- Email: [email protected]
Ignition of low-density expandable polystyrene foam by a hot particle
ARTICLE IN PRESS
JID: CNF
[m5G;September 15, 2015;21:23]
Combustion and Flame 000 (2015) 1–7
Contents lists available at ScienceDirect
Combustion and Flame journal homepage: www.elsevier.com/locate/combustflame
Ignition of low-density expandable polystyrene foam by a hot particle Supan Wang a, Xinyan Huang b, Haixiang Chen a,∗, Naian Liu a, Guillermo Rein b a b
State Key Laboratory of Fire Science, University of Science and Technology of China, Jinzhai Road, Hefei, Anhui 230027, China Department of Mechanical Engineering, Imperial College London, London SW7 2AZ, UK
a r t i c l e
i n f o
Article history: Received 22 May 2015 Revised 19 August 2015 Accepted 19 August 2015 Available online xxx Keywords: Building fire dynamics EPS Insulation materials Embedding ignition Rolling ignition Mixing time
a b s t r a c t Insulating materials are ubiquitous in modern buildings for improving energy efficiency, but their high flammability becomes a significant fire safety issue. Many large fires in high-rise buildings were caused by the ignition of insulating materials by hot particles from fireworks and welding processes. Such ignition event is fundamentally different from the traditional flame or radiation driven ignition assumed in the literature, and still presents significant knowledge gaps. In this work, we study experimentally the ignition of a widely used insulation materials, expandable polystyrene (EPS) foam, by a hot steel particle under different conditions. In the experiments, a small spherical particle (6∼14 mm in diameter) was heated to a high temperature (>900 °C), and then placed on a bench-scale low-density (18 or 27 kg/m3 ) foam sample. It was observed that flaming ignition could only occur on the foam surface during its rolling process (rolling ignition) or before it was fully embedded (embedding ignition). The measurements suggested that larger particles held lower critical temperatures for ignition, which decreased from 1030 to 935 °C for diameters increasing from 6 to 14 mm. Compared to higher-density forest fuels in the literature, the critical particle temperature of EPS foam is much higher, with a narrower transition region for ignition probability of 5–95% and has a weaker dependence on the particle size. Results also show that both the sample density and thickness have a negligible influence on the ignition probability and mass-loss ratio. Theoretical analysis suggested that the hot particle acts as both heating and pilot sources, and the ignition of EPS foam is controlled by the competition between the gas mixing time and the particle residence time. © 2015 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
1. Introduction For improving energy efficiency, the use of plastic insulation materials such as polystyrene, polyisocyanurate and polyurethane is increasing in buildings and industrial sites because of the low cost and the superior insulating performance. However, these organic insulation materials are flammable [1,2], and their fire safety becomes a serious public concern. In the last decade, the ignition of external wall insulation by hot metallic particles (mainly generated from fireworks display and welding processes) has caused several disastrous fires, such as the Germany Düsseldorf Airport fire (1996), the China Centre Television fire (2009), and the high-rise residential hall fire in Shanghai Jing’an district (2010) [3]. More importantly, the process of hot-particle ignition is a fundamentally different ignition event, compared to the traditional flame or radiation driven ignition assumed in the fire science literature [1,2]. Significant knowledge gaps present on this complex ignition phenomenon. Therefore, a better understanding of the conditions for igniting insulation materials by hot metal
∗
Corresponding author. fax: +86 551 63601669. E-mail address: [email protected] (H. Chen).
particles is of great importance for the building fire safety. These ignition conditions, such as the critical temperature for particle size, can help determine the safe distance between the fireworks display space and high-rise buildings [4], and optimize the safety of welding operations [5]. Most building insulation materials are polymeric and porous fuels, such as expandable polystyrene (EPS) foam (close-cell) and polyurethane (PU) spray foam (open-cell) [6]. Most of these insulation materials have a low density (<50 kg/m3 ) to assure a superior insulation and weight performance. When the insulation materials are hit by a hot particle, it will be heated and pyrolysed locally. If the particle is hot and large enough, a flammable gaseous mixture can be generated and ignited into a flame [1]. Some fuels can also go through smouldering ignition [7]. In general, the ignition by a hot particle is a complex process, combining heat and mass transport, phase change, and chemical reactions. The ignition propensity depends on the fuel bed, particle, environment, and the landing conditions [2]. There are a limited number of studies on the ignition of fuel beds by hot metal particles in the literature (e.g., [8–19]). Wang et al. [8] numerically studied the ignition process on the low-density (18 kg/m3 ) EPS foam by small steel particles, and found the ignition required an extremely high particle temperature (>1000 °C).
http://dx.doi.org/10.1016/j.combustflame.2015.08.017 0010-2180/© 2015 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF 2
ARTICLE IN PRESS
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
Fig. 1. Schematic of experimental setup for the hot-particle ignition of EPS form samples.
All other studies used the high-density forest fuel beds (>200 kg/m3 ). Tanaka [9] conducted an experimental study to investigate the ignition condition of sawdust with various moisture contents by the welding spatters. Rowntree and Stokes [10–12] focused on the ignition of forest fuels, such as barley grass, hardwood forest litter and pine needle, exposed to incandescent particles from electrical arcing or frictional sparks. More recently, Fernandez-Pello and co-workers [13–16] have conducted a series of studies of the ignition of alphacellulose powder by steel, aluminium, copper and brass spheres over a range of initial temperatures and diameters (2 ∼ 19.1 mm). Their work simulated the ignition of forest fuels by the hot particles generated by power lines clashing, welding, grinding and various forms of hot work. Other studies focused on the ignition of forest fuels by firebrands (burning wooden particles) generated from wildfires and fires at wildland–urban interface (WUI) [17–19]. Compared with forest fuels, the insulation materials are of low density and melt when heated. Therefore, the ignition mechanism of insulation materials by a hot particle may be different from that of forest fuels. The aim of the present work is to study the ignition process of insulation materials by a hot particle and to provide a basic understanding of their ignition mechanism. In this work, a well-controlled experiment was designed to reproduce the ignition process of insulation materials by a hot particle in the laboratory. Various flaming ignition phenomena are observed and described in detail. The ignition limit, relating to the critical size and temperature of particle, was measured through a statistical analysis of a series of experiments. Afterwards, the hot-particle ignition mechanism of insulation materials is analysed using heat and mass transport theories and compared with that of forest fuel bed found in the literature. 2. Experimental setup
which was in contact with the particle surface. The temperature difference between T(particle) and T(tube) was not constant for the full temperature range but less than 50 °C, and remained fairly constant for the given particle size and furnace temperature. Under a steady tube temperature, the particle was quickly heated up and stabilized at a temperature (Tp ). Since the Bi number of the particle was very small (Bi ∼ 0.02 < 0.1 [20], see Appendix), the temperature inside the particle was uniform after reaching steady-state. Then, the particle was released to slip along the inclined tube (5° slope). The tube outlet rested on top of the sample to minimize the impact of the particle’s vertical momentum. The fuel sample was placed above the insulating base. All experiments were conducted without forced air flow to minimize ambient cooling and avoid disturbance to the emitted pyrolysis gases. The whole process was recorded by a video camera (at 25 fps), placed horizontally to the sample top surface to image the front view. Spherical steel particles with diameter of 6, 8, 10, 12 and 14 mm were tested, and the steady tube temperatures increased from 900 to 1100 °C in steps of 10 °C. Preliminary experiments found that 4mm (or smaller) particles required a very high temperature for ignition (> 1200 °C), close to the melting point of steel and beyond the furnace upper temperature limit. Therefore, particle temperatures above 1100 °C are not explored here. The EPS foam used in this work was produced by a local manufacturer via catalytic polymerization of styrene without flame retardant. For building insulation, the density of EPS foam is in the range of 15–40 kg/m3 , thus two typical foam densities of 18 and 27 (±0.5) kg/m3 were tested. The foam was cut into small samples of 100 × 100 mm2 cross section (Ao ) and 20– 100 mm thickness (H). For any given experimental condition, 10–20 repeats were performed to capture the full experimental uncertainty. The mass of the EPS foam sample in each run was measured before and after the experiment. 3. Experimental results In the experiment, only flaming ignition or no-ignition was observed. Unlike some forest fuels [15,16], smouldering ignition by a hot particle was never observed. We define a successful ignition as the presence of a visible flame that persisted for more than 1 s. The outcome of each run was categorized as “ignition” or “no ignition”. We report the ignition probability as the ratio of ignition times (Nig ) to the number of total repeating runs (Ntot ):
Pig =
Nig × 100% Ntot
(1)
Accordingly, we define the initial particle temperature with Pig = 50% as the critical temperature for ignition, and quantify a transition ignition region between Pig = 5% and 95%. 3.1. Ignition phenomena
A schematic of the experimental apparatus is illustrated in Fig. 1. A ceramic tube furnace was used to heat the hot particle up to a pre-set temperature. The tube temperature was monitored and adjusted based on the measurement of a Pt/Rh thermocouple. When the tube temperature was steady, a metal particle, held by a long-tail spoon, was placed into the centre of the tube. The particle temperature was monitored by a 0.5-mm K-type thermocouple
Flaming ignition occurred on the top surface of the sample during particle’s rolling over the sample surface (called “rolling ignition”), or after the rolling has ceased while before the hot particle is fully embedded (called “embedding ignition”). Once ignition occurred, flame always lasted for more than 1 s, differing from the unstable flash. Figure 2 presents a group of snapshots to illustrate a typical case of
Fig. 2. Illustrative case of the embedding ignition for the EPS foam (18 kg/m3 and 40 mm thick) by a hot steel particle (D = 6 mm and Tp = 1022 °C), recorded at 25 fps.
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF
ARTICLE IN PRESS
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
3
Fig. 3. The ignition process for the EPS foam (18 kg/m3 and 100 mm thick) by a hot steel particle with D = 12 mm (top) Tp = 935 °C “ignition”; and (lower) Tp = 925 °C “no ignition”, recorded at 25 fps.
embedding ignition, recorded by the front-view camera, for an EPS foam sample with a density of 18 kg/m3 and a thickness of 30 mm. This particle had a small diameter (D) of 6 mm, and an initial temperature of 1013 °C, near the critical ignition temperature. The time zero was set to the moment when the particle landed on the foam surface. Once landed, the particle rolled horizontally for 3∼4 cm, and stopped at 160 ms. In the next frame at 200 ms, a flash was observed, while the particle was partially embedded into the foam at a depth of d (
Fig. 4. Cross sections of the residual EPS foam (18 kg/m3 and 80 mm thick) hit by a hot steel particle (8 mm and 990 °C) in (a) no-ignition; and (b) ignition cases.
served on the rough top inner wall in Fig. 4(b), indicating that flame only attached to the fuel of top 2 cm before extinction. Such a neck structure limits the oxygen supply and prevents the flame spread to the bottom, so the flame could only last for around 2 s after ignition. 3.2. Ignition probability of EPS foam by hot particle Figure 5 shows the observed ignition probability distribution versus the particle size and initial temperature for EPS foam samples of two densities (18 and 27 kg/m3 ). Because of the complex ignition process, a large experiment uncertainty was found at each tested particle size, resulting in a non-uniform ignition transition zone. In order to enforce a comparison, the critical ignition temperature at Pig = 50% and the transition region bounded by Pig = 5% and 95% were selected and compared. Samples of both densities have a similar trend: for a smaller particle, a higher temperature was needed for ignition. For example, in order to ignite an 18 kg/m3 EPS sample with 50% probability, the critical particle temperature needs to increase from 935 to 1030 °C as the particle diameter decreases from 14 to 6 mm. The temperature range for the transition region is 40∼50 °C, similar for all particle sizes. In the experiments we used 20∼100 mm thick samples, and observed that the sample thickness had little influence on the ignition outcome. Comparison between Fig. 5(a and b) shows that samples of two densities have very small differences (less than 20 °C) in critical ignition temperatures, suggesting a weak density dependence on ignition
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
ARTICLE IN PRESS
JID: CNF 4
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
Fig. 5. Ignition probability as function of the initial temperature and diameter of hot particles for the EPS foam densities of (a) 18 kg/m3 , and (b) 27 kg/m3 .
Fig. 6. Critical embedding ratio (d f /D) for the transition region (ignition probability between 5% and 95%) versus the initial temperature and diameter of hot particle with a foam density of 18 kg/m3 where critical ratio seen at d f /D = 1/4.
under the tested range of particle sizes. This weak dependence on fuel density has also been found for the hot-particle ignition of cellulose fuels (290 kg/m3 in [14] and 338 kg/m3 in [16]). 3.3. Embedding condition for ignition For the ignition event, a flash always occurred during rolling (d → 0) or after the particle stopped, it was partially embedded into the foam (0 < d < D). Unlike the stable and long diffusion flame, the flash could follow the rolling particle (Fig. 3), and later covered a large volume compared to the particle volume of the particle for less than 100 ms. Therefore, this flash was most likely a premixed flame piloted by the hot particle. In other words, the hot particle acted as both heating and pilot sources. At the flashing moment, the embedding depth (d f ) can be directly measured from the video (see Figs. 2 and 3), so as to get the embedding ratio (d f /D). Figure 6 presents the observed ratio for the foam density of 18 kg/m3 versus the particle size and initial temperature in the transition region (i.e. Pig = 5 ∼ 95%). The error bars combine the uncertainty from two neighbouring frames and all repeating runs. As particle temperature increases, the embedding ratio decreases, and the chance for rolling ignition increases. For all particle sizes, the flash only occurred before half of the particle was embedded into the foam, and mostly occurred at the ratio of d f /D ≈ 1/4. Such an ignition would also explain why the sample thickness has little influence as long as it was thicker than the particle diameter.
Fig. 7. Fuel mass-loss ratio (m/m0 ) versus particle diameter (D) for the transition region (ignition probability between 5% and 95%) in the cases of ignition, no-ignition, and minimum fuel loss.
Moreover, as the particle was embedded deeper into the foam, the chance of pilot ignition actually decreased. It was probably because the descending of the particle also led to the removal of the pilot source. Measurements from the video showed that the embedding duration for d f /D ≈ 1/4 took about 40 ms (D = 6 mm) and 80 ms (D = 12 mm). For a small particle, the embedded ratio shows a large effect on the ignition probability because (1) the small particle is easily affected by the random factors in repeating runs, and (2) the time scale controlling the ignition by a small particle is also small, discussed more in Section 4.
3.4. Mass loss The measured mass-loss ratio, i.e. the mass loss normalized to the initial mass (m/m0 ), of EPS foam samples versus particle diameter in the ignition transition region are presented in Fig. 7. Each data point is averaged over all repeating runs of both densities (18 and 27 kg/m3 ) and all sample thicknesses (H = 20 ∼ 100 mm). The small error bar manifests a good experimental repeatability, and suggests that both the sample density and thickness have little influence on the mass-loss ratio, probably because a particle near ignition temperature is hot enough to penetrate all samples from top to bottom. Once fully penetrated, the minimum volume loss equals to the volume of tube path (π D2 H/4). Thus, the minimum mass-loss ratio becomes π D2 /4Ao which is plotted as reference in Fig. 7, where Ao is the sample cross-section area 100 × 100 mm2 .
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF
ARTICLE IN PRESS
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
As expected, a particle much hotter than the critical ignition temperature may produce a large fire to consume the entire foam sample, and a smaller or cooler particle cannot fully penetrate the foam. However, in the transition region, the fuel mass-loss ratio is always about twice the minimum value no matter ignition occurred or not, as shown in Fig. 7. This phenomenon implied that after flaming ignition, the heating effect from neither the hot particle nor the flame was strong enough to support fire spread. It is because under an intensive heating, the low-density EPS foam would quickly melt and shrink back. Therefore, a direct and continuous contact between hot particle and foam is crucial to the fuel mass-loss ratio. During the particle’s descent inside the foam, gravity promoted such a good contact between hot particle and fuel to pyrolyse or burn the fuel around its path, while the oxygen supply becomes limited.
4. Discussion
D, g
=2+
0.589Ra1/4 D 1 + (0.469/Pr)9/16
4/9
(4)
where Pr is the Prandtl number; and the Rayleigh number depends on the particle diameter as RaD ∼ D3 . Therefore, the mixing time (t ) mix, g
should increase with the particle size. Calculations (see Table A1 in Appendix) estimate that t = 29 ms (D = 6 mm, 1030 °C) and mix, g
70 ms (D = 12 mm, 930 °C), and can be fitted into t
mix, g
∼ D1.2 . For
the embedding ignition near particle stops, we choose the time between d f /D = 0 and 1/4 as the residence time: tr ≈ tD/4 . From the video, this residence time was measured to be 40 ms (D = 6 mm), 40 ∼ 80 ms (D = 8 or 10 mm), 80 ms (D = 12 mm) and 80 ∼ 120 ms (D = 14 mm), approximately linear with the particle size (tr ∼ D), and was comparable to t .
mix, g
(2)
The residence time increases as the moving speed (u) of particle decreases, tr ∼ 1/u. Therefore, the ignition tends to occur near the stop of particle and before its full embedding, which corresponds to the largest residence time. In general, the gas-phase chemical time is short, ∼O(10–4 ) s [21]. Therefore, the processes of pyrolysis and mixing determine the ignition delay. In current experiments, it is suggested that sufficient fuel should be pyrolysed within a very short time, because (1) the particle temperature (>900 °C) is much higher than the characteristic pyrolysis temperature of EPS (around 450 °C [22,23]), which can be quickly reached by the direct contact between hot particle and foam; and (2) the energy of particle is much larger than the pyrolysis energy to reach the lower flammability limit of the pyrolysate. Other three factors: (1) the weak particle-temperature dependence on foam density (Fig. 5), (2) ignition occurring before the particle half embedded (Fig. 6), and (3) similar fuel mass-loss ratio for two fuel densities and both ignition and no-ignition events (Fig. 7) also suggest that the pyrolysis time is very short. Therefore, the solid-phase and pyrolysistime based ignition theory [1,21], such as the critical ignition temperature, minimum energy or critical mass flux [24], is unlikely to explain the observed ignition phenomena. Instead, it is highly possible that the gas-phase mixing time controls the ignition delay in this work (tig ≈ tmix ≤ tr ). Because the particle acts as both ignition and heating source, the mixing time can be estimated as the diffusion time across the flammable gas mixture around the particle [21],
δ 2 (D/NuD )2 ≈ α α
Nu
mix, g
For pilot ignition, material needs to be heated to decompose and release flammable gases (pyrolysate) [2]. The duration of this process is characterized by a pyrolysis time, tpy . Then, pyrolysate needs to mix with the air, and reaches the lower flammability limit, characterized by a mixing time, tmix . After the pilot, the flammable mixture needs another moment (tchem ) for gas-phase chemical reactions to produce a flame [21]. Since the moving hot-particle also acts as the pilot source as seen before, a successful ignition requires an ignition delay shorter than the residence time of the particle
tmix ∼
a still sphere [25,26] can be adopted,
Figure 8(a) illustrates the particle size dependences on the mixing time by natural convection (t ) and the embedding residence
4.1. Hot-particle ignition mechanism of EPS foam
tig = tpy + tmix + tchem ≤ tr
5
(3)
where δ is the thickness of boundary layer, and α is the diffusivity of the mixture. In general, tmix decreases with increasing temperature because of the increasing diffusivity. Assuming the flash occurs near the particle stops, the Nusselt number for natural convection around
time (tr ). As discussed above, both time scales increase with the particle diameter but at different rates, and ignition occurs when they match each other (tig = t = tr ). For example, a large particle (Dc ) mix, g
has both large mixing time and large residence time, so a relatively low particle temperature (blue line) is required for ignition (at blue solid point). As the particle size reduces from Dc to Dh , no ignition occurs (at blue empty point) because the residence time becomes shorter than the mixing time (tr < t ). To ignite the fuel, a higher mix, g
particle temperature is required to reduce the mixing time (red line). At the same time, it is observed in experiment that tr changes little with the temperature in the transition region, so ignition occurs (at red solid point). Therefore, in order to ignite the fuel, a higher particle temperature is required for a smaller particle, agreeing with the experimental results in Fig. 5. This concise theory provides a qualitative analysis for the experimental temperature-diameter ignition criteria in Fig. 5. A quantitative analysis for ignition condition requires an accurate estimate of mixing time, and can be achieved with a comprehensive numerical simulation in future work. As the temperature increases to Pig = 95% or slightly above, a rolling ignition often occurs before the particle stops (see Fig. 3). The rolling residence time decreases with the particle speed (tr ≈ D/u), as illustrated in Fig. 8(b). The video camera recorded that the average moving speed is u¯ ≈ 0.3 m/s, leading to an average rolling residence time of t¯r ≈ 30 ms. The rolling of particle tends to slow down due to the friction. Meanwhile, the mixing time is controlled by the forced convection, so the Nusselt number [26,27] may be estimated as
Nu
D, u
= 2 + 0.6Re1/2 Pr1/3 D
(5)
where the Reynolds number increases with the speed (Re ∼ u). Therefore, the mixing time (t ) also decreases with the particle mix, u
speed. Calculations (see Table A1 in Appendix) estimate that t
mix, u
=
21 ms (D = 6 mm, 1030 °C) and 59 ms (D = 12 mm, 930 °C) at u = 0.3 m/s. Such the mixing time is comparable to t¯r ≈ 30 ms, and −0.8 . can be fitted to t ∼ u mix, g
Figure 8(b) illustrates the mixing time by forced convection (t ) and the rolling residence time (tr ) versus the particle speed. mix, u
As calculated above, both time scales decrease with the particle speed but at different rates. For a hotter particle (red line), the “rolling ignition” occurs when they match (tig,h = t = tr ) before the rolling mix, u
ceases. As the temperature decreases (blue line), the t
mix, u
increases,
which is higher than tr during the rolling. Then, near the full stop, the particle’s vertical embedding motion and the mixing by natural convection take over (t < t ), leading to an embedding mix, u
mix, g ,
ignition. This explains why rolling ignition tends to occur in the
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
ARTICLE IN PRESS
JID: CNF 6
Fig. 8. (a) Mixing time by natural convection (t
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
mix, g
) and embedding residence time (tr ) versus particle size; and (b) mixing time by forced convection (t
mix, u
) and rolling residence
time (tr ) versus particle speed. The red and blue lines represent a higher and a lower particle temperature, respectively; the solid and empty points represent ignition and no ignition, respectively.
Fig. 9. Comparison of the transition region of hot-particle ignition between EPS foam (18 kg/m3 ) and cellulose powder (338 kg/m3 ) [16] versus particle diameter and temperature.
higher-temperature part of transition region while embedding ignition occurs in the lower-temperature part.
Moreover, the transition ignition region of EPS foam was much narrower than that of cellulose powders in the whole range of particle sizes. One reason is that compared to commercial EPS foam samples, the fuel bed of cellulose powder packed in the laboratory (e.g., [9–16]) was less homogeneous, increasing the uncertainty in repeating runs. The other reason is that the cellulosic fuels can go through smouldering which can be ignited at a lower temperature or heat flux [7]. This smouldering ignition creates a shortcut to flaming ignition, especially under external air flow, as observed in [15,16], expanding the transition region (see Fig. 9). When the particle temperature is much higher than the 95% ignition probability point, the “rolling ignition” could generate a large fire, following the particle and attaching to the EPS foam surface, as occasionally observed in the experiment. Subsequently, the flames were able to spread over the sample and consume it entirely, defining a “point for ignition and fire spread”. Such fire event should require a much higher particle temperature than the current ignition (flaming) point, as illustrated in Fig. 9. For example, in the preliminary experiments, the particle with 16 mm diameter and 1133 °C temperature was obtained to result in the ignition and fire spread to the entire sample. So far, due to the limitation of furnace temperature and particle melting point, the point for ignition and fire spread is a reasonable prediction while has not been quantitatively reported for neither building insulation materials nor forest fuels, and thus deserve a separate study in future work. 5. Conclusions
4.2. Comparison to forest fuels Figure 9 compares the critical particle temperatures and diameters for igniting EPS foam (18 kg/m3 ) with those for cellulose powders (338 kg/m3 [16]). For all fuel types, a significantly higher particle temperature (∼1000 °C) is required for smaller particles. However, the high-density forest fuel beds can be ignited with much cooler particles (600 ∼ 700 °C) by large sizes (D > 6 mm) for two reasons: (1) the pyrolysis temperature of cellulose powders is around 350 °C [28], slightly lower than 450 °C for EPS [22,23]; and more importantly (2) cellulose powder has a density large enough to hold the hot particle and to enforce effective contact and heat transfer. Thus, even if the particle temperature is too low (around 700 °C) to initiate a pilot ignition, a large particle would still have sufficient energy and heat transfer into the cellulose powder, producing a flammable mixture and going through a spontaneous ignition. In other words, the ignition of high-density cellulose by a lower-temperature particle may be controlled by the pyrolysis time (tpy ), different from the ignition of the low-density EPS foam by the gas-phase mixing time (tmix ).
This paper presented an experimental study on the ignition of expandable polystyrene (EPS) foam by a hot metallic particle. The flaming ignition was observed on the foam surface before the particle was embedded or during its rolling process. The critical particle temperature for flaming ignition (50% probability) was found to decrease sensibly from 1030 to 935 °C as the particle diameter increases from 6 to 14 mm. In the ignition transition region, no fire spread or fuel burnout occurred after flame ignition, and whether ignition occurred or not made little difference in fuel mass-loss ratio. Tests also showed that the sample density and thickness had negligible influence on either the ignition limit or fuel mass-loss ratio. Theoretical analysis suggested that the hot particle acts as both heating and pilot sources, and ignition of EPS foam is controlled by the competition between the mixing time in the gas phase and the residence time of the particle. Therefore, the critical particle temperature for EPS ignition was much higher, accompanied by a narrower transition region, and less dependent on the particle size, differing from higher-density forest fuel beds found in the literature. This is the first study of the hot-particle ignition mechanism of building insulation
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
ARTICLE IN PRESS
JID: CNF
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
7
Table A1 Exemplary calculations of the mixing time (1) by natural convection for different particle sizes in the “embedding” ignition and (2) by the forced convection for different particle speeds in the “rolling” ignition at room temperature of 24 °C where RaD = gβ(Tp − T∞ )D3 /αν and β = 2/(Tp + T∞ ). “Embedding” ignition (u = 0 m/s)
Thermal properties Tp
ν × 10−6
α × 10−6
Pr
D
Ra
Nu
(°C)
(m2 /s)
(m2 /s)
(−)
(mm)
(−)
1030
84.9
120
0.71
109
0.70
6 8 10 12 14 6 8 10 12 14
261 620 1210 2092 3321 307 729 1424 2460 3906
930
7.64
t
(−)
(mm)
(ms)
3.83 4.27 4.68 5.07 5.45 3.90 4.36 4.79 5.20 5.59
1.57 1.87 2.14 2.37 2.57 1.54 1.84 2.09 2.31 2.51
29 41 54 66 78 31 44 57 70 82
D, g
materials, helping to further understand the fire hazard of other lowdensity fuels. Acknowledgments This work was sponsored by National Basic Research Program of China (973 Program, NO. 2012CB719702) and National Natural Science Foundation of China (51576184 and 51120165001). H.C. was supported by Fundamental Research Funds for the Central University (No. WK2320000020). All authors would like to thank Prof. Forman Williams (UC San Diego) for valuable comments. Appendix The Biot number for the largest particle with 14 mm diameter [20] is estimated as:
“Rolling” ignition (u > 0 m/s)
δ
400 W/ m2 ×K × 13 × 0.007[mm] hL ∼ Bi = = 0.02 < 0.1 k 51.9[W/(m × K)] References [1] D. Drysdale, An Introduction To Fire Dynamics, third ed., John Wiley & Sons Ltd, Chichester, 2011. [2] V. Babrauskas, Ignition Handbook: Principles and Applications to Fire Safety Engineering, Fire Investigation, Risk Management and Forensic Science, Fire Science Publishers, Issaquah, WA, 2003. [3] J. Sun, L. Hu, Y. Zhang , Theor. Appl. Mech. Lett. 4 (2013) 042001. [4] J. Song, S. Wang, H. Chen, Theor. Appl. Mech. Lett. 4 (2014) 11–034005.
mix, g
D
u
Re
Nu
δ
t
(mm)
(m/s)
(−)
(−)
(mm)
(ms)
0.2 0.3 0.4 0.5 0.6 0.2 0.3 0.4 0.5 0.6
77 116 155 193 232 31 47 63 79 94
9.8 11.6 13.1 14.4 15.6 4.99 5.66 6.23 6.73 7.18
0.61 0.52 0.46 0.42 0.39 2.41 2.12 1.93 1.78 1.67
26 21 18 16 14 76 59 49 42 37
6
12
D, u
mix, u
[5] Z. Lu, W. Chen, Fire Safety Sci. 18 (2009) 15–18 (in Chinese). [6] R. Bynum, R.T. Bynum, Insulation Handbook, McGraw Hill Professional, 2001, p. 512. [7] G. Rein, Smouldering Combustion, chapter 19, in: SFPE Handbook of Fire Protection Engineering, 5th ed., Springer, 2015 (in press). [8] S. Wang, H. Chen, N. Liu, J. Hazard. Mater. 283 (2015) 536–543. [9] T. Tanaka, Rep. Natl. Res. Inst. Police Sci. 30 (1977) 151–158. [10] A.D. Stokes, J. Electr. Electron. Eng. 10 (1990) 175–187. [11] A.D. Stokes, J. Electr. Electron. Eng. 10 (1990) 188–194. [12] G.W.G. Rowntree, A.D. Stokes, J. Electr. Electron. Eng. 14 (1994) 117–123. [13] R.M. Hadden, S. Scott, C. Lautenberger, A.C. Fernandez-Pello, Fire Technol. 47 (2011) 341–355. [14] J.L. Urban, C.D. Zak, C. Fernandez-Pello, 8th US National Combustion Meeting, Canyons resort in Park City, Utah 4 (2013) 3673–3680. [15] C.D. Zak, J.L. Urban, C. Fernandez-Pello, Combust. Sci. Technol. 186 (2014) 1618– 1631. [16] J.L. Urban, C.D. Zak, C. Fernandez-Pello, P. Combust. Inst. 35 (2015) 2707–2714. [17] S.L. Manzello, T.G. Cleary, J.R. Shields, J.C. Yang, Int. J. Wildland Fire 15 (2006) 427– 431. [18] S.L. Manzellon, T.G. Cleary, J.R. Shields, J.C. Yang, Fire Mater 30 (2006) 77–87. [19] P. Yin, N. Liu, H. Chen, J. Lozano, Y. Shan, Fire Technol. 50 (2014) 79–91. [20] F.P. Incropera, D.P. DeWitt, T.L. Bergman, A.S. Lavine, Fundamentals of Heat and Mass Transfer, 7th ed., John Wiley & Sons, New York, 2011. [21] J.G. Quintiere, Fundamentals of Fire Phenomena, John Wiley & Sons Ltd, 2006. [22] P. Kannan, J.J. Biernacki, D.P. Visco Jr, W. Lambert, J. Anal. Appl. Pyrol. 84 (2009) 139–144. [23] J. Li, S.I. Stoliarov, Combust. Flame 160 (2013) 1287–1297. [24] D. Rich, C. Lautenberger, J.L. Torero, J.G. Quintiere, C. Fernandez-Pello, P. Combust. Inst. 31 (2007) 2653–2660. [25] S. Churchill, Heat Exchanger Design Handbook, in: K.J. Bell (Ed.), Hemisphere Publishing, NY, 1983. [26] F.P. Incropera, Introduction to Heat Transfer, 6th ed., John Wiley & Sons, New York, 2011. [27] W. Ranz, W. Marshall, Chem. Eng. Prog. 48 (1952) 141–146. [28] H. Yang, R. Yan, H. Chen, D.H. Lee, C. Zheng, Fuel 86 (2007) 1781–1788.
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF
[m5G;September 15, 2015;21:23]
Combustion and Flame 000 (2015) 1–7
Contents lists available at ScienceDirect
Combustion and Flame journal homepage: www.elsevier.com/locate/combustflame
Ignition of low-density expandable polystyrene foam by a hot particle Supan Wang a, Xinyan Huang b, Haixiang Chen a,∗, Naian Liu a, Guillermo Rein b a b
State Key Laboratory of Fire Science, University of Science and Technology of China, Jinzhai Road, Hefei, Anhui 230027, China Department of Mechanical Engineering, Imperial College London, London SW7 2AZ, UK
a r t i c l e
i n f o
Article history: Received 22 May 2015 Revised 19 August 2015 Accepted 19 August 2015 Available online xxx Keywords: Building fire dynamics EPS Insulation materials Embedding ignition Rolling ignition Mixing time
a b s t r a c t Insulating materials are ubiquitous in modern buildings for improving energy efficiency, but their high flammability becomes a significant fire safety issue. Many large fires in high-rise buildings were caused by the ignition of insulating materials by hot particles from fireworks and welding processes. Such ignition event is fundamentally different from the traditional flame or radiation driven ignition assumed in the literature, and still presents significant knowledge gaps. In this work, we study experimentally the ignition of a widely used insulation materials, expandable polystyrene (EPS) foam, by a hot steel particle under different conditions. In the experiments, a small spherical particle (6∼14 mm in diameter) was heated to a high temperature (>900 °C), and then placed on a bench-scale low-density (18 or 27 kg/m3 ) foam sample. It was observed that flaming ignition could only occur on the foam surface during its rolling process (rolling ignition) or before it was fully embedded (embedding ignition). The measurements suggested that larger particles held lower critical temperatures for ignition, which decreased from 1030 to 935 °C for diameters increasing from 6 to 14 mm. Compared to higher-density forest fuels in the literature, the critical particle temperature of EPS foam is much higher, with a narrower transition region for ignition probability of 5–95% and has a weaker dependence on the particle size. Results also show that both the sample density and thickness have a negligible influence on the ignition probability and mass-loss ratio. Theoretical analysis suggested that the hot particle acts as both heating and pilot sources, and the ignition of EPS foam is controlled by the competition between the gas mixing time and the particle residence time. © 2015 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
1. Introduction For improving energy efficiency, the use of plastic insulation materials such as polystyrene, polyisocyanurate and polyurethane is increasing in buildings and industrial sites because of the low cost and the superior insulating performance. However, these organic insulation materials are flammable [1,2], and their fire safety becomes a serious public concern. In the last decade, the ignition of external wall insulation by hot metallic particles (mainly generated from fireworks display and welding processes) has caused several disastrous fires, such as the Germany Düsseldorf Airport fire (1996), the China Centre Television fire (2009), and the high-rise residential hall fire in Shanghai Jing’an district (2010) [3]. More importantly, the process of hot-particle ignition is a fundamentally different ignition event, compared to the traditional flame or radiation driven ignition assumed in the fire science literature [1,2]. Significant knowledge gaps present on this complex ignition phenomenon. Therefore, a better understanding of the conditions for igniting insulation materials by hot metal
∗
Corresponding author. fax: +86 551 63601669. E-mail address: [email protected] (H. Chen).
particles is of great importance for the building fire safety. These ignition conditions, such as the critical temperature for particle size, can help determine the safe distance between the fireworks display space and high-rise buildings [4], and optimize the safety of welding operations [5]. Most building insulation materials are polymeric and porous fuels, such as expandable polystyrene (EPS) foam (close-cell) and polyurethane (PU) spray foam (open-cell) [6]. Most of these insulation materials have a low density (<50 kg/m3 ) to assure a superior insulation and weight performance. When the insulation materials are hit by a hot particle, it will be heated and pyrolysed locally. If the particle is hot and large enough, a flammable gaseous mixture can be generated and ignited into a flame [1]. Some fuels can also go through smouldering ignition [7]. In general, the ignition by a hot particle is a complex process, combining heat and mass transport, phase change, and chemical reactions. The ignition propensity depends on the fuel bed, particle, environment, and the landing conditions [2]. There are a limited number of studies on the ignition of fuel beds by hot metal particles in the literature (e.g., [8–19]). Wang et al. [8] numerically studied the ignition process on the low-density (18 kg/m3 ) EPS foam by small steel particles, and found the ignition required an extremely high particle temperature (>1000 °C).
http://dx.doi.org/10.1016/j.combustflame.2015.08.017 0010-2180/© 2015 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF 2
ARTICLE IN PRESS
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
Fig. 1. Schematic of experimental setup for the hot-particle ignition of EPS form samples.
All other studies used the high-density forest fuel beds (>200 kg/m3 ). Tanaka [9] conducted an experimental study to investigate the ignition condition of sawdust with various moisture contents by the welding spatters. Rowntree and Stokes [10–12] focused on the ignition of forest fuels, such as barley grass, hardwood forest litter and pine needle, exposed to incandescent particles from electrical arcing or frictional sparks. More recently, Fernandez-Pello and co-workers [13–16] have conducted a series of studies of the ignition of alphacellulose powder by steel, aluminium, copper and brass spheres over a range of initial temperatures and diameters (2 ∼ 19.1 mm). Their work simulated the ignition of forest fuels by the hot particles generated by power lines clashing, welding, grinding and various forms of hot work. Other studies focused on the ignition of forest fuels by firebrands (burning wooden particles) generated from wildfires and fires at wildland–urban interface (WUI) [17–19]. Compared with forest fuels, the insulation materials are of low density and melt when heated. Therefore, the ignition mechanism of insulation materials by a hot particle may be different from that of forest fuels. The aim of the present work is to study the ignition process of insulation materials by a hot particle and to provide a basic understanding of their ignition mechanism. In this work, a well-controlled experiment was designed to reproduce the ignition process of insulation materials by a hot particle in the laboratory. Various flaming ignition phenomena are observed and described in detail. The ignition limit, relating to the critical size and temperature of particle, was measured through a statistical analysis of a series of experiments. Afterwards, the hot-particle ignition mechanism of insulation materials is analysed using heat and mass transport theories and compared with that of forest fuel bed found in the literature. 2. Experimental setup
which was in contact with the particle surface. The temperature difference between T(particle) and T(tube) was not constant for the full temperature range but less than 50 °C, and remained fairly constant for the given particle size and furnace temperature. Under a steady tube temperature, the particle was quickly heated up and stabilized at a temperature (Tp ). Since the Bi number of the particle was very small (Bi ∼ 0.02 < 0.1 [20], see Appendix), the temperature inside the particle was uniform after reaching steady-state. Then, the particle was released to slip along the inclined tube (5° slope). The tube outlet rested on top of the sample to minimize the impact of the particle’s vertical momentum. The fuel sample was placed above the insulating base. All experiments were conducted without forced air flow to minimize ambient cooling and avoid disturbance to the emitted pyrolysis gases. The whole process was recorded by a video camera (at 25 fps), placed horizontally to the sample top surface to image the front view. Spherical steel particles with diameter of 6, 8, 10, 12 and 14 mm were tested, and the steady tube temperatures increased from 900 to 1100 °C in steps of 10 °C. Preliminary experiments found that 4mm (or smaller) particles required a very high temperature for ignition (> 1200 °C), close to the melting point of steel and beyond the furnace upper temperature limit. Therefore, particle temperatures above 1100 °C are not explored here. The EPS foam used in this work was produced by a local manufacturer via catalytic polymerization of styrene without flame retardant. For building insulation, the density of EPS foam is in the range of 15–40 kg/m3 , thus two typical foam densities of 18 and 27 (±0.5) kg/m3 were tested. The foam was cut into small samples of 100 × 100 mm2 cross section (Ao ) and 20– 100 mm thickness (H). For any given experimental condition, 10–20 repeats were performed to capture the full experimental uncertainty. The mass of the EPS foam sample in each run was measured before and after the experiment. 3. Experimental results In the experiment, only flaming ignition or no-ignition was observed. Unlike some forest fuels [15,16], smouldering ignition by a hot particle was never observed. We define a successful ignition as the presence of a visible flame that persisted for more than 1 s. The outcome of each run was categorized as “ignition” or “no ignition”. We report the ignition probability as the ratio of ignition times (Nig ) to the number of total repeating runs (Ntot ):
Pig =
Nig × 100% Ntot
(1)
Accordingly, we define the initial particle temperature with Pig = 50% as the critical temperature for ignition, and quantify a transition ignition region between Pig = 5% and 95%. 3.1. Ignition phenomena
A schematic of the experimental apparatus is illustrated in Fig. 1. A ceramic tube furnace was used to heat the hot particle up to a pre-set temperature. The tube temperature was monitored and adjusted based on the measurement of a Pt/Rh thermocouple. When the tube temperature was steady, a metal particle, held by a long-tail spoon, was placed into the centre of the tube. The particle temperature was monitored by a 0.5-mm K-type thermocouple
Flaming ignition occurred on the top surface of the sample during particle’s rolling over the sample surface (called “rolling ignition”), or after the rolling has ceased while before the hot particle is fully embedded (called “embedding ignition”). Once ignition occurred, flame always lasted for more than 1 s, differing from the unstable flash. Figure 2 presents a group of snapshots to illustrate a typical case of
Fig. 2. Illustrative case of the embedding ignition for the EPS foam (18 kg/m3 and 40 mm thick) by a hot steel particle (D = 6 mm and Tp = 1022 °C), recorded at 25 fps.
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF
ARTICLE IN PRESS
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
3
Fig. 3. The ignition process for the EPS foam (18 kg/m3 and 100 mm thick) by a hot steel particle with D = 12 mm (top) Tp = 935 °C “ignition”; and (lower) Tp = 925 °C “no ignition”, recorded at 25 fps.
embedding ignition, recorded by the front-view camera, for an EPS foam sample with a density of 18 kg/m3 and a thickness of 30 mm. This particle had a small diameter (D) of 6 mm, and an initial temperature of 1013 °C, near the critical ignition temperature. The time zero was set to the moment when the particle landed on the foam surface. Once landed, the particle rolled horizontally for 3∼4 cm, and stopped at 160 ms. In the next frame at 200 ms, a flash was observed, while the particle was partially embedded into the foam at a depth of d (
Fig. 4. Cross sections of the residual EPS foam (18 kg/m3 and 80 mm thick) hit by a hot steel particle (8 mm and 990 °C) in (a) no-ignition; and (b) ignition cases.
served on the rough top inner wall in Fig. 4(b), indicating that flame only attached to the fuel of top 2 cm before extinction. Such a neck structure limits the oxygen supply and prevents the flame spread to the bottom, so the flame could only last for around 2 s after ignition. 3.2. Ignition probability of EPS foam by hot particle Figure 5 shows the observed ignition probability distribution versus the particle size and initial temperature for EPS foam samples of two densities (18 and 27 kg/m3 ). Because of the complex ignition process, a large experiment uncertainty was found at each tested particle size, resulting in a non-uniform ignition transition zone. In order to enforce a comparison, the critical ignition temperature at Pig = 50% and the transition region bounded by Pig = 5% and 95% were selected and compared. Samples of both densities have a similar trend: for a smaller particle, a higher temperature was needed for ignition. For example, in order to ignite an 18 kg/m3 EPS sample with 50% probability, the critical particle temperature needs to increase from 935 to 1030 °C as the particle diameter decreases from 14 to 6 mm. The temperature range for the transition region is 40∼50 °C, similar for all particle sizes. In the experiments we used 20∼100 mm thick samples, and observed that the sample thickness had little influence on the ignition outcome. Comparison between Fig. 5(a and b) shows that samples of two densities have very small differences (less than 20 °C) in critical ignition temperatures, suggesting a weak density dependence on ignition
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
ARTICLE IN PRESS
JID: CNF 4
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
Fig. 5. Ignition probability as function of the initial temperature and diameter of hot particles for the EPS foam densities of (a) 18 kg/m3 , and (b) 27 kg/m3 .
Fig. 6. Critical embedding ratio (d f /D) for the transition region (ignition probability between 5% and 95%) versus the initial temperature and diameter of hot particle with a foam density of 18 kg/m3 where critical ratio seen at d f /D = 1/4.
under the tested range of particle sizes. This weak dependence on fuel density has also been found for the hot-particle ignition of cellulose fuels (290 kg/m3 in [14] and 338 kg/m3 in [16]). 3.3. Embedding condition for ignition For the ignition event, a flash always occurred during rolling (d → 0) or after the particle stopped, it was partially embedded into the foam (0 < d < D). Unlike the stable and long diffusion flame, the flash could follow the rolling particle (Fig. 3), and later covered a large volume compared to the particle volume of the particle for less than 100 ms. Therefore, this flash was most likely a premixed flame piloted by the hot particle. In other words, the hot particle acted as both heating and pilot sources. At the flashing moment, the embedding depth (d f ) can be directly measured from the video (see Figs. 2 and 3), so as to get the embedding ratio (d f /D). Figure 6 presents the observed ratio for the foam density of 18 kg/m3 versus the particle size and initial temperature in the transition region (i.e. Pig = 5 ∼ 95%). The error bars combine the uncertainty from two neighbouring frames and all repeating runs. As particle temperature increases, the embedding ratio decreases, and the chance for rolling ignition increases. For all particle sizes, the flash only occurred before half of the particle was embedded into the foam, and mostly occurred at the ratio of d f /D ≈ 1/4. Such an ignition would also explain why the sample thickness has little influence as long as it was thicker than the particle diameter.
Fig. 7. Fuel mass-loss ratio (m/m0 ) versus particle diameter (D) for the transition region (ignition probability between 5% and 95%) in the cases of ignition, no-ignition, and minimum fuel loss.
Moreover, as the particle was embedded deeper into the foam, the chance of pilot ignition actually decreased. It was probably because the descending of the particle also led to the removal of the pilot source. Measurements from the video showed that the embedding duration for d f /D ≈ 1/4 took about 40 ms (D = 6 mm) and 80 ms (D = 12 mm). For a small particle, the embedded ratio shows a large effect on the ignition probability because (1) the small particle is easily affected by the random factors in repeating runs, and (2) the time scale controlling the ignition by a small particle is also small, discussed more in Section 4.
3.4. Mass loss The measured mass-loss ratio, i.e. the mass loss normalized to the initial mass (m/m0 ), of EPS foam samples versus particle diameter in the ignition transition region are presented in Fig. 7. Each data point is averaged over all repeating runs of both densities (18 and 27 kg/m3 ) and all sample thicknesses (H = 20 ∼ 100 mm). The small error bar manifests a good experimental repeatability, and suggests that both the sample density and thickness have little influence on the mass-loss ratio, probably because a particle near ignition temperature is hot enough to penetrate all samples from top to bottom. Once fully penetrated, the minimum volume loss equals to the volume of tube path (π D2 H/4). Thus, the minimum mass-loss ratio becomes π D2 /4Ao which is plotted as reference in Fig. 7, where Ao is the sample cross-section area 100 × 100 mm2 .
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
JID: CNF
ARTICLE IN PRESS
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
As expected, a particle much hotter than the critical ignition temperature may produce a large fire to consume the entire foam sample, and a smaller or cooler particle cannot fully penetrate the foam. However, in the transition region, the fuel mass-loss ratio is always about twice the minimum value no matter ignition occurred or not, as shown in Fig. 7. This phenomenon implied that after flaming ignition, the heating effect from neither the hot particle nor the flame was strong enough to support fire spread. It is because under an intensive heating, the low-density EPS foam would quickly melt and shrink back. Therefore, a direct and continuous contact between hot particle and foam is crucial to the fuel mass-loss ratio. During the particle’s descent inside the foam, gravity promoted such a good contact between hot particle and fuel to pyrolyse or burn the fuel around its path, while the oxygen supply becomes limited.
4. Discussion
D, g
=2+
0.589Ra1/4 D 1 + (0.469/Pr)9/16
4/9
(4)
where Pr is the Prandtl number; and the Rayleigh number depends on the particle diameter as RaD ∼ D3 . Therefore, the mixing time (t ) mix, g
should increase with the particle size. Calculations (see Table A1 in Appendix) estimate that t = 29 ms (D = 6 mm, 1030 °C) and mix, g
70 ms (D = 12 mm, 930 °C), and can be fitted into t
mix, g
∼ D1.2 . For
the embedding ignition near particle stops, we choose the time between d f /D = 0 and 1/4 as the residence time: tr ≈ tD/4 . From the video, this residence time was measured to be 40 ms (D = 6 mm), 40 ∼ 80 ms (D = 8 or 10 mm), 80 ms (D = 12 mm) and 80 ∼ 120 ms (D = 14 mm), approximately linear with the particle size (tr ∼ D), and was comparable to t .
mix, g
(2)
The residence time increases as the moving speed (u) of particle decreases, tr ∼ 1/u. Therefore, the ignition tends to occur near the stop of particle and before its full embedding, which corresponds to the largest residence time. In general, the gas-phase chemical time is short, ∼O(10–4 ) s [21]. Therefore, the processes of pyrolysis and mixing determine the ignition delay. In current experiments, it is suggested that sufficient fuel should be pyrolysed within a very short time, because (1) the particle temperature (>900 °C) is much higher than the characteristic pyrolysis temperature of EPS (around 450 °C [22,23]), which can be quickly reached by the direct contact between hot particle and foam; and (2) the energy of particle is much larger than the pyrolysis energy to reach the lower flammability limit of the pyrolysate. Other three factors: (1) the weak particle-temperature dependence on foam density (Fig. 5), (2) ignition occurring before the particle half embedded (Fig. 6), and (3) similar fuel mass-loss ratio for two fuel densities and both ignition and no-ignition events (Fig. 7) also suggest that the pyrolysis time is very short. Therefore, the solid-phase and pyrolysistime based ignition theory [1,21], such as the critical ignition temperature, minimum energy or critical mass flux [24], is unlikely to explain the observed ignition phenomena. Instead, it is highly possible that the gas-phase mixing time controls the ignition delay in this work (tig ≈ tmix ≤ tr ). Because the particle acts as both ignition and heating source, the mixing time can be estimated as the diffusion time across the flammable gas mixture around the particle [21],
δ 2 (D/NuD )2 ≈ α α
Nu
mix, g
For pilot ignition, material needs to be heated to decompose and release flammable gases (pyrolysate) [2]. The duration of this process is characterized by a pyrolysis time, tpy . Then, pyrolysate needs to mix with the air, and reaches the lower flammability limit, characterized by a mixing time, tmix . After the pilot, the flammable mixture needs another moment (tchem ) for gas-phase chemical reactions to produce a flame [21]. Since the moving hot-particle also acts as the pilot source as seen before, a successful ignition requires an ignition delay shorter than the residence time of the particle
tmix ∼
a still sphere [25,26] can be adopted,
Figure 8(a) illustrates the particle size dependences on the mixing time by natural convection (t ) and the embedding residence
4.1. Hot-particle ignition mechanism of EPS foam
tig = tpy + tmix + tchem ≤ tr
5
(3)
where δ is the thickness of boundary layer, and α is the diffusivity of the mixture. In general, tmix decreases with increasing temperature because of the increasing diffusivity. Assuming the flash occurs near the particle stops, the Nusselt number for natural convection around
time (tr ). As discussed above, both time scales increase with the particle diameter but at different rates, and ignition occurs when they match each other (tig = t = tr ). For example, a large particle (Dc ) mix, g
has both large mixing time and large residence time, so a relatively low particle temperature (blue line) is required for ignition (at blue solid point). As the particle size reduces from Dc to Dh , no ignition occurs (at blue empty point) because the residence time becomes shorter than the mixing time (tr < t ). To ignite the fuel, a higher mix, g
particle temperature is required to reduce the mixing time (red line). At the same time, it is observed in experiment that tr changes little with the temperature in the transition region, so ignition occurs (at red solid point). Therefore, in order to ignite the fuel, a higher particle temperature is required for a smaller particle, agreeing with the experimental results in Fig. 5. This concise theory provides a qualitative analysis for the experimental temperature-diameter ignition criteria in Fig. 5. A quantitative analysis for ignition condition requires an accurate estimate of mixing time, and can be achieved with a comprehensive numerical simulation in future work. As the temperature increases to Pig = 95% or slightly above, a rolling ignition often occurs before the particle stops (see Fig. 3). The rolling residence time decreases with the particle speed (tr ≈ D/u), as illustrated in Fig. 8(b). The video camera recorded that the average moving speed is u¯ ≈ 0.3 m/s, leading to an average rolling residence time of t¯r ≈ 30 ms. The rolling of particle tends to slow down due to the friction. Meanwhile, the mixing time is controlled by the forced convection, so the Nusselt number [26,27] may be estimated as
Nu
D, u
= 2 + 0.6Re1/2 Pr1/3 D
(5)
where the Reynolds number increases with the speed (Re ∼ u). Therefore, the mixing time (t ) also decreases with the particle mix, u
speed. Calculations (see Table A1 in Appendix) estimate that t
mix, u
=
21 ms (D = 6 mm, 1030 °C) and 59 ms (D = 12 mm, 930 °C) at u = 0.3 m/s. Such the mixing time is comparable to t¯r ≈ 30 ms, and −0.8 . can be fitted to t ∼ u mix, g
Figure 8(b) illustrates the mixing time by forced convection (t ) and the rolling residence time (tr ) versus the particle speed. mix, u
As calculated above, both time scales decrease with the particle speed but at different rates. For a hotter particle (red line), the “rolling ignition” occurs when they match (tig,h = t = tr ) before the rolling mix, u
ceases. As the temperature decreases (blue line), the t
mix, u
increases,
which is higher than tr during the rolling. Then, near the full stop, the particle’s vertical embedding motion and the mixing by natural convection take over (t < t ), leading to an embedding mix, u
mix, g ,
ignition. This explains why rolling ignition tends to occur in the
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
ARTICLE IN PRESS
JID: CNF 6
Fig. 8. (a) Mixing time by natural convection (t
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
mix, g
) and embedding residence time (tr ) versus particle size; and (b) mixing time by forced convection (t
mix, u
) and rolling residence
time (tr ) versus particle speed. The red and blue lines represent a higher and a lower particle temperature, respectively; the solid and empty points represent ignition and no ignition, respectively.
Fig. 9. Comparison of the transition region of hot-particle ignition between EPS foam (18 kg/m3 ) and cellulose powder (338 kg/m3 ) [16] versus particle diameter and temperature.
higher-temperature part of transition region while embedding ignition occurs in the lower-temperature part.
Moreover, the transition ignition region of EPS foam was much narrower than that of cellulose powders in the whole range of particle sizes. One reason is that compared to commercial EPS foam samples, the fuel bed of cellulose powder packed in the laboratory (e.g., [9–16]) was less homogeneous, increasing the uncertainty in repeating runs. The other reason is that the cellulosic fuels can go through smouldering which can be ignited at a lower temperature or heat flux [7]. This smouldering ignition creates a shortcut to flaming ignition, especially under external air flow, as observed in [15,16], expanding the transition region (see Fig. 9). When the particle temperature is much higher than the 95% ignition probability point, the “rolling ignition” could generate a large fire, following the particle and attaching to the EPS foam surface, as occasionally observed in the experiment. Subsequently, the flames were able to spread over the sample and consume it entirely, defining a “point for ignition and fire spread”. Such fire event should require a much higher particle temperature than the current ignition (flaming) point, as illustrated in Fig. 9. For example, in the preliminary experiments, the particle with 16 mm diameter and 1133 °C temperature was obtained to result in the ignition and fire spread to the entire sample. So far, due to the limitation of furnace temperature and particle melting point, the point for ignition and fire spread is a reasonable prediction while has not been quantitatively reported for neither building insulation materials nor forest fuels, and thus deserve a separate study in future work. 5. Conclusions
4.2. Comparison to forest fuels Figure 9 compares the critical particle temperatures and diameters for igniting EPS foam (18 kg/m3 ) with those for cellulose powders (338 kg/m3 [16]). For all fuel types, a significantly higher particle temperature (∼1000 °C) is required for smaller particles. However, the high-density forest fuel beds can be ignited with much cooler particles (600 ∼ 700 °C) by large sizes (D > 6 mm) for two reasons: (1) the pyrolysis temperature of cellulose powders is around 350 °C [28], slightly lower than 450 °C for EPS [22,23]; and more importantly (2) cellulose powder has a density large enough to hold the hot particle and to enforce effective contact and heat transfer. Thus, even if the particle temperature is too low (around 700 °C) to initiate a pilot ignition, a large particle would still have sufficient energy and heat transfer into the cellulose powder, producing a flammable mixture and going through a spontaneous ignition. In other words, the ignition of high-density cellulose by a lower-temperature particle may be controlled by the pyrolysis time (tpy ), different from the ignition of the low-density EPS foam by the gas-phase mixing time (tmix ).
This paper presented an experimental study on the ignition of expandable polystyrene (EPS) foam by a hot metallic particle. The flaming ignition was observed on the foam surface before the particle was embedded or during its rolling process. The critical particle temperature for flaming ignition (50% probability) was found to decrease sensibly from 1030 to 935 °C as the particle diameter increases from 6 to 14 mm. In the ignition transition region, no fire spread or fuel burnout occurred after flame ignition, and whether ignition occurred or not made little difference in fuel mass-loss ratio. Tests also showed that the sample density and thickness had negligible influence on either the ignition limit or fuel mass-loss ratio. Theoretical analysis suggested that the hot particle acts as both heating and pilot sources, and ignition of EPS foam is controlled by the competition between the mixing time in the gas phase and the residence time of the particle. Therefore, the critical particle temperature for EPS ignition was much higher, accompanied by a narrower transition region, and less dependent on the particle size, differing from higher-density forest fuel beds found in the literature. This is the first study of the hot-particle ignition mechanism of building insulation
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017
ARTICLE IN PRESS
JID: CNF
[m5G;September 15, 2015;21:23]
S. Wang et al. / Combustion and Flame 000 (2015) 1–7
7
Table A1 Exemplary calculations of the mixing time (1) by natural convection for different particle sizes in the “embedding” ignition and (2) by the forced convection for different particle speeds in the “rolling” ignition at room temperature of 24 °C where RaD = gβ(Tp − T∞ )D3 /αν and β = 2/(Tp + T∞ ). “Embedding” ignition (u = 0 m/s)
Thermal properties Tp
ν × 10−6
α × 10−6
Pr
D
Ra
Nu
(°C)
(m2 /s)
(m2 /s)
(−)
(mm)
(−)
1030
84.9
120
0.71
109
0.70
6 8 10 12 14 6 8 10 12 14
261 620 1210 2092 3321 307 729 1424 2460 3906
930
7.64
t
(−)
(mm)
(ms)
3.83 4.27 4.68 5.07 5.45 3.90 4.36 4.79 5.20 5.59
1.57 1.87 2.14 2.37 2.57 1.54 1.84 2.09 2.31 2.51
29 41 54 66 78 31 44 57 70 82
D, g
materials, helping to further understand the fire hazard of other lowdensity fuels. Acknowledgments This work was sponsored by National Basic Research Program of China (973 Program, NO. 2012CB719702) and National Natural Science Foundation of China (51576184 and 51120165001). H.C. was supported by Fundamental Research Funds for the Central University (No. WK2320000020). All authors would like to thank Prof. Forman Williams (UC San Diego) for valuable comments. Appendix The Biot number for the largest particle with 14 mm diameter [20] is estimated as:
“Rolling” ignition (u > 0 m/s)
δ
400 W/ m2 ×K × 13 × 0.007[mm] hL ∼ Bi = = 0.02 < 0.1 k 51.9[W/(m × K)] References [1] D. Drysdale, An Introduction To Fire Dynamics, third ed., John Wiley & Sons Ltd, Chichester, 2011. [2] V. Babrauskas, Ignition Handbook: Principles and Applications to Fire Safety Engineering, Fire Investigation, Risk Management and Forensic Science, Fire Science Publishers, Issaquah, WA, 2003. [3] J. Sun, L. Hu, Y. Zhang , Theor. Appl. Mech. Lett. 4 (2013) 042001. [4] J. Song, S. Wang, H. Chen, Theor. Appl. Mech. Lett. 4 (2014) 11–034005.
mix, g
D
u
Re
Nu
δ
t
(mm)
(m/s)
(−)
(−)
(mm)
(ms)
0.2 0.3 0.4 0.5 0.6 0.2 0.3 0.4 0.5 0.6
77 116 155 193 232 31 47 63 79 94
9.8 11.6 13.1 14.4 15.6 4.99 5.66 6.23 6.73 7.18
0.61 0.52 0.46 0.42 0.39 2.41 2.12 1.93 1.78 1.67
26 21 18 16 14 76 59 49 42 37
6
12
D, u
mix, u
[5] Z. Lu, W. Chen, Fire Safety Sci. 18 (2009) 15–18 (in Chinese). [6] R. Bynum, R.T. Bynum, Insulation Handbook, McGraw Hill Professional, 2001, p. 512. [7] G. Rein, Smouldering Combustion, chapter 19, in: SFPE Handbook of Fire Protection Engineering, 5th ed., Springer, 2015 (in press). [8] S. Wang, H. Chen, N. Liu, J. Hazard. Mater. 283 (2015) 536–543. [9] T. Tanaka, Rep. Natl. Res. Inst. Police Sci. 30 (1977) 151–158. [10] A.D. Stokes, J. Electr. Electron. Eng. 10 (1990) 175–187. [11] A.D. Stokes, J. Electr. Electron. Eng. 10 (1990) 188–194. [12] G.W.G. Rowntree, A.D. Stokes, J. Electr. Electron. Eng. 14 (1994) 117–123. [13] R.M. Hadden, S. Scott, C. Lautenberger, A.C. Fernandez-Pello, Fire Technol. 47 (2011) 341–355. [14] J.L. Urban, C.D. Zak, C. Fernandez-Pello, 8th US National Combustion Meeting, Canyons resort in Park City, Utah 4 (2013) 3673–3680. [15] C.D. Zak, J.L. Urban, C. Fernandez-Pello, Combust. Sci. Technol. 186 (2014) 1618– 1631. [16] J.L. Urban, C.D. Zak, C. Fernandez-Pello, P. Combust. Inst. 35 (2015) 2707–2714. [17] S.L. Manzello, T.G. Cleary, J.R. Shields, J.C. Yang, Int. J. Wildland Fire 15 (2006) 427– 431. [18] S.L. Manzellon, T.G. Cleary, J.R. Shields, J.C. Yang, Fire Mater 30 (2006) 77–87. [19] P. Yin, N. Liu, H. Chen, J. Lozano, Y. Shan, Fire Technol. 50 (2014) 79–91. [20] F.P. Incropera, D.P. DeWitt, T.L. Bergman, A.S. Lavine, Fundamentals of Heat and Mass Transfer, 7th ed., John Wiley & Sons, New York, 2011. [21] J.G. Quintiere, Fundamentals of Fire Phenomena, John Wiley & Sons Ltd, 2006. [22] P. Kannan, J.J. Biernacki, D.P. Visco Jr, W. Lambert, J. Anal. Appl. Pyrol. 84 (2009) 139–144. [23] J. Li, S.I. Stoliarov, Combust. Flame 160 (2013) 1287–1297. [24] D. Rich, C. Lautenberger, J.L. Torero, J.G. Quintiere, C. Fernandez-Pello, P. Combust. Inst. 31 (2007) 2653–2660. [25] S. Churchill, Heat Exchanger Design Handbook, in: K.J. Bell (Ed.), Hemisphere Publishing, NY, 1983. [26] F.P. Incropera, Introduction to Heat Transfer, 6th ed., John Wiley & Sons, New York, 2011. [27] W. Ranz, W. Marshall, Chem. Eng. Prog. 48 (1952) 141–146. [28] H. Yang, R. Yan, H. Chen, D.H. Lee, C. Zheng, Fuel 86 (2007) 1781–1788.
Please cite this article as: S. Wang et al., Ignition of low-density expandable polystyrene foam by a hot particle, Combustion and Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2015.08.017