Introduction

Smoldering is an important, but often overlooked aspect of fire safety. Smolder detection is difficult due to the fact that the reaction temperatures are relatively low and occur within the porous material. For this reason, the smolder reaction can progress for long periods of time, undetected, and then undergo a sudden transition to flaming.

Transition from somldering to flames

The project objective is to perform smolder experiments on both ground- and space- based facilities, whereupon theoretical models of the process can be developed. The experiments are conducted using polyurethane foam as fuel and mixtures of oxygen/nitrogen as oxidizer. Thermocouples measure the temperature at several locations of the sample; and from the resulting temperature histories, smolder propagation velocity and smolder reaction temperature, as functions of the oxidizer flow velocity and oxygen concentration, can be obtained. The experimental results will serve to verify and to improve the theoretical models of smoldering combustion. Additional experiments study the conditions and mechanisms involved in the potential transition from smoldering to flaming.

Smoldering is important both as a fundamental combustion problem (i.e., the propagation of a heterogeneous, non-flaming, surface combustion reaction through a porous combustible material) and as a fire safety problem (i.e., the production of toxic compounds and the potential initiation of a fire by way of a transition to flaming). Microgravity introduces additional questions about the transport of mass and heat to and from the reaction zone, which must be resolved so that smolder behavior can be better predicted in a space-based environment.

Two Dimensional Smoldering and the Transition to Flaming

Experiments have been conducted to study the upward smolder characteristics and the phenomenon of transition to flaming of a porous combustible material in the presence of an air/porous-solid interface. External air flow velocities ranging from 0 m/s to 2.5 m/s are forced upward and forward across the exposed, flat, vertical surface of a high void fraction, flexible, polyurethane foam to examine how a smoldering fuel responds to increased oxygen supply and heat transfer. The smolder results of varying the air flow velocity reveal three regimes. First, for no flow (natural convection) to 0.25 m/s, smoldering either dies almost immediately following the ignition period or propagates very slowly. Second, for flow velocities between 0.25 m/s and 2.0 m/s, transition to flaming occurs with decreasing smolder duration relative to increasing air flow velocity. Visual observations and thermocouple histories, when compared to analyses of differential shear interferometry images of the gas phase at the interface, indicate that transition to flaming occurs inside the hot char region below the smoldering front and not at the air/porous-solid interface. Finally, as the air velocity is increased further (> 2.0 m/s), the smolder reaction is initially strong following the ignition period but always ends in almost abrupt extinction, due ultimately to convective cooling. These results indicate that smolder propagation with transition to flaming is the result of increased oxidizer supply and reduced heat losses.

Smoldering is defined as a non-flaming, self-sustaining, propagating, exothermic, surface reaction, deriving its principal heat from heterogeneous oxidation of the fuel (direct attack of oxygen on the surface of a solid phase fuel)[1]. Smoldering constitutes a serious fire risk because it typically yields a substantially higher conversion of fuel to toxic compounds than does flaming (though more slowly), is difficult to detect (and extinguish) in the interior of a porous material, and provides a pathway to flaming that can be initiated by heat sources much too weak to cause a flame directly. This transition process from slow smoldering to rapid flaming (fast, exothermic, gas-phase reactions) is also of particular interest as a fundamental combustion problem, with, surprisingly, very little information currently available.

Ortiz-Molina et al. [2] studied the relative smoldering tendency of different flexible polyurethane foams, in a horizontal configuration, by varying the ambient oxygen concentrations. A few experiments showed transition to flaming for very high oxygen concentrations. However, the work primarily concerned itself with the threshold conditions at which transition to extinguishment occurs. Chen et al. [3] examined the behavior of cellulosic materials (grain and wood byproducts and paper) in 20 cm layers at external air stream velocities up to 6 m/s. In this case, the fuel layer was set horizontally into the bottom of a flow tunnel so that its top surface was initially flush with the tunnel floor and the bottom surface was heated uniformly. Glowing combustion with subsequent transition to flaming occurred for air velocities less than 3 m/s. Ohlemiller's work [4] with cellulose remains the most detailed study on the matter to date. He examined smoldering of thick, horizontal layers of permeable fuel (cellulosic insulation) in the presence of flowing air and found that while opposed smolder responded only weakly to an increased air flow with no transition to flaming at flow velocities up to 5 m/s, forward smolder responded strongly to increased air flow and yielded transition to flaming at about 2 m/s.

It is well known that a forward, upward smoldering fuel responds to an increased oxygen supply by becoming faster and hotter until, eventually, flames erupt [5, 6, 7,]. This paper presents observations of two-dimensional, upward smoldering of open cell, flexible, polyurethane foam in the presence of forward, forced air flow at a vertical, air/porous-solid interface and the transition to flaming. In the particular smoldering configuration treated here, the foam is ignited at the bottom; and the smolder wave propagates upward. Thus, the resulting smolder may be termed forward smolder because the smolder wave propagates in the same direction as the forced air flow as well as the upward, buoyancy induced flow. In this work, the use of polyurethane foam, a very common material whose geometric structure tends to be preserved during and after the passage of a smolder wave [8], permits upward burning experiments without the fuel collapse and material erosion problems that occur with cellulose and other loose material. Ortiz-Molina et al. [3] finds that the smoldering behavior of flexible polyurethane foams can in large part be interpreted in terms of the mechanism postulated for cellulose. Therefore, the use of this material provides us with a degree of generality, in addition to advantageous thermophysical properties. Upward smolder is an intrinsically unsteady process controlled not only by the supply of oxidizer, but where the accumulation of heat in the foam, transported by the products of combustion, increases the smoldering velocity, and may lead eventually to gas phase pyrolysis and transition to flaming [6, 7]. Forward smolder, as described in Refs. [5, 6], responds strongly to increased air flow and is known to yield transition to flaming. Consequently, such an arrangement is employed in this work.

From analyses of the physical transformations and temperature histories involved in the smoldering process, we can formulate the following broad model of the behavior of flexible polyurethane foams in the smolder process [9, 10, 11]:

For our upward, forward smolder configuration, as the heat from the product gases is convected downstream of the smolder front, the virgin foam becomes preheated, thereby reducing the additional thermal energy needed to pyrolyze the foam and initiate its oxidation. Consequently, local temperatures and smolder propagation velocities can increase, accelerating perhaps to a transition to flaming. Although this is a possible mechanism for the phenomenon [2, 6, 7], experiments indicate that transition from smoldering to flaming in the present arrangement proceeds by way of a different path.

Steps 1 and 2 comprise the smolder mechanism of flexible polyurethane foam, whereby a complex polymer, C1.0H1.7N0.07O0.3, is converted to a char-like material of approximately C7.2H5NO (which is clearly not a simple carbonaceous char) [6]. Noteworthy, the char is also oxidizable, exhibiting experimentally that it is somewhat more resistant to oxidation than the original foam, but once reacted, is more exothermic. This process is documented in the literature, Ref. [2, 6-11], and is commonly referred to as secondary char oxidation. Ohlemiller [2] recognizes that the course of secondary char oxidation can be energetic enough to ignite flammable gases, initiating a transition to flaming. As it is shown below, it is this process that provides the pathway to flaming in the present experimental setup.

Forward Flow Smoldering Combustion

Within the framework of the Microgravity Smoldering Combustion Experiment the forward flow smolder study is being conducted to examine the effects that control smolder in this configuration and the requirements for ignition of a forward self-sustaining smolder. The mechanisms controlling the subsequent transition from smoldering to flaming will also be examined. A Fig. seen below illustrates the fundamental differences between forward and opposed flow smolder.

Ground based experiments are conducted in a forward flow configuration within a cylindrical, vertically oriented combustion chamber. Igniter power and total energy are varied under several flow conditions to determine the effects of both input heat flux and energy. Variable flow conditions enable the experimentalist an opportunity to quantify the effects of oxygen mass flow rate to the reaction zone and convective cooling.

The experimental results show that the ignition behavior of a porous combustible such as open-cell, unretarded, polyurethane is controlled by the igniter power, time of power input, and igniter temperature, and can be effectively modeled by a simple energy balancing scheme. This model substitutes the external ignition source for an internal on-going smolder reaction, then balances this with the energy flux required to bring the fuel ahead of the reaction up to the smolder temperature at a rate equal to the smolder velocity. The results show a minimum energy and temperature requirement for a given input heat flux to initiate a self-propagating smolder front. With input powers ranging from 30 to 90W, self propagating smolder was ignited on the order of 1000 seconds.

An examination of smoldering velocities, for the conditioins tested to date, has shown that the smolder velocity is independent of the ignition conditions for both opposed and forward flow smolder, as would be expected. No trend between smolder velocity and ignition power is evident. The smolder velocity is correlated with the air flow velocity used during the self-propagating portion of the experiment, however.

Under some cases the thermal decomposition of the fuel forms a tough skin at the edge of the char, virgin fuel interface. This skin is much less permeable than the foam or char and inhibits oxidizer transport, weakening the reaction to the point of extinction. The thermal decomposition of the foam (melting) has been also noted in a similar experiment utilizing an opposed flow configuration. It appears that this melting may be related to the input heat flux and the time the igniter is powered. Further experimental work is required to examine the cause of extinguishment of the smolder due to these melt phenomena.

Future work will examine the effect of buoyancy, oxygen concentration, and ignition flow rates on the ignition process. The effect of buoyancy, oxygen concentration, and the smolder forced air flow rate on the smolder velocity and temperature will also be examined. Buoyancy effects will be determined through the comparison of upward and downward smoldering experiments. The results of these experiments will also be compared with data obtained from future NASA microgravity Shuttle Experiments.

Opposed Flow Smoldering

Within the framework of the Microgravity Smoldering Combustion Experiment the opposed flow smolder study is being conducted to examine the effects that control smolder in this configuration and the requirements for ignition of an opposed self-sustaining smolder.

Ground based experiments are conducted in an opposed flow configuration within a vertically oriented combustion chamber.

Igniter power and total energy are varied under several flow conditions to determine the effects of both input heat flux and energy. Variable flow conditions enable the experimentalist an opportunity to quantify the effects of oxygen mass flow rate to the reaction zone and convective cooling. Comparison with results from similar microgravity experiments identifies the role of buoyancy to the overall reaction.

The experimental results from these tests show that the ignition behavior of a porous combustible such as open-cell, unretarded, polyurethane is controlled by the total power, time of power input, and igniter temperature, and can be effectively modeled by a simple energy matching scheme. This model equates the external ignition source to the heat released by an internal ongoing smolder reaction. The results show a minimum energy and temperature requirement for a given input heat flux to initiate a self-propagating smolder front for both opposed flow and forward flow smolder. For the opposed flow case, with input powers ranging from 30 to 90W, self propagating smolders were ignited on the order of 1000 seconds.

An examination of smoldering velocities shows that the smolder velocity is independent of the ignition conditions for opposed flow smolder, as would be expected. No trend between smolder velocity and ignition power is evident. The smolder velocity is correlated with the air flow velocity used during the self-propagating portion of the experiment, however. Under some test cases the thermal decomposition of the fuel forms a tough skin at the edge of the void. This skin has diffusive properties quite different from the virgin foam or residual char, and is much less permeable. The thermal decomposition of the foam (melting) is noted in both experimental configurations (forward and opposed smolder) and it appears that it may be related to the input heat flux and the time the igniter is powered. Further experimental work is required to examine the cause of extinguishment of the smolder due to these melt phenomena. Future work will examine the effect of buoyancy, oxygen concentration, and ignition flow rates, (Vign), on the ignition process. T he effect buoyancy, oxygen concentration, and the smolder flow rate, (Vox), on the smolder velocity and temperature will also be examined. Buoyancy effects will be determined through the comparison of upward and downward smoldering experiments for both the forward and opposed flow cases. The results of these experiments will also be compared with data obtained from the NASA microgravity Shuttle Experiments.

Opposed Flow Smoldering

These movies show the evolution of the permeability (left) and temperature (right) as a smolder wave penetrates into a cylindrical polyurethane foam fuel sample. The permeability is taken as a line of sight average and transformed, assuming axisymmetry, using the Abel transform. The igniter was located at the top of the sample subject to natural convection conditions. Each image represents 80-100 seconds that are required to scan the sample. The ultrasound pulse consists of a 6 period, 40kHz sine wave with an amplitude of 80Vp-p.

An ultrasonic imaging technique has been developed to examine the propagation of a smolder reaction through the interior of a porous combustible material. The technique provides information about the location of a propagating smolder front, as well as any permeability variations of the smoldering material. The method represents an important tool in the study of smoldering combustion, and has potential application in studies of other heterogeneous combustion processes. The method is based on the fact that, for an ultrasonic frequency defined by the porous medium properties, the transmission of an ultrasonic signal through a porous material increases with increasing permeability. Thus, since a propagating smolder reaction leaves behind char with a higher permeability than the original material, ultrasound transmission can be employed to differentiate virgin material from char. Furthermore, if the char continues to react, the technique can be used to investigate the evolution of the permeability of the char. In this work, experiments are presented where the technique is applied to smoldering combustion in a two-dimensional geometry. The results of applying the technique have furthered the understanding of two-dimensional smolder, especially in elucidating the controlling mechanisms leading to the transition from smoldering to flaming. The applicability of ultrasonic tomography to smoldering combustion has also been successfully applied.

The ultrasonic imaging technique developed here is based on the evidence that, for an ultrasonic frequency defined by the porous medium properties, the degree of attenuation of an ultrasonic signal through a porous material decreases with increasing permeability. Thus, since a propagating smolder reaction leaves behind char with a higher permeability than the original material, the relative attenuation of ultrasonic transmissions can be employed to differentiate virgin material from char and to monitor any permeability variations of a smoldering material.

In the method developed here, an acoustic frequency in the ultrasonic regime is employed because of its spatial resolution and distinction from background noise. In addition, the frequency to be employed had to be the highest frequency (for spatial resolution purposes) that could provide a detectable transmission (at reasonable speaker output powers) through virgin foam samples of the thickness (~ 16 cm) used throughout the present smolder study. Furthermore, noticeable relative magnitudes of transmission between virgin foam and char had to exist such that a smoldering interface could be visualized. The 40 kHz frequency applied in this work performed best under these criteria, with commercially available pressure transducers (speakers/microphones). The spatial resolution is of the order of 1 cm, corresponding to the ultrasonic wavelength and the size of the diaphragm of the microphone. Only a limited number of commercial pressure transducers were tested; and conceivably, pressure transducers with frequency responses higher than 40kHz that provide better resolution and adaptability may be available and should be investigated.

Another design consideration is whether to employ a continuous ultrasonic wave or an ultrasonic wave-train pulse. Although slightly more difficult to employ, an ultrasonic wave-train pulse allows for a more precise transmission measurement because the first peak in the received wave-form identifies the desired transmitted signal through the foam/char, distinguishing it from reflection and other interference signals. In addition, the frequency at which the wave-train pulses are sent can be reduced, permitting the speakers to be driven at higher powers without overheating. Moreover, the time of flight of the wave-train pulse can be used to measure the average temperature along its propagation path. Although this feature is not employed in this work, it is the subject of future work.

Operation procedure, along with design considerations, for the ultrasonic imaging technique for a single set of speaker and microphone are as follow:

1) A speaker emits a 40 kHz ultrasonic sinusoidal wave-train pulse through the porous medium. The wave-train pulse consists of a given number of cycles (6). The duration of the wave-train pulse is experimentally determined to maximize amplitude magnitude versus time spread of the transmitted wave-train pulse due to superposition of diffracted and reflected signals in the porous medium.

2) A microphone receives a wave-form which includes the transmitted wave-train pulse along with reflection and other interference signals. This wave-form is amplified and converted to an RMS signal. Parenthetically, we are only interested in relative magnitudes of transmission amplitude modulation; thus the absolute value of the RMS signal is not critical. Based on the shortest path length through the sample, the first peak in the received wave-form identifies the desired transmitted signal through the foam/char.

3) The received RMS wave-form is digitally sampled by a computer and stored into memory. The entire received wave-form constitutes a single ultrasonic transmission data point, where the attenuation of the transmitted wave-train pulse is deduced in post-processing. Figure 1 shows the wave-form received by a microphone, after amplification and RMS conversion.

4) The next ultrasonic transmission data point is taken. The frequency of wave-train pulses is very important. The time between wave-train pulses must be longer than the time of flight for a single pulse. Enough time must also be allowed to minimize superposition effects from previously sent wave-trains that can be reflected back into the propagation path of interest. The time interval is limited by the desired rate of ultrasonic transmission data points to be taken.

In the present setup , linear arrays of speaker/microphones are employed, to reduce the time needed to scan the sample. Line-of-sight transmission projections are produced by scanning the samples with horizontal arrays of speakers and microphones, mounted to a 2-axis moving assembly. 40 kHz sinusoidal waves in 6 period bursts are produced by a synthesized function generator, triggered by a function generator. The pulsed signals are amplified and are sent sequentially, via relay modules, from an array of 8 ultrasonic speakers. An array of 8 ultrasonic microphones receive the transmitted pulsed signals. These received wave-forms are pre-amplified, multiplexed, amplified by a gain programmable differential amplifier, converted to a RMS/DC signal, sampled at 250 kHz by a high speed A/D board (which is triggered by the function generator), and recorded onto a personal computer.

The resulting set of data points gives a 2-D image of the line-of-sight average of the attenuation of the sample. Even at this point, much information about the sample can be obtained, however, we can go one step further. 3-D tomographic techniques can be applied. Here, we have used the Abel Transform due to its relative simplicity. It requires an axisymetric sample, but the sample need only be scanned in one direction. Phantoms and smoldering sample have been successfully imaged to date.

The innovative application of ultrasonic imaging employed here brings the study of optically inaccessible heterogeneous combustion into a new perspective, where new insight into the mechanisms controlling smoldering combustion is revealed via visualization of evolving material properties. The technique is especially informative concerning char permeability evolution and propagation of the smolder front. Development of this technique into an ultrasonic tomography of non-axisymmetric material permeability and temperature fields for a smoldering fuel in real time is worth pursuing.