Flame spread over solid fuel beds
INTRODUCTION
Flame
spread over thin solid fuels is a useful paradigm for studying the behavior of
more complex flame spread problems, e.g., in building fires. Our experiments on
flame spread in non-standard atmospheres have shown that the unequal rates of
diffusion of thermal energy and oxidizer (the "Lewis number" effect)
are important influences which have not received much attention in prior
studies. We have also discovered a new type of flame spread process in which
the flame front is cellular and explained this in terms of Lewis number effects
in conjunction with reactant interdiffusion near extinction. Our theoretical
modeling has confirmed the importance of Lewis number effects and shown that
such effects can be described via extensions of classical theory.
Current
efforts focus on the effects of buoyancy-induced convection and radiative heat
transport on flame spread over both thin and thick fuel bed materials.
Experiments have been conducted at earth gravity and at microgravity (µg) in
the drop towers at the NASA-Glenn Research Center in Cleveland, OH. Most
experiments were conducted in a 20 liter combustion chamber (Fig. 1). Exactly the
same apparatus was used for 1g and µg tests. In most thin-fuel experiments 5 cm
wide Kimwipe samples 15 cm long were used and were held by aluminum quenching
plates. Thick-fuel samples studied have included polystyrene, polyurethane,
polyphenolic and carbon foams. The samples were ignited by an
electrically-heated Kanthal wire. The flame spread process was imaged via three
video cameras and a laser shearing interferometer.
The
objectives of these experiments is to study the effect of diluent type, and
particularly the radiative properties of the diluent gas, on flame spread rates
and extinction conditions. Our
current state of understanding of these topics is discussed below.
EFFECT
OF DILUENT TYPE
The
effect of inert gases He, Ar, N2, CO2 and SF6
on flame spread over thermally-thin fuels were tested since these inerts
provide a variety of radiative properties and oxygen Lewis numbers. For He, N2
and Ar diluents, the flame spread rates (Sf) at µg were always lower
than the 1g values and the minimum oxygen concentrations that would support
flame spread were lower at 1g than µg. These findings are consistent with prior
studies in O2-N2 atmospheres and results from the greater
radiative heat losses at µg due to the increased flame thickness (d) at µg. In contrast, for CO2 diluent,
Sf was slightly higher at g and the minimum O2
concentration was lower (Fig. 2). For SF6 diluent, Sf was
substantially higher at µg for all oxygen concentrations and the minimum
O2 concentration was significantly lower (Fig. 2). These
effects was attributed to changes in the character of radiative transfer for CO2
and SF6. He, Ar and N2 diluents do not emit thermal
radiation and thus only the H2O and CO2 combustion
products radiate significantly. For our test conditions tested the Planck mean
absorption length (LP) of the combustion products in these diluents
is typically 1 m, which is much larger than d,
consequently, radiative transport is optically thin. However, for CO2
and especially SF6 diluents, LP is comparable to d or smaller, thus reabsorption effects
cannot be neglected. With reabsorption, radiation emitted near the flame is not
lost to the surroundings and instead augments conventional thermal conduction
to increase Sf above radiation-free values. Interferometer images
(not shown) showed that d increased at µg for all
diluents but most dramatically for radiatively-active diluents, most likely due
to these reabsorption effects.
Experiments on flame spread over thermally-thick fuels have been
conducted using foam fuels to obtain low density and thermal conductivity, and
thus large spread rate (Sf) over thermally-thick fuels compared to
dense fuels such as PMMA. This
scheme enabled meaningful results to be obtained even in 2.2 second drop tower
experiments. It was found that, in
contrast conventional understanding; steady spread can occur over thick fuels
in quiescent microgravity environments, especially when a radiatively active
diluent gas such as CO2 is employed (Figs. 3, 4). This is proposed to be due to radiative
transfer from the flame to the fuel surface that can lead to steady spread even
when conductive heat transfer from the flame to the fuel bed is
negligible. Radiative effects are
more significant at microgravity conditions because the flame thickness is
larger and thus the volume of radiating combustion products is larger at
microgravity. The effects of
oxygen concentration and pressure are shown and the transition from
thermally-thick to thermally-thin behavior with decreasing bed thickness is
demonstrated. Radiative flux
measurements confirm the proposed effects of diluent type and gravity level. These results are particularly
noteworthy considering that the International Space Station employs CO2
fire extinguishers; our results suggest that helium may be a better inerting agent on both mass and mole bases
at microgravity even though CO2 is much better on a mole basis at
earth gravity.
|
|
|
|
Figure 1. Schematic of
drop frame and interferometer apparatus. The fuel bed is mounted inside the
chamber parallel to the plane of the page. |
Figure 2. Flame spread
rates vs. O2 mole fraction at 1 atm in helium, carbon dioxide and
sulfur hexafluoride diluents. |
|
|
|
|
Figure 3. Effect of oxygen concentration on
spread rates over thick solid fuel beds at µg and earth gravity. |
Figure 4. Effect of pressure on spread rate over
thick solid fuel beds at µg and earth gravity (polyphenolic foam, density
0.0267 g/cm3, atmosphere 40% O2 - 60% CO2.). |
FLAME SPREAD
IN PARTIALLY-PREMIXED ATMOSPHERES
Fires in
enclosures usually burn in under-ventilated conditions, leading to atmospheres
containing unburned fuel or intermediates such as CO. With this motivation, thin-fuel
flame spread experiments were conducted at 1g and g in atmospheres containing
sub-flammability-limit concentrations of gaseous fuels in O2-N2
atmospheres. CO and CH4 were used for the gaseous fuels in
partially-premixed atmosphere tests, plus H2, C3H8
and NH3 for 1g tests only. 1g tests showed that that for some fuels
such as CO and H2, there is a substantial effect of gaseous fuel on
Sf, whereas for other fuels such as NH3, there was
practically no effect. Remarkably, for CO fuel (Fig. 5), a very important case
for practical applications, Sf was higher and the minimum O2
concentration was lower when a given number of oxygen atoms in the ambient
atmosphere was present in the form of CO rather than O2. Moreover,
these data do not even account for the fact that in practical fires the
vitiated air will be hotter than ambient due to the heat release associated
with the partial oxidation. For all gaseous fuels enhancement of Sf
correlated well with the characteristic chemical reaction rate of the premixed
fuel and no correlation with the heating value or diffusive properties of the
premixed fuel was observed. In the current flight definition study, this work
will be extended to thick fuels using longer-duration g experiments.
It was found
that the effect of adding gaseous fuel to the ambient atmosphere was
qualitatively similar at 1g and g but the effect is stronger at g than 1g, and
in fact Sf is actually higher at g than 1g at high premixed fuel
concentrations (Fig. 6). Also, the effect of added gaseous fuel was found to be
more substantial at higher oxygen concentration and with CO fuel. All of these
results are consistent with the simple theoretical model proposed by the PI
which shows the effect of the premixed fuel is to cause a partially-premixed flame
sheet to occur upstream of the conventional non-premixed flame. This additional
flame increases the total heat flux to the fuel bed and thus Sf.
Finite-rate chemistry of the premixed flame was found to affect the additional
heat flux, even when the nonpremixed flame is at the high Damkohler number
(mixing-limited) condition.
|
|
|
|
Figure 5. Effect of O2
mole fraction and gaseous CO addition on Sf. Solid lines and data
points: experiment; dashed lines: theoretical predictions. f is the equivalence ratio of the
premixed atmosphere. |
Figure 6. Measured and
predicted flame spread rates vs. f. Atmosphere: 18% O2 in N2
at 1 atm with CO as added gaseous fuel. |
CONCURRENT-FLOW
FLAME SPREAD
A study of upward
flame spread over solid fuels was conducted to clarify the mechanisms of spread
rates for concurrent-flow flame spread and in particular, buoyancy effects on
this process. It was proposed that, contrary to many prior theoretical
predictions, upward flame spread could be steady because convective losses to
the sides of the fuel samples and/or surface radiative losses prevent the flame
length and thus spread rate from growing indefinitely. These losses were argued
to be unavoidable because the flame length will grow until these losses balance
the heat generation rate. Scaling relations for the spread rates in the
presence of convective and radiative losses, laminar and turbulent flow,
buoyant and forced convection, and thin and thick fuels were derived. Tests of
some of these relations were conducted for upward-propagating flames over tall,
thermally-thin fuel samples, subject to buoyant convection only, for a range of
pressures, oxygen mole fractions, diluents and fuel bed thicknesses. In this
manner a seven-decade range of Grashof number, defined as gW3/n2, where g is the
gravitational acceleration, W the fuel bed width and n the kinematic viscosity, was studied.
Only conditions away from quenching were studied to minimize chemical
influences. Flames were found to achieve steady values of both Sf and
flame length when the sample was sufficient tall. Measured values of Sf,
normalized by the opposed-flow (downward) spread rate with the same atmosphere
and fuel bed (Sf,opp), are shown in Fig. 7. At low GrW, Sf/Sf,opp
~ GrW1 with the value of the proportionality constant
being slightly different for different atmospheres. At higher W, Sf
is independent of GrW, indicating a transition to
radiatively-stabilized flame spread. At intermediate GrW, there is some
indication of a region where Sf/Sf,opp ~ GrW4/7
as would be characteristic of turbulent buoyant regime. The data deviate from Sf/Sf,opp
~ GrW1 behavior towards Sf/Sf,opp ~
GrW4/7 near GrW = 20,000, which is close to
where a predicted transition from laminar to turbulent behavior occurs.
Furthermore, visually the flames were observed to change to a turbulent
structure near this value of GrW.
Figure 7. Effect of Grashof number (GrW)
on upward flame spread rate, normalized by downward spread rate, over thin fuel
samples.
PUBLICATIONS
Greenberg, J.
B., Ronney, P. D., "Analysis of Lewis Number Effects in Flame
Spread," International Journal of Heat and Mass Transfer, Vol. 36, pp. 315-323
(1993).
Zhang, Y.,
Ronney, P. D., Roegner, E., Greenberg, J. B., "Lewis Number Effects on
Flame Spreading Over Thin Solid Fuels," Combustion and Flame, Vol. 90, pp. 71-83
(1992).