Flame Spread over Liquid Fuels and Heat Transfer Characters of Subsurface Convection Flow


Student thesis: Doctoral Thesis

View graph of relations


  • Manhou LI


Awarding Institution
Award date27 Jul 2015


In recent years, the rapid and continuing development of the global economy has demonstrated an increasing dependence on energy supplement. Oil, which is a traditional fossil energy resource, still accounts for the largest percentage of the total energy consumption in the world (approximately 33% in 2013). However, statistics indicates that a great number of serious fire or oil explosion accidents have occurred during the production, storage, transportation, and usage processes of liquid fuels. Once leaking fuel is exposed to an ignition source, a pool fire, which is accompanied by the typical phenomenon of flame spread across a liquid fuel surface, is easily established. This behavior also possibly occurs in an environment with an elevated altitude, such as air crashes during the takeoff or landing stages of an aircraft or fuel leakage in long-distance oil pipelines, which tend to cover a wide range of altitudes. Moreover, flame spread over liquid fuel is a complicated heat and mass transfer problem that has not been understood comprehensively. Considering both the potential catastrophe hazard and the scientific significance that involves flame spread over liquid fuels, as stated earlier, research on this field both in plain and plateau environments is highly important to approximate large-scale fuel-spilling fire behavior.

Considering noticeable behavioral variations between flame spread over hydrocarbon and alcohol fuels, two typical hydrocarbon fuels (aviation kerosene of RP-5 and 0# diesel) and a pure alcohol fuel (n-butanol) are chosen to study flame behaviors under various experimental conditions (initial fuel temperature, pool width fuel depth, and ambient pressure). According to the results of these experiments, several important fire parameters are established, including flame pulsation frequency, and wavelength, flame spread rate, subsurface convection flow size, and the temperature profile of the liquid surface. Based on the experimental data, a comprehensive model is established to explain the heat transfer characters of subsurface convection flow for liquid phase-controlled flame spread. The main conclusions drawn from the present experimental conditions are summarized as follows:

Flame spread over thick liquid fuels in a normal pressure plain environment is achieved. Two types of flame evidently exist in flame spread over hydrocarbon fuels: a blue precursor flame located ahead of a yellow main flame. The main flame is a flame shape under diffusion combustion, whereas the precursor flame belongs to premixed combustion. For a given initial fuel temperature, the pulsation wavelength of the precursor flame is longer for 0# diesel than for RP-5. In hydrocarbon fuels, diffusion flame spread is a coupling and periodical switching process between an advancing phase and a retreating phase. By contrast, flame spread over n-butanol pulsates in a jumping-crawling-jumping manner, in which a flame proceeds slowly, or even stops after a rapid advancement, but never retreats. Flame pulsation frequency increases nearly linearly as initial fuel temperature rises, which is predicted using Fick’s second law. For a given initial fuel temperature, the flame pulsation frequency of RP-5 is larger than that of 0# diesel. The existence of a gas-phase recirculation cell is a key factor in flame pulsation. A gas-phase recirculation cell should develop ahead of the flame tip because counter-current buoyancy dominates hot gas expansion effects, which causes pulsating flame spread. Otherwise, a uniform flame spread is achieved. The critical transition temperature between the uniform and pulsating flame spread regimes of n-butanol is predicted to be 28 °C, which is acceptably consistent with the experimental value of 30.2 °C. Flame spread rate increases slowly as initial fuel temperature rises until it approaches the critical transition temperature between liquid phase- and gas phase-controlled flame spread, which is approximately 82.5 °C for RP-5 and 92.2 °C for 0# diesel. The Clausius-Clapeyron relation predicts that an increase in flame spread rate with initial pool temperature is primarily caused by an increase in the initial amount of fuel vapor in the gas phase. For a given initial fuel temperature, the velocity of flame spread over n-butanol is significantly higher than that over hydrocarbon fuels. In the oscillatory regime of flame spread over n-butanol, non-dimensional pulsating velocity (VfmaxVfmin)/Vf varies directly proportionally with the square root of non-dimensional initial temperature (THbT0)/THb. The scaling analysis of gas phase-controlled flame spread indicates that the variation in flame spread rate Vf is directly proportional to 1/T01/3(TbT0)2, regardless of fuel type. The temperature profiles near the gas-liquid interface are measured. A steady-step temperature rise and a preheating time occur as subsurface convection flow proceeds because of the arrival of a thermal vortex front. For a given initial fuel temperature, steady-step temperature rise is larger in RP-5 than in diesel fuel, but preheating time is shorter. The measured temperature distributions normal to the oil surface reveal that liquid temperature in the upper layer is higher than that in the lower layer.

Flame spread over liquid fuels in a low-pressure and low-oxygen plateau environment is analyzed. Flame luminance is relatively weak at an elevated altitude, whereas flame height is high. Flame pulsates more frequently at a high initial fuel temperature or at an elevated altitude. In particular, the flame pulsation period of n-butanol, which varies with the dimensionless initial fuel temperature (T0 Tho)/Tho, correlates well with a semi-logarithmic fit, regardless of altitude. Flame spread rate at an elevated altitude is considerably larger than at a normal altitude; thus, the fire risk of a fuel spill accident is potentially higher in a plateau. The influence of altitude on flame spread rate is predicted by variations in flash point and surface tension with altitude. Steady-step temperature rise in Lhasa is slightly higher than that in Hefei, whereas preheating time and subsurface convection flow length are considerably shorter.

Flame spread over thin liquid fuels of aviation kerosene is revealed. Flame spread over RP-5 is divided into shallow pools for fuel depths below 4 mm and deep pools for fuel depths above 4 mm. Similarly, flame spread is divided into narrow and wide pools for a pool width of 12 cm. An increase in fuel depth or pool width for thin or narrow pool regimes noticeably decreases flame pulsation frequency, but increases flame pulsation wavelength. However, fuel depth or pool width for a deep or wide pool regime does not change both pulsating flame frequency, and wavelength. The division for deep and shallow pools is also verified by characteristic length scale ratio (h*/L*) and Fr /√Ma values. Moreover, for a thin or narrow pool regime, flame spreads more rapidly as fuel depth or pool width increases. The decrease in flame spread velocity for a shallow pool is mainly attributed to viscous shear effects from the metal substrate, whereas heat loss is relatively less important. A flame cannot spread forward unless fuel depth is larger than critical liquid depth. Critical liquid depth is inversely proportional to pool width when pool width is less than 16 cm. When pool width is above 16 cm, critical liquid depth stabilizes at approximately 1.2 mm. At critical liquid depth, surface deformation is a significant factor for flame extinction. The theoretical analysis indicates that surface deformation amount reaches 69% under critical liquid depth.

The energy transfer and preheating model of subsurface convection flow is established. The motion of subsurface convection flow is largely attributed to the Marangoni force and only minimally to buoyancy force. Subsurface convection flow velocities are calculated for both thin and thick pools based on surface tension variations in the liquid phase. The results indicate that subsurface convection flow velocity increases as pool depth or initial pool temperature increases; however, the theoretical values are considerably larger than the experimental data. Finally, heat transfer calculation proves that liquid-phase convection is the main heat transfer mode in flame spread over liquid fuels.

    Research areas

  • Flame spread, liquid fuel, initial fuel temperature, pool dimension, critical liquid depth, high altitude, subsurface convection flow