [5] E. Degroote, Control parameters of flame spreading in a fuel container, Journal of Thermal
Analysis
and
Calorimetry,
87
(2007)
149-151.
http://dx.doi.org/10.1007/s10973-006-7838-1. [6] E. Degroote, P.L. Garcia-Ybarra, Flame propagation over liquid alcohols. Part III. Pulsating regime, Journal of Thermal Analysis and Calorimetry, 80 (2005) 555-558. http://dx.doi.org/10.1007/s10973-005-0693-7. [7] E. Degroote, P.L. Garcia-Ybarra, Flame propagation over liquid alcohols. Part II. Steady propagation regimes, Journal of Thermal Analysis and Calorimetry, 80
(2005) 549-553. http://dx.doi.org/10.1007/s10973-005-0692-8. [8] D. White, C.L. Beyler, C. Fulper, J. Leonard, Flame spread on aviation fuels, Fire Safety
Journal,
28
(1997)
1-31.
http://dx.doi.org/10.1016/s0379-7112(96)00070-7. [9] J. Zhou, G. Chen, P. Li, B. Chen, C. Wang, S. Lu, Analysis of flame spread over aviation
kerosene,
Chinese
Science
Bulletin,
55
(2010)
1822-1827.
http://dx.doi.org/10.1007/s11434-010-3014-x. [10] R. Mackinven, J.G. Hansel, I. Glassman, Influence of laboratory parameters on flame spread across liquid fuels, Combustion Science and Technology, 1 (1970) 293-306. http://dx.doi.org/10.1080/00102206908952209. [11] J.P. Burelbach, M. Epstein, M.G. Plys, Initiation of flame spreading on shallow subflash
fuel
layers,
Combustion
and
Flame,
114
(1998)
280-282.
http://dx.doi.org/10.1016/s0010-2180(97)00290-3. [12] N. Wu, G. Kolb, J.L. Torero, The effect of weathering on the flammability of a slick of crude oil on a water bed, Combustion Science and Technology, 161 (2000) 269-308. http://dx.doi.org/10.1080/00102200008935820. [13] H.D. Ross, Ignition of and flame spread over laboratory-scale pools of pure liquid fuels, Progress in Energy and Combustion Science, 20 (1994) 17-63. http://dx.doi.org/10.1016/0360-1285(94)90005-1. [14] I. Glassman, F.L. Dryer, Flame spreading across liquid fuels, Fire Safety Journal, 3 (1981) 123-138. http://dx.doi.org/10.1016/0379-7112(81)90038-2. [15] J. Guo, S. Lu, C. Wang, Effect of fuel depth on flame spread over aviation
kerosene,
Journal
of
Fire
Sciences,
32
(2014)
43-51.
http://dx.doi.org/10.1177/0734904113491705. [16] D.N. Schiller, H.D. Ross, W.A. Sirignano, Computational analysis of flame spread across alcohol pools, Combustion Science and Technology, 118 (1996) 203-255. http://dx.doi.org/10.1080/00102209608951980. [17] E.M. Remick, Flame spread over liquid fuels, Ph.D. thesis, Cornell University, New York, 1981. [18] M. Li, S. Lu, J. Guo, R. Chen, K.-L. Tsui, Initial Fuel Temperature Effects on Flame Spread over Aviation Kerosene in Low- and High-Altitude Environments, Fire
Technology,
51
(2015)
707-721.
http://dx.doi.org/10.1007/s10694-014-0395-4. [19] K. Akita, Some problems of flame spread along a liquid surface, Proceedings of the
Combustion
Institute,
14
(1973)
1075-1083.
http://dx.doi.org/10.1016/s0082-0784(73)80097-9. [20] F. Tang, L. Hu, Q. Wang, Z. Ding, Flame pulsation frequency of conduction-controlled rectangular hydrocarbon pool fires of different aspect ratios in a sub-atmospheric pressure, International Journal of Heat and Mass Transfer,
76
(2014)
447-451.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2014.04.047. [21] D.N. Schiller, W.A. Sirignano, Opposed-flow flame spread across n-propanol pools, Proceedings of the Combustion Institute, 26 (1996) 1319-1325. http://dx.doi.org/10.1016/s0082-0784(96)80350-x.
[22] D.N. Schiller, H.D. Ross, W.A. Sirignano, Computational Predictions of Flame Spread over Alcohol Pools, in: 31st Aerospace Sciences Meeting & Exhibit, American Institute of Aeronautics and Astronautics, Reno, NV, 1993. http://dx.doi.org/10.2514/6.1993-825. [23] V. Babrauskas, Ignition handbook, Fire Science Publishers, Issaquah WA, USA. Co-published by the Society of Fire Protection Engineers, 2003. [24] J. Crank, The Mathematics of Diffusion, 2ed ed., Clarendon Press, London, 1975. [25] T. Konishi, Flame spread over liquid fuels, Ph.D. thesis, Oita University, Oita, Japan, 1998. [26] M. Li, S. Lu, J. Guo, R. Chen, K.-L. Tsui, Flame spread over n-butanol at sub-flash temperature in normal and elevated altitude environments, Journal of Thermal
Analysis
and
Calorimetry,
119
(2015)
401-409.
http://dx.doi.org/10.1007/s10973-014-4190-8. [27] A. Chakraborty, Thermodynamic trends for the adsorption of non polar gases on activated carbons employing a new adsorption isotherm modelling, Applied Thermal
Engineering,
105
(2016)
189-197.
http://dx.doi.org/10.1016/j.applthermaleng.2016.05.160. [28] F.J. Miller, H.D. Ross, Smoke visualization of the gas-phase flow during flame spread across a liquid pool, Proceedings of the Combustion Institute, 27 (1998) 2715-2722. http://dx.doi.org/10.1016/S0082-0784(98)80127-6. [29] J. Ji, S. Lin, C. Zhao, K. Li, Z. Gao, Experimental study on initial temperature influence on flame spread characteristics of diesel and gasoline–diesel blends,
Fuel, 178 (2016) 283-289. http://dx.doi.org/10.1016/j.fuel.2016.03.072. [30] M. Li, S. Lu, R. Chen, C. Wang, Pulsating behaviors of flame spread across n-butanol fuel surface, Applied Thermal Engineering, 112 (2017) 1445-1451. http://dx.doi.org/10.1016/j.applthermaleng.2016.10.001. [31] M. Mahdavi, S. Qiu, S. Tiari, Numerical investigation of hydrodynamics and thermal performance of a specially configured heat pipe for high-temperature thermal energy storage systems, Applied Thermal Engineering, 81 (2015) 325-337. http://dx.doi.org/10.1016/j.applthermaleng.2015.02.031. [32] B.M.C. Brantley, Quantitative analysis of a turbulent wind tunnel with obstructions for use in liquid flame spread experiments, M.S. thesis, Air University, Ohio, 2006. [33] G.N. Badami, A. Egerton, The Determination of Burning Velocities of Slow Flames, Proceedings Of The Royal Society Of London Series A-Mathematical And Physical Sciences, 228 (1955) 297-322. http://dx.doi.org/10.2307/99660. [34] Standard Test Method for Flash and Fire Points by Cleveland Open Cup Tester, ASTM D92 - 12b. 1-10. http://dx.doi.org/10.1520/D0092-12B.
Figure Captions Fig. 1. Experimental arrangement for flame spreading over RP-5. Fig. 2. Periodic appearance of precursor flame. Fig. 3. Leading edge positions of diffusion and precursor type flames versus time. Fig. 4. Precursor flame length as a function of spreading time. Fig. 5. Pulsation frequency and wavelength of precursor flame versus initial fuel temperature. Fig. 6. Schematic illustration of gas- and liquid-phase flows in flame spread. Fig. 7. Velocity of flame spread versus initial temperature of RP-5. Fig. 8. Infrared images of oil surface for flame spreading over RP-5. Fig. 9. Surface temperatures for flame spread over oil surface. Fig. 10. Oil surface temperatures under flame front. Fig. 11. Details of the thermocouple readings.
Fig. 1. Experimental arrangement for flame spreading over RP-5.
Fig. 2. Periodic appearance of precursor flame.
Fig. 3. Leading edge positions of diffusion and precursor type flames versus time.
Fig. 4. Precursor flame length as a function of spreading time.
Fig. 5. Pulsation frequency and wavelength of precursor flame versus initial fuel temperature.
Fig. 6. Schematic illustration of gas- and liquid-phase flows in flame spread.
Fig. 7. Velocity of flame spread versus initial temperature of RP-5.
Fig. 8. Infrared images of oil surface for flame spreading over RP-5.
Fig. 9. Surface temperatures for flame spread over oil surface.
Fig. 10. Oil surface temperatures under flame front.
Fig. 11. Details of the thermocouple readings.
Table Caption Table 1 Environmental testing conditions and technical parameters of RP-5. Experimental parameters
Measurement unit
Data
Environment temperature
o
C
13-20
Ambient humidity
%
36-39
Testing oil temperature
o
16.5-78.8
Equivalent molecular formula
-
C11.9H22.2
Latent heat of evaporation
kJ/kg
383
Boiling point
o
181
Fire point
o
95
Flashpoint
o
66
Dynamic viscosity (20 oC)
mPa·s
3.02
Surface tension (20 oC)
mN/m
25.8
Density (20 oC)
g/cm3
0.820
C
C C C
S1359-4311(16)33753-X http://dx.doi.org/10.1016/j.applthermaleng.2017.02.041 ATE 9922
To appear in:
Applied Thermal Engineering
Received Date: Revised Date: Accepted Date:
29 November 2016 20 January 2017 10 February 2017
Please cite this article as: M. Li, C. Wang, S. Yang, J. Zhang, Precursor flame characteristics of flame spread over aviation fuel, Applied Thermal Engineering (2017), doi: http://dx.doi.org/10.1016/j.applthermaleng.2017.02.041
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Precursor flame characteristics of flame spread over aviation fuel Manhou Li a, b,*, Changjian Wang a, *, Shenlin Yang a, Jiaqing Zhang c a
School of Civil Engineering, Hefei University of Technology, Hefei 230009, China
b
State Key Laboratory of Fire Science, University of Science and Technology of China, Hefei 230026,
China c
State Grid Anhui Electric Power Research Institute, Hefei 230601, China
ABSTRACT Research on flame spreading over liquids is indispensable to make a fire risk assessment of large-sized pool fires in their incipient stages. This precursor flame that is a unique behavior for flame spread over hydrocarbon oils is still not exhaustively understood. A series of tests on flame spread over aviation fuel of RP-5 is well designed and conducted by using a high-speed camera, an infrared camera and several high-sensitive thermocouples. Pulsating performance, spreading velocity, and temperature distribution of this flame are achieved and compared with previous hypotheses. The controlling mechanism of precursor flame is interpreted by coupling effects of gas- and liquid-phase flows in flame spread. The pulsation frequency is qualitatively predicted by Fick’s second law and Raoult partial pressure law. The precursor flame attribute is achieved according to spreading velocity. This precursor flame velocity synchronously illustrates that hydrocarbon fuel spilling is potentially more hazardous than alcohol fuel owing to significant large speed of precursor flame.
*Corresponding author. E-mail address: [email protected] (M. Li); [email protected] (C. Wang). Tel: (86) 551-62901443.
The precursor flame propagation is directly related to liquid surface temperature, hence, spatial temperature distribution near oil surface is revealed. The results of this study will help us understand hydrocarbon spilling fires and possess implications on fire protection design of such flames. Keywords: Precursor flame; flame spread; aviation fuel; initial fuel temperature; fuel pulsation 1. Introduction No. 5 aviation fuel, also called as RP-5 is primarily used as energy sources for turbine engine of carrier-based aircraft. In recent years, fuel leakage in aircraft crash, ship and other transportations frequently occur with the rapid development of Chinese Navy; as a result, fire accidents originated from spilling fuels continue to cause disastrous consequences [1]. Research on flame spreading over liquids seems indispensable to make a fire risk assessment of large-sized pool fires in their incipient stages [2]. Besides, this study possesses scientific significance to appreciate complicated combustion and thermal exchange involving in this issue. Therefore, flame spread over liquid fuel is an important topic in liquid fire dynamics and the related research results have been applied to the prevention and control of oil fires [3]. For alcohol fuels, merely a tall diffusion flame was observed during flame spreading [4-7]. By contrast, observation of flame spread over hydrocarbons verified that a low-height unsteady precursor flame periodically existed ahead of diffusion type flame. The precursor flame phenomenon, also called flash flame has been independently observed by several pioneering researchers [8-13]. Glassman and
co-workers [14] described oscillatory and cyclical characteristics of precursor flame for flame spreading over n-decane, and the authors supposed precursor flame speed as the order of laminar flame velocity. Zhou et al. [9] revealed that the liquid-phase flow preheating non-burning region performed an essential way in periodical appearance of precursor flame. Very recently, Guo et al. [15] discovered that the frequency of precursor flame over oil floating on water decreased with fuel depth increasing, whereas amplitude increased. Schiller and associates [16] supposed that the precursor flame was a lean premixed fuel and air flame in their numerical models. However, in the treatment of the combustion process of flame propagation, Remick [17] assumed the precursor flame as a diffusion burning flame. Among the majority of previous studies, some typical characteristics of precursor flame are still unclear. First, most of previous studies focused on phenomenal description of this flame and less quantitative conclusion i.e., flame speed, oscillation frequency is presented. Second, whether the precursor flame is a premixed flame or a diffusion type flame is still a controversial issue. Third, the temperature evolution near oil surface is very important for precursor flame propagation but still neglected subject. Traditionally, fire risk of spilling accident merely considers igniting possibility imposed by well-developed diffusion flame on nearby unburnt oils. In fact, the precursor flame is more likely to act as fire source due to it locating ahead of diffusion flame. The purposes of current research are to get a better understanding of important issues involved with precursor flame spreading, like pulsating performance, velocity, temperature distribution and controlling mechanisms. In order to find
answers for above-mentioned issues, a group of lab-scale tests and analyses on flame spread over sub-flashpoint RP-5 are conducted. 2. Experimental apparatus The experimental apparatus is depicted in Fig. 1 that has been specified in detailed in Ref. [18]. The environmental conditions as well as technical parameters of RP-5 are displayed in Table 1. The open steel tray size was 100 cm long by 4 cm wide by 10 cm deep. Akita [19] noted that the tray widths did not affect flame pulsating mechanism. Hence, in the present study, the tray of 4 cm wide was used, although the width would be insufficient for obtaining the results independent of the tray width at initial liquid temperatures below flashpoint. The tray walls were made up with two Pyrex widows whose coefficient of thermal conductivity was significantly smaller than steel, so the heat conduction through the walls was greatly reduced. The temperature of the test oils was calibrated by using an electric heating board combing with a thermocouple measurement. For all tests, there was a 1 cm free board height above the oil surface to the tray rim. A baffle barrier was set up at one extreme of the tray and it was made up of a 4 cm-long bottomless cube. In each test, 2 ml heptane was poured into the baffle barrier to establish an ignition pilot flame. The position of the baffle barrier was set to minimize heat flux from the ignition flame, so that the flame spreading should not be altered by the pilot flame. After the heptane fuel burned out, the baffle barrier was removed to allow the flame spread forward. When the flame had spread over the entire oil surface, it was quenched by using a thin fire-retardant plate that could completely isolate the oxygen supply.
Fig. 1. Experimental arrangement for flame spreading over RP-5. Table 1 Environmental testing conditions and technical parameters of RP-5.
Three measurements were available for recording flame spread behaviors. First, the spreading process was registered by a lateral Sony NEX-FS700RH type HD video camera at photographing frequency of 120 frames per second. Second, fuel surface nearby temperature measurements were obtained using six 0.1 mm S-type fine-thermocouples. The horizontal and vertical distances for adjacent measurement points were 10 cm and 5.0 mm respectively. Third, a SAT HY6850 infrared thermal imager with a spectral range of 8 μm to 14 μm was placed approximately 1 m above oil surface to achieve the 2D temperature field of oil surface. The spectral range of IR imager was 8 to 14 μm and the emissivity of RP-5 was about 0.95. 3. Experimental results and discussion 3.1. Precursor flame appearance Fig. 2 is an array of flame images with a full pulsating period of precursor flame for flame spreading over RP-5 at 20 °C. The onset of a pulsation period is defined as the time when no precursor flame is observed ahead of the diffusion flame. The precursor flame extends for 7.0 cm at 83 ms and reaches its maximum value of 15.7 cm ( l ) at 183 ms. Then, it retreats to the head of diffusion flame where it is starting point of precursor propagation. This phenomenon occurs at the end of pulsation period of the precursor flame. The process is repeated several times and the precursor flame
eventually heats the surface temperature to support a relatively stable flame, or diffusion flame [14]. The precursor flame extends more than ten centimeters in a short period. This finding indicates that the fire-protection distance of adjacent oil tanker of hydrocarbon fuel spilling should not be smaller this value because large extended distance may ignite fuels to spilling site.
Fig. 2. Periodic appearance of precursor flame. 3.2. Pulsating behaviors of precursor flame Fig. 3 displays the transient positions of diffusion and precursor flame positions which are obtained by a self-programming procedure with the MATLAB software. The starting points of the horizontal axis in Fig. 3 are selected as the time when the blue precursor flame could be viewed by the camera. The length of precursor flame is achieved to display in Fig. 4 by subtracting the leading distance of diffusion flame from the precursor flame front movement. The solid points in Fig. 4 represent the maximum values of the length of the precursor flame. It can be seen that the precursor flame length is periodic variation with the spreading time in a very fast frequency.
Fig. 3. Leading edge positions of diffusion and precursor type flames versus time.
Fig. 4. Precursor flame length as a function of spreading time. In this research, this maximum value of the precursor flame length ( l ) is labeled as the pulsation wavelength of the precursor flame, whereas the pulsation period of
precursor flame is labeled the spreading time when an entire propagation is completed. The average values of pulsation frequency and wavelength of precursor flame varying with initial temperature are achieved by utilizing an FFT (Fast Fourier Transformation) technique [20], and the experimental results are plotted in Fig. 5. In order to interpret precursor flame oscillations, the lengths of subsurface convection flow from Ref. [18] are added into this figure. Evidently, the pulsation frequency of precursor flame gradually increases with an increase in liquid temperature, whereas the pulsation wavelength decreases.
Fig. 5. Pulsation frequency and wavelength of precursor flame versus initial fuel temperature. The precursor flame wavelength is directly proportional to the liquid-phase flow length as well as the size of gas-phase recirculation cell established in front of precursor flame [21]. The data shown in Fig. 5 verify that the length of subsurface flow declines when the initial temperature of RP-5 increases. Additionally, the size of gas-phase recirculation cell was discovered smaller in a higher initial fuel temperature using a smoke tracking technique [22]. Therefore, both the liquid- and gas-phase measures predict that the precursor flame wavelength should be smaller at a higher initial fuel temperature. After imposing an external heat flux, the liquid temperature increases to vaporization temperature with a period of vaporization time, t . Then, increasing vap
amount of fuel vapor diffuses into air to produce a combustible gaseous mixture close
to oil surface. This period is named as the flammable mixture time (t ). As time mix
elapses, combustion reaction within the flammable mixture reinforces to overcome heat losses, so that the precursor propagation will arise. This period is called as the induction time (tin). The pulsating period of precursor flame time (t) is proposed by the sum of vaporization time, flammable mixture time and induction time, it yields, t tvap tmix tin
(1)
The induction time (t ) as well as flammable mixture time (t ) is significantly in
smaller than t
vap
mix
[12]. Thus, merely the fuel vaporization time is considered. Actually,
the precursor flame advances to a position where the saturated fuel vapor concentration out of quenching layer distance is assumed to be lean flammability limit [23]. The vaporization time (t ) is essentially equivalent to the diffusion time (t ) for vap
dif
fuel vapor concentration satisfying lean flammability limit. When the diffusion distance is very short, the vapor concentration is non-uniform from oil surface (or there is a concentration gradient), and the concentration gradient changes with time, the relationship between vapor concentration and time can be expressed by Fick’s second law [24]:
C f t
D
2C f z 2
(2)
where C f is fuel vapor concentration, D is diffusion coefficient, z is the distance above oil surface, t is diffusion time. Eq. (2) is integrated by one initial condition [25]: t 0,C f 0 , and two boundary conditions: z 0,C f C f 0 and z ,C f 0 . It yields,
C f ( z, t ) C f 0erfc(
z ) 2 Dt
(3)
Therefore, the diffusion time for the fuel vapor concentration meeting lean flammability limit, tq , is predicted by [26],
tq
q2 D(1 C f q / C f 0 )2
(4)
where q is quenching layer distance, about 3~4 mm for hydrocarbon fuels [13], C f q is fuel vapor concentration at quenching layer distance, and C f 0 is initial fuel
vapor concentration. The initial fuel vapor concentrations near oil surface C f 0 under various liquid temperature are predicted by Raoult partial pressure law and Clausius–Clapeyron equation [27],
Cf 0
h 1 1 Psat exp[ vap m ( )] P R Tb T0
(5)
where Psat is saturated vapor pressure in liquid temperature of T0 , Tb is boiling point temperature, P is ambient pressure, and vap hm is latent heat of evaporation which is a constant for a given liquid. Eq. (5) indicates that the saturated vapor pressure Psat should increase as the liquid temperature rises, causing a larger initial fuel vapor concentration. Consequently, more fuel vapors evaporate from oil surface and less time is needed for the fuel/air mixture meeting burning condition, so that the precursor flame oscillates more frequently in a higher initial pool temperature. Moreover, as presented in Fig. 5, the subsurface convection flow length is longer at lower temperature [18], thus larger heat transfer areas are available and more energy will be transferred from the subsurface convection flow to bulk of cold fuels. The
larger heat losses may reduce fuel evaporation rate and further result in a smaller pulsation frequency. 3.3. Mechanisms of precursor flame Fig. 6 demonstrates the schematic diagram of gas- and liquid-phase flows in flame spread. Pulsating spread of precursor flame is related to a gas-phase recirculation cell [28-30]. The formation of the gas cell is competitive result of liquid-phase flow induced vapor motion and gas-phase buoyancy induced opposed natural convection. In the process of flame propagation, the moving flame front produces temperature gradients along the pool surface. The oil temperature below the flame tip is higher, while the fuel surface temperature far away from the flame leading edge is lower. It is well believed that the surface tension coefficient decreases with liquid temperature [31], thus, the warm interfacial liquid under the spot is pulled by the cold, highly tensioned liquid on the surface toward the end of the pool.
Fig. 6. Schematic illustration of gas- and liquid-phase flows in flame spread. This forward moving liquid flow drives gas-phase fuel/air mixture near the oil surface to the same direction of the flame spread by no-slip condition. Further off the pool surface, an opposed buoyant flow moves to the flame front owing to the entrainment of flame to supply oxidizer for supporting fuel combustion and flame spread. This opposite stream combined with the no-slip condition-driven flow products produces a gas-phase recirculation cell outside the quenching distance ahead of flame front [28]. The fuel vapor concentration inside the gas cell meets the lean
flammability limit at the time when the oil surface temperature reaches the oil’s flashpoint. Then, a blue precursor flame close to the oil surface starts to rapidly proceed ahead of the well-developed yellow diffusion flame. The precursor flame jumps through the mixture at a high rate of speed and then it extinguishes due to both a lack of flammable mixtures induced by the hot gas expansion and quenching by the liquid surface. Meanwhile, the mixture inside the gas-phase recirculation cell is no longer in the flammability limit and it is destroyed by the hot gas expansion that is formed ahead of the flame tip [32]. Therefore, the hot gas expansion in front of flame tip plays a significant role in the precursor flame oscillation by periodically destroying the gas cell. 3.4. Velocity of precursor flame The velocities of precursor flame and diffusion flame spreading [18] are displayed in Fig. 7. Apparently, the diffusion flame speed increases nearly exponentially with temperature, whereas the velocity of the precursor flame remains stable at ~85 cm/s. The diffusion flame speed is significantly smaller than the counterpart of precursor spreading. This indicates that the hydrocarbon fuel spilling accidents are potentially more hazardous than those of alcohol fuels owing to the significant large spreading rate of precursor flame. The laminar burning velocity is measured for a range of hydrocarbon fuels and it is found as approximately 40 cm/s at the stoichiometric concentration [33]. Thus, the measured velocity of precursor flame is in contradiction with Glassman’s hypothesis who supposed the precursor flame as a premixed laminar combustion flame [14]. Moreover, the maximum value of diffusion combustion flame
velocity should not be greater than 10 cm/s [13], thus, the precursor flame is a premixed type flame in essence.
Fig. 7. Velocity of flame spread versus initial temperature of RP-5. According to our previous measures [18], the velocity of diffusion flame within lean flammability limit is approximately 7.53 cm/s. This suggests a distinct phenomenology between diffusion flame and precursor flame. Although we are not sure the intrinsic reasons for such occurrence, some probable interpretations are presented. First, the diffusion flame always exists during the flame spreading process, while the precursor flame is periodic existence and disappearance. More energy is transferred to support the continuous propagation of diffusion flame, causing a relative slow flame spread rate. Second, the diffusion flame possesses taller flame height and higher temperature than the precursor flame, thus the radiant heat losses are larger for the diffusing combustion flame. 3.5. Temperature profile of oil surface Fig. 8 shows two infrared images for flame spreading over RP-5 under pool temperature of 30 °C. Fig. 8 (a) is in the initial stage of a pulsation period and Fig. 8 (b) is 80 ms after the onset of this pulsation period. At beginning, the preheating area length is largest (about 16.4 cm). As time elapses, the oil surface temperature as well as the gas mixture concentration increases due to the coupling preheating effects by warm subsurface convection flow and flame radiation. As illustrated in Fig. 8 (b), when the fuel vapor concentration reaches the lean flammability limit of RP-5, the
precursor flame “jumps” 6.1 cm forward. This advancement is potentially equivalent to the length of precursor flame at this time.
Fig. 8. Infrared images of oil surface for flame spreading over RP-5.
Fig. 9. Surface temperatures for flame spread over oil surface. Fig. 9 is the oil surface oil surface temperature distributions, corresponding to the occurrence times of the IR images in Fig. 8. The whole oil surface can be separated into four regions at t + 80 ms: a diffusion flame area, a precursor flame area, a preheating area and the bulk of cold oil area. The oil surface temperatures below the precursor flame are not very accurate owing to the emissivity difference between the flame contour and the liquid fuel. The oil surface temperature gradually increases with the distance close to flame area due to the preheating effects by both liquid-phase thermal vortex and external energy import by flame radiation. This oil surface temperature reaches 66 °C at the precursor flame tip and 95 °C at the diffusion flame tip, which are supposed as the flashpoint and fire point temperature of the RP-5, respectively. The fire point is the temperature at which the production rate of vapor is enough to sustain a steady flame and it is typically higher than the flashpoint for hydrocarbon [34]. The oil surface temperatures both at diffusion flame and precursor flame fronts are measured as a function of initial temperature of RP-5. Fig. 10 identifies that the diffusion flame proceeds to a point where the liquid temperature is equivalent to fire point. By contrast, the precursor flame locates at a position where
the oil surface temperature corresponds to flashpoint temperature [14].
Fig. 10. Oil surface temperatures under flame front.
Fig. 11. Details of the thermocouple readings. A representative thermocouple temperature measurement is presented in Fig. 11. The x-coordinate indicates the horizontal distance from the first thermocouple, while the y-coordinate identifies the vertical distance from oil surface. The evolutions of TC-1 reflect a higher temperature as the convective flow front arriving (21 s). When the flame’s leading edge approaches the measuring point, the sudden peak in recorded temperatures by gas-phase thermocouples indicates this event. Then, the readings of oil surface thermocouples fluctuate seriously after the diffusion flame tip passes over the detection point of this thermocouple. The oil surface temperature is close to or above the fire point temperature, but lower than the boiling point of RP-5 (181 °C), indicating that the oils under the diffusion flame root are not heated to boil during the whole flame spread test. Moreover, the interval time between two gas-phase thermocouples i.e., TC-2 and TC-4 is 6 s, so that the flame speed equals to ~1.67 cm/s. As mentioned above, the velocity of precursor flame is ~85 cm/s, thus the gas-phase thermocouple measured flame spread rate should be corresponding to diffusion flame type, not precursor flame. The thermocouples placed in gas phase are unaffected by the precursor flames and the second sudden temperature rise corresponds to arrival of diffusion flame tip.
4. Conclusions In order to get a better understanding of the important issues involved with precursor flame spreading, a great group of lab-scale tests and analyses on flame spread over sub-flashpoint RP-5 are conducted. Main conclusions include: (1) The pulsation frequency of precursor flame steadily increases with fuel temperature, whereas the pulsation wavelength diminishes. (2) The oscillation of precursor flame is attributed to periodical existence and disappearance of gas-phase recirculation cell. (3) The velocity of precursor flame is ~85 cm/s under different initial fuel temperatures, which is significantly larger than the diffusion flame spread rate in low flammability limit. On the basis of the flame speed, the precursor flame is a premixed combustion flame in essence. (4) The diffusion flame proceeds to a point where the liquid temperature is equivalent to fire point. By contrast, the precursor flame locates at a position where the oil surface temperature corresponds to flashpoint temperature. Acknowledgement The authors would like to thank the Opening Fund of State Key Laboratory of Fire Science (No. HZ2016-KF09), and the Fundamental Research Funds for the Central Universities (No. JZ2015HGBZ0502) for their supports. Reference [1] C.G. Fan, J.Q. Zhang, K.J. Zhu, K.Y. Li, An experimental study of temperature and heat flux in a channel with an asymmetric thermal plume, Applied Thermal
Engineering,
113
(2017)
1128-1136.
http://dx.doi.org/10.1016/j.applthermaleng.2016.11.017. [2] T. Hirano, Flame Spread Across Condensed Combustibles, in: W.A. Sirignano, A.G. Merzhanov, L.D. Luca (eds.) Advances in Combustion Science: In Honor of Ya. B. Zel'dovich, American Institute of Aeronautics and Astronautics, 1997, pp. 131-144. http://dx.doi.org/10.2514/4.866456. [3] F. Tang, L. Hu, Z. Qiu, Q. Wang, A global model of plume axial temperature profile transition from axisymmetric to line-source pool fires in normal and reduced
pressures,
Fuel,
130
(2014)
211-214.
http://dx.doi.org/10.1016/j.fuel.2014.04.053. [4] H.D. Ross, F.J. Miller, Understanding flame spread across alcohol pools, in: Proceedings of the sixth international symposium, International Association for Fire
Safety
Science,
Poitiers,
France,
2000,
pp.
77-94.
http://dx.doi.org/10.3801/IAFSS.FSS.6-77. [5] E. Degroote, Control parameters of flame spreading in a fuel container, Journal of Thermal
Analysis
and
Calorimetry,
87
(2007)
149-151.
http://dx.doi.org/10.1007/s10973-006-7838-1. [6] E. Degroote, P.L. Garcia-Ybarra, Flame propagation over liquid alcohols. Part III. Pulsating regime, Journal of Thermal Analysis and Calorimetry, 80 (2005) 555-558. http://dx.doi.org/10.1007/s10973-005-0693-7. [7] E. Degroote, P.L. Garcia-Ybarra, Flame propagation over liquid alcohols. Part II. Steady propagation regimes, Journal of Thermal Analysis and Calorimetry, 80
(2005) 549-553. http://dx.doi.org/10.1007/s10973-005-0692-8. [8] D. White, C.L. Beyler, C. Fulper, J. Leonard, Flame spread on aviation fuels, Fire Safety
Journal,
28
(1997)
1-31.
http://dx.doi.org/10.1016/s0379-7112(96)00070-7. [9] J. Zhou, G. Chen, P. Li, B. Chen, C. Wang, S. Lu, Analysis of flame spread over aviation
kerosene,
Chinese
Science
Bulletin,
55
(2010)
1822-1827.
http://dx.doi.org/10.1007/s11434-010-3014-x. [10] R. Mackinven, J.G. Hansel, I. Glassman, Influence of laboratory parameters on flame spread across liquid fuels, Combustion Science and Technology, 1 (1970) 293-306. http://dx.doi.org/10.1080/00102206908952209. [11] J.P. Burelbach, M. Epstein, M.G. Plys, Initiation of flame spreading on shallow subflash
fuel
layers,
Combustion
and
Flame,
114
(1998)
280-282.
http://dx.doi.org/10.1016/s0010-2180(97)00290-3. [12] N. Wu, G. Kolb, J.L. Torero, The effect of weathering on the flammability of a slick of crude oil on a water bed, Combustion Science and Technology, 161 (2000) 269-308. http://dx.doi.org/10.1080/00102200008935820. [13] H.D. Ross, Ignition of and flame spread over laboratory-scale pools of pure liquid fuels, Progress in Energy and Combustion Science, 20 (1994) 17-63. http://dx.doi.org/10.1016/0360-1285(94)90005-1. [14] I. Glassman, F.L. Dryer, Flame spreading across liquid fuels, Fire Safety Journal, 3 (1981) 123-138. http://dx.doi.org/10.1016/0379-7112(81)90038-2. [15] J. Guo, S. Lu, C. Wang, Effect of fuel depth on flame spread over aviation
kerosene,
Journal
of
Fire
Sciences,
32
(2014)
43-51.
http://dx.doi.org/10.1177/0734904113491705. [16] D.N. Schiller, H.D. Ross, W.A. Sirignano, Computational analysis of flame spread across alcohol pools, Combustion Science and Technology, 118 (1996) 203-255. http://dx.doi.org/10.1080/00102209608951980. [17] E.M. Remick, Flame spread over liquid fuels, Ph.D. thesis, Cornell University, New York, 1981. [18] M. Li, S. Lu, J. Guo, R. Chen, K.-L. Tsui, Initial Fuel Temperature Effects on Flame Spread over Aviation Kerosene in Low- and High-Altitude Environments, Fire
Technology,
51
(2015)
707-721.
http://dx.doi.org/10.1007/s10694-014-0395-4. [19] K. Akita, Some problems of flame spread along a liquid surface, Proceedings of the
Combustion
Institute,
14
(1973)
1075-1083.
http://dx.doi.org/10.1016/s0082-0784(73)80097-9. [20] F. Tang, L. Hu, Q. Wang, Z. Ding, Flame pulsation frequency of conduction-controlled rectangular hydrocarbon pool fires of different aspect ratios in a sub-atmospheric pressure, International Journal of Heat and Mass Transfer,
76
(2014)
447-451.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2014.04.047. [21] D.N. Schiller, W.A. Sirignano, Opposed-flow flame spread across n-propanol pools, Proceedings of the Combustion Institute, 26 (1996) 1319-1325. http://dx.doi.org/10.1016/s0082-0784(96)80350-x.
[22] D.N. Schiller, H.D. Ross, W.A. Sirignano, Computational Predictions of Flame Spread over Alcohol Pools, in: 31st Aerospace Sciences Meeting & Exhibit, American Institute of Aeronautics and Astronautics, Reno, NV, 1993. http://dx.doi.org/10.2514/6.1993-825. [23] V. Babrauskas, Ignition handbook, Fire Science Publishers, Issaquah WA, USA. Co-published by the Society of Fire Protection Engineers, 2003. [24] J. Crank, The Mathematics of Diffusion, 2ed ed., Clarendon Press, London, 1975. [25] T. Konishi, Flame spread over liquid fuels, Ph.D. thesis, Oita University, Oita, Japan, 1998. [26] M. Li, S. Lu, J. Guo, R. Chen, K.-L. Tsui, Flame spread over n-butanol at sub-flash temperature in normal and elevated altitude environments, Journal of Thermal
Analysis
and
Calorimetry,
119
(2015)
401-409.
http://dx.doi.org/10.1007/s10973-014-4190-8. [27] A. Chakraborty, Thermodynamic trends for the adsorption of non polar gases on activated carbons employing a new adsorption isotherm modelling, Applied Thermal
Engineering,
105
(2016)
189-197.
http://dx.doi.org/10.1016/j.applthermaleng.2016.05.160. [28] F.J. Miller, H.D. Ross, Smoke visualization of the gas-phase flow during flame spread across a liquid pool, Proceedings of the Combustion Institute, 27 (1998) 2715-2722. http://dx.doi.org/10.1016/S0082-0784(98)80127-6. [29] J. Ji, S. Lin, C. Zhao, K. Li, Z. Gao, Experimental study on initial temperature influence on flame spread characteristics of diesel and gasoline–diesel blends,
Fuel, 178 (2016) 283-289. http://dx.doi.org/10.1016/j.fuel.2016.03.072. [30] M. Li, S. Lu, R. Chen, C. Wang, Pulsating behaviors of flame spread across n-butanol fuel surface, Applied Thermal Engineering, 112 (2017) 1445-1451. http://dx.doi.org/10.1016/j.applthermaleng.2016.10.001. [31] M. Mahdavi, S. Qiu, S. Tiari, Numerical investigation of hydrodynamics and thermal performance of a specially configured heat pipe for high-temperature thermal energy storage systems, Applied Thermal Engineering, 81 (2015) 325-337. http://dx.doi.org/10.1016/j.applthermaleng.2015.02.031. [32] B.M.C. Brantley, Quantitative analysis of a turbulent wind tunnel with obstructions for use in liquid flame spread experiments, M.S. thesis, Air University, Ohio, 2006. [33] G.N. Badami, A. Egerton, The Determination of Burning Velocities of Slow Flames, Proceedings Of The Royal Society Of London Series A-Mathematical And Physical Sciences, 228 (1955) 297-322. http://dx.doi.org/10.2307/99660. [34] Standard Test Method for Flash and Fire Points by Cleveland Open Cup Tester, ASTM D92 - 12b. 1-10. http://dx.doi.org/10.1520/D0092-12B.
Figure Captions Fig. 1. Experimental arrangement for flame spreading over RP-5. Fig. 2. Periodic appearance of precursor flame. Fig. 3. Leading edge positions of diffusion and precursor type flames versus time. Fig. 4. Precursor flame length as a function of spreading time. Fig. 5. Pulsation frequency and wavelength of precursor flame versus initial fuel temperature. Fig. 6. Schematic illustration of gas- and liquid-phase flows in flame spread. Fig. 7. Velocity of flame spread versus initial temperature of RP-5. Fig. 8. Infrared images of oil surface for flame spreading over RP-5. Fig. 9. Surface temperatures for flame spread over oil surface. Fig. 10. Oil surface temperatures under flame front. Fig. 11. Details of the thermocouple readings.
Fig. 1. Experimental arrangement for flame spreading over RP-5.
Fig. 2. Periodic appearance of precursor flame.
Fig. 3. Leading edge positions of diffusion and precursor type flames versus time.
Fig. 4. Precursor flame length as a function of spreading time.
Fig. 5. Pulsation frequency and wavelength of precursor flame versus initial fuel temperature.
Fig. 6. Schematic illustration of gas- and liquid-phase flows in flame spread.
Fig. 7. Velocity of flame spread versus initial temperature of RP-5.
Fig. 8. Infrared images of oil surface for flame spreading over RP-5.
Fig. 9. Surface temperatures for flame spread over oil surface.
Fig. 10. Oil surface temperatures under flame front.
Fig. 11. Details of the thermocouple readings.
Table Caption Table 1 Environmental testing conditions and technical parameters of RP-5. Experimental parameters
Measurement unit
Data
Environment temperature
o
C
13-20
Ambient humidity
%
36-39
Testing oil temperature
o
16.5-78.8
Equivalent molecular formula
-
C11.9H22.2
Latent heat of evaporation
kJ/kg
383
Boiling point
o
181
Fire point
o
95
Flashpoint
o
66
Dynamic viscosity (20 oC)
mPa·s
3.02
Surface tension (20 oC)
mN/m
25.8
Density (20 oC)
g/cm3
0.820
C
C C C