Investigating the Effect of Structure to Structure Separation Distance on Firebrand Accumulation

Wind plays an important role in the built environment. Large outdoor fires in the built environment are no exception. Under strong wind, firebrands fly far, which leads to quick fire spread. In this study, the effect of structure to structure separation distance on firebrand accumulation was investigated by using a custom designed firebrand generator installed in a real scale wind tunnel. Firebrands accumulated at 4 and 6 m s−1, but no firebrand accumulation zone was observed at 8 and 10 m s−1, regardless of separation distance (SD). Experimental results were compared with a simple CFD flow simulation (no firebrands included). The size of firebrand accumulation zone as well as distance from the structure front was compared with SD in the cases of 4 and 6 m s−1 wind speeds. It was found that firebrands behave differently from SD = 1 to 2 m, to that of SD = 2 to 3 m. The results of this study are the first to explore these important interactions between firebrands and structure separation distances. The results of this work will help develop and design sustainable communities that may better resist the destruction of increasing large outdoor fire outbreaks worldwide, as well as help develop the next generation of CFD models needed to grasp the important large outdoor fire problem and associated firebrand processes.


INTRODUCTION
Wind plays an important role in the built environment, from ventilation within buildings, wind force on buildings, effects on rain, to pollution dispersion (Shah and Ferziger, 1997;Blocken and Carmeliet, 2004;Quyang et al., 2006;Chaves et al., 2011;Yuan and Ng, 2012;Razak et al., 2013). The importance of wind applies to fire safety, where an increasing number of large outdoor fires, such as urban fires, informal settlement fires, wildland-urban interface (WUI) fires and wildland fires, are a major concern (Manzello et al., 2018).
One of important measures in areas prone to large outdoor fires are home ignition zones (HIZ) (Cohen, 2000;Syphard et al., 2012;Biswas et al., 2013). It is important to remove all the combustibles including sheds or mulches, in case of any combustibles being ignited by firebrands, which will lead to ignition of homes. Unfortunately, it is difficult to implement in practice (Mell and Maranghides, 2009). More understanding on firebrand behavior around structures is needed. To this end, experiments were performed to investigate the firebrand behavior around structures, with a special emphasis on the separation distance between structures.

EXPERIMENTS Experimental Description
The continuous-feed firebrand generator (NIST Dragon) was used for all the experiments. The details of this apparatus are described in (Manzello and Suzuki, 2014), so a brief overview is provided here. The continuous-feed NIST Dragon was made of a continuous-feed part connected to the firebrand generator ( Figure 1). The continuous-feed part has the storage of wood pieces (for firebrands) connected with a pipe to the NIST Dragon, with two gates to mitigate fire spread from the apparatus to the feeding system. The firebrand generator has a blower that was set to 3 m s −1 at the exit of firebrand generator in order to loft the generated firebrands. This blower velocity is selected to be able to produce smoldering firebrands. The feeding rate was 800 g/min,  Structures were symmetrically placed at 7.5 m downwind (to the leading edge of structures) from the NIST Dragon (Suzuki and Manzello, 2017a). This distance was far enough to investigate the firebrand behavior around a wall (Suzuki and Manzello, 2017a). The dimensions of the structures were 1, 1.5 or 2 m wide (shown as α in Figure 2) × 2.44 m long × 2.44 m high depending on the selected separation distance (SD) either 3, 2, and 1 m (shown as β in Figure 2). The SD considered in this study were 1, 2, and 3 m as the SD of approximately 2 m is allowed in USA (Maranghides and Johnsson, 2008). Schematics of experimental settings are provided in Figure 2A feeding time of 10 min was selected for most cases as it was reported that (Suzuki and Manzello, 2017a) the firebrand deposition reached a peak and remain the same after a certain time. For 10 m s −1 wind speed cases with SD 1 and 2 m, 5 min feeding time was selected for safety, since it was  Frontiers in Mechanical Engineering | www.frontiersin.org January 2021 | Volume 6 | Article 628510 possible to produce outside fires due to the large effluent of firebrands outside the wind facility. Figure 3 shows the images of firebrand deposition process in the experiments under 4 m s −1 and SD 1 m. The firebrand accumulation zone was not completely symmetric. Separate experiments were performed to verify the characteristics of firebrands produced under different wind speeds, specifically, 4, 6, 8 and 10 m s −1 . Firebrands were produced under a desired wind and collected in water pans. After the collection, firebrands were dried at 104°C in the oven. The mass of each firebrand was measured with a scale, and a picture of each firebrand was taken. The image analysis was performed to measure the projected area. Characteristics of firebrands produced under different wind speeds is shown in Figure 4.

Experimental Results and Discussion
Experiments showed a clear firebrand accumulation zone in front of the separation zone (upwind side) at lower wind speeds, 4 and 6 m s −1 while no accumulation zones were observed at higher wind speeds, 8 and 10 m s −1 (Figure 5). In Figure 5, no images are provided for experiments with 8 or 10 m s −1 , due to the absence of accumulation zones. Figure 5 shows that the firebrand accumulation zone was not completely symmetric. In the case of SD 3 m under 4 m s −1 , the firebrand accumulation was observed between structures, also on the downwind side ( Figure 5E). Under 6 m s −1 wind with SD 1 m, firebrands accumulated into two zones ( Figure 5D). Under 8 m s −1 wind with SD 1 m, a small number of firebrands showed a tendency to accumulate, rolling on the floor together, however those accumulations were not sustained at the end. This behavior is shown in Figure 6. The distance from the front of the structure (D f ), the maximum length (L), width (W) of firebrand accumulated area, and the total area of firebrand accumulation were measured after each experiment (Figure 7). The accumulation distance in front of a structure is plotted against SD in Figure 8A, as well as the maximum length and width of firebrand accumulation zone in Figures 8B and 8C, respectively. Figure 8A showed that the D f had a peak at SD 2 m for experiments at 4 m s −1 , while the effect of SD is less clear in experiments at 6 m s −1 , due to two firebrand accumulation zones observed at SD 1 m. In Figure 8B, the length of firebrand accumulation zone was longer at 4 than 6 m s −1 .
The length of the accumulation zone varies depending on SD, and that behavior for experiments at 4 m s −1 was the opposite to those at 6 m s −1 wind speed. The similarity between the two was a peak at 2 m SD, which was the same as the case for D f under 4 m s −1 wind shown in Figure 8A. Figure 8C shows that the width of firebrand accumulation zone was wider at 4 than 6 m s −1 , which is the same for length shown in Figure 8B. The total area of the accumulation zone was plotted in Figure 8D. It was observed that the higher wind speed led to a smaller accumulation zone. This is similar behavior to very simple wall experiments (Suzuki and Manzello, 2017a). Under 4 m s −1 wind speed, the size of the firebrand accumulation zone was similar, regardless of SD, yet it changed significantly under a 6 m s −1 wind. This, in some sense, is similar to the accumulation distance, as it indicated that wind profile changes more significantly between SD 1 and 2 m than between 2 and 3 m. This effect needs to be more carefully investigated as the minimum SD between buildings is 2 m. The firebrand accumulation behind the separation (downstream)

SIMULATIONS
The Fire Dynamic Simulator (FDS) was used to simulate the wind field of these experiments (McGrattan et al., 2013). FDS does not contain firebrands; however, it was used to describe the firebrand accumulation behavior based on predicted wind profiles (Suzuki and Manzello, 2017a). Firebrands follow the wind as they are  The mesh size was 10 cm × 10 cm × 10 cm. The conditions of simulation are summarized in Table 1. After 10 s, modeling shows that the wind has stabilized. The simulation was performed up to 20 s for all cases in order to observe the repeated wind behavior. In a simple calculation performed in (Suzuki and Manzello, 2017a), the wind speed of 2.3 m s −1 was the lowest wind speed for firebrands to move on the floor where the surface was relatively smooth considering the balance between friction force between a firebrand and the floor (gypsum board) and the wind force: And Therefore, μ is the friction coefficient between a gypsum board and smoldering firebrands, m firebrand is the average mass of a firebrand, ρ air is the density of the air, g is gravitation acceleration, v is wind speed on a firebrand and A is the average projected area of a firebrand. As seen in Figure 4, the mass and the size of firebrands under 4, 6, 8 and 10 m s −1 can be considered within the range of uncertainties (±10%), average of mass (0.05 g) and projected area (0.78 cm 2 ) was used for calculation. The   unknown parameter was μ, as there is no data available for a smoldering firebrand and a gypsum board. Therefore μ 0.5 was used based on data for wood on wood (Japanese Society of Mechanical Engineers, 2004). Applying 2.3 m s −1 for the lowest wind speed for firebrands to move, we consider the stagnation zone (wind speed less than 2.3 m s −1 ) as the firebrand accumulation zone. An FDS simulation was performed using the conditions shown in Table 1. The FWTRF was simulated. Figures 9 and  10 show the FDS simulation results under 4 and 8 m s −1 in the cases of SD 1 m and 3 m. Figures 9 and 10 shows the wind speed 2.3 m s −1 being marked in black. As shown in Figure 9, as the wind speed increases, the distance to the stagnation zone from the front of structures, D f , decreases, and as SD become larger, wind between structures become less turbulent. Figure 11 shows the distance to the stagnation zone from the front of structures, D f , vs. time (10-20 s) in the case of 1 m SD for example. Overall, the D f increases as the wind speed increases. Data from 4 m s −1 is relatively stable, while those from 10 m s −1 is changing a lot. Those from 6 m s −1 are fluctuating. It is assumed the two firebrand accumulation zones observed in experiments are due to this behavior. Figure 12 shows the 10 s average, D f, against wind speeds along with experimental results. Results from FDS shows that D f increases as the wind speed increases. D f also increases as SD increases with the exception of SD 1 and 2 m under 6 m s −1 wind. This is interesting as experiments showed the wind profile around the structures may change between SD from 1 to 2 m, and SD from 2 to 3 m.
Experimental data with SD 1 m has similar D f to FDS results while experimental results does not match well when SD 2 or 3 m. For the same SD, as the wind speed increases the wind flow around structures becomes complicated, and it is expected firebrands may not follow the wind completely; the calculation deduced from simple analysis no longer works.

SUMMARY
Wind has an impact on the built environment. For large outdoor fire safety, wind plays a critical role, from enhancing flame spread processes to firebrand spotting. Firebrands fly far under high wind, which presents risks to many houses. This study focused on understanding firebrand behavior between two structures by varying the separation distances and wind speeds experimentally and by simulation. No significant accumulation was observed under 8 and 10 m s −1 , regardless of structure separation distance. The size of firebrand accumulation as well as distance from structure front was compared with separation distance in the cases of 4 and 6 m s −1 wind speeds. It was found that firebrands behave differently from SD 1-2 m, to that of SD 2-3 m. The results of this study are the first to explore these important interactions between firebrands and structure separation distances.