Influence of Angle Orientation on Firebrand Production from the Combustion of Surrogate Photovoltaic (PV) Panel Assemblies Exposed to Applied Wind Fields

Abstract

A shared feature in the rapid spread of large outdoor fires are the production or generation of new, far smaller combustible fragments from the original fire source referred to as firebrands. A simplified experimental protocol has been developed that allows for the study of firebrand generation processes from various structural materials exposed to an applied wind field. The influence of angle of orientation on the firebrand production process is investigated. The thickest firebrands were produced with experiments with 45° angle under 8 m/s. The influence of angle was found to have the same trend under the tested wind speeds and to be more apparent at 8 m/s than 6 m/s. As installation angles are a key factor for photovoltaic panel (PV) efficiency, often only the solar energy efficiency is considered in PV panel orientation decisions. Yet, this study demonstrates that the types of firebrands generated in the event of large outdoor fires were sensitive to the angle of installation for structural materials used as surrogates for PV panels. The work is unique in that is begins the discussion on firebrand production from cutting edge home technologies, such as PV panels. These results have implications for how installation angles may influence firebrand production in the event of large outdoor fire outbreaks.

1. Introduction

Across the globe, there exist many recent examples of large outdoor fires [1]. Perhaps the most common are wildland fires that approach urban areas. These are often called Wildland-Urban Interface (WUI) fires [2]. Two recent examples are the 2019 WUI fires that occurred in South Korea and those in 2018 in Northern California in the United States [3, 4]. At the writing of this paper, WUI fires are also raging in California once again. In countries that are less developed, there have been large fires that have occurred in informal settlements. Some recent examples are those in the Philippines as well as South Africa, both in 2017 [5]. There have also been large urban fires in Japan, a country with no large wildland fire problem or informal settlement situation. Such urban fires have been recorded for centuries in Japan’s history [6].

Large outdoor fires spread via direct flame contact, thermal radiation or firebrands [7]. A shared feature in the rapid spread of large outdoor fires is the production or generation of new, far smaller combustible fragments from the original fire source referred to as firebrands [8, 9]. As communities become involved in large outdoor fires, firebrands are generated from the combustion of various structural fuel types [7]. These include common building materials such as plywood, oriented strand board (OSB), and wood supporting members. A far less investigated aspect is the production of firebrands from cutting edge home technologies, such as photovoltaic panels or PV panels.

At first glance, the release of firebrands from these combustion processes appears straightforward but is in fact quite complex. As a result, there is a need to develop simplified experimental protocols to be able to investigate these combustion processes in a cost-effective manner. Such experiments have the possibility to aid in the development of computational models to understand these complex firebrand release processes from structural fuels. At the same time, such experiments may also lead to new insights into the physics of firebrand generation from various materials perhaps suggest new methods of material fabrication to lessen firebrand liberation processes from these combustion reactions [7, 10, 11].

An important factor that has not been investigated to the authors knowledge is related to the installation of PV panel assemblies on roofs and dangers in the events of large outdoor fires. While PV panel installations are predicated on the best angle to collect solar energy [12], it is not considered on how these installation angles itself may influence the production of firebrands in the event of large outdoor fire outbreaks.

There have been numerous documented fires for buildings equipped with PV panel installations all over the world [13]. Most of the prior work on PV panel fire safety has been focused on fires that occur inside buildings and how these fires effect the overall roofing assembly combustion due to PV panel installations or if new fires are caused from the electrical connections inherent in such systems [13–17]. While the focus of those investigations is different than the present work, they are still reviewed here for completeness.

In the USA, Underwriters Laboratories (UL) conducted numerous studies on how the installation of PV panel installations influences roofing assembly fire performance [16]. In their studies, they reported that these installations may greatly impact the severity of roofing assembly combustion. Due to the global use of PV panel installations and the overall lack of harmonized fire testing standards for these systems, the need for test methods has been reported in European countries [15]. Additionally, more fundamental investigations have looked at the mechanisms on how PV panels may increase roofing assembly combustion, with fire induced re-radiation underneath the PV panel assembly identified as an important aspect [13–14]. Furthermore, there are even more simplified studies using cone calorimeters to determine toxic gas production from the PV panel arrays [17]. These are fascinating investigations, but none are directed to understand firebrand production from PV panel installation itself in the event of large outdoor fires, such as disastrous WUI and urban fires seen all over the world.

The unique aspects of this investigation are the development of new experimental technique that affords the investigation of firebrand production from structural materials over various angles in a wind field. As a first step, simplified mock-ups of actual structural assemblies, made from inexpensive building materials, are used a surrogate for PV panels. As has been identified in the literature [16], the sheer number of PV panel assemblies in existence make it intractable to experiment on every type of material used world-wide. Since it has been suggested that a surrogate for PV panel experimentation is needed, the authors, as a first step, undertook experiments using common, combustible building materials to investigate firebrand production.

2. Experimental Description

All experiments used mock-ups of structural assemblies and were performed in a wind facility at the National Research Institute of Fire and Disaster (NRIFD). NRIFD’s wind facility has a 4 m diameter fan. The flow field was measured to be within ± 10 % over a cross-section of 2.0 m by 2.0 m. Experiments were performed within this cross-section.

Mock-ups used in this study were intended to be low cost surrogates for PV assemblies, constructed with OSB and wood studs, with the dimensions of 610 mm (W) x 610 mm (H) (Figs. 1 and 2). These assemblies were exposed to applied wind fields and the angle of the assembly with respect to the wind was varied to values of 25°, 45°, and 65 °. The angle was adjusted using a series of custom mounting frames fabricated for this study.

Figure 1.

Image of mock-up assemblies used for the experiments. The scale of assembly used is shown as is the wind direction.

Figure 2.

Schematic of mock-up assemblies used for the experiments. The angle was varied from 25° to 65°. For illustration purposes, an arbitrary is shown.

The details of experimental procedures were described in [11] and only a short description is provided here. It is important to understand that prior work was only presented for fixed angle orientation. Mock-up assemblies were ignited by a T-shaped propane gas burner with heat release rate (HRR) of 32 kW ± 10 % for 10 min. Ignition under no wind was found to be necessary for repeatable ignition conditions as this methodology provides constant flame contact area for the assembly ignition investigations. The total ignition time was selected to be 10 min based on preliminary experiments [11]. After the burner was turned off, a desired wind speed (6 m/s or 8 m/s) was applied. Experiments were stopped when the combustion of the assemblies was completed or the assemblies were not able to support themselves anymore. The firebrand collection methodology is the same as the one described in [10–11, 18–19] using the pans filled with water. Collected firebrands were dried at 104 °C. Dried firebrands were measured with a scale and pictures were taken for image analysis. The image analysis method follows the same procedure described in [10–11, 18]. In short, the easiest method to visualize the image analysis process is seen in Fig. 3, as firebrands are often of very thin thickness, so the largest projected area is of interest (Fig. 3). Images of specific shapes that have areas that may be simply calculated were used to determine the ability of the image analysis method to calculate the projected area [10–11,18]. The standard uncertainty in determining the projected area was ± 10 %. Repeat measurements of known calibration masses were measured by the balance which was used for the firebrand mass analysis. The standard uncertainty in the firebrand mass was approximately ± 1 %. In addition, the maximum thickness of a firebrand was measured with calipers. As shown in Fig. 3, firebrands may not always possess a uniform thickness. The maximum thickness for a given (measured) projected area was measured. The uncertainty in the thickness measurement of approximately ± 10 % was calculated from the repeated measurements of known objects as well as firebrands.

Figure 3.

Example of maximum firebrand projected area and maximum thickness determination.

3. Results and Discussions

3.1. Repeatability of firebrand data

Experiments were repeated under the same conditions and the characteristics of firebrands were compared in order to investigate the robustness of the experimental protocol [11]. Fig. 4 shows the firebrand data from two experiments under the same conditions (65° and 6 m/s). Fig. 4 focuses on firebrands with the projected area smaller than 2500 mm². Firebrand characteristics from repeated experiments are in a good agreement.

Figure 4.

Repeatability of firebrand data. Data at an angle of 65° is shown.

3.2. Comparison with the projected area and mass of firebrands under different attack-of-angle

Figs. 5 displays a comparison of the firebrand size and mass distribution collected from these experiments. To more clearly show the results, linear curve fits are applied to the data. As may be seen, the angle of orientation had very little influence on the size and mass of released firebrands for an applied wind speeds of 6 m/s. As the wind speed was increased, for a given projected area, the mass of firebrands generated was sensitive to the angle of orientation.

Figure 5.

The mass and the projected area of firebrands collected under 6 m/s (top) and 8 m/s (bottom). Data is shown for mock-up assemblies oriented at three different angles with respect to the applied wind speeds.

It is known that the wind force plays an important role in releasing firebrands from the combustion of fuels. In its most simplistic representation, the wind force is proportional to the square of the applied wind speed. A more quantitative description for the wind force (pressure) to a structural element, p_wall, is described as follows [20];

$$p_{wall}=\frac{1}{2}{m}_{air}\times E\times V^2\times {\text{C}}_p$$ (1)

Here m_air is the mass of the air, E is the exposure coefficient, C_p is the wind pressure coefficient, which consists of drag and lift coefficients, and V is the wind speed. C_p depends on many parameters including the angle of attack [21]. E depends on the structure shape or surface roughness, E can be assumed to be constant for this experimental series [21]. The drag coefficient increases with the angle of attack, with a peak around 25° and 40°, then begins to decrease again. The lift coefficient increases with the angle of attack [21–23]. These interesting force relationships suggest that the angle of the mock-up should influence the physical characteristics of the firebrands.

As the projected area of the released firebrands demonstrated a linear relationship, the values of the slopes presented in Figs. 5 are summarized in Fig. 6. The error bars represent R². The mass of a given firebrand may be expressed as follows:

$$m_F={\rho }_F{d}_{ave}{A}_{proj}$$ (2)

$$m_F/A_{proj}={\rho }_F{d}_{ave}=CF{d}_{max}$$ (3)

where m_F is the firebrand mass, ρ_F is the firebrand density, d_ave is the average firebrand thickness, A_proj is the projected area, d_max is the maximum firebrand thickness, and CF is the correction factor. As all the firebrands are released from the same materials in these experiments, it is reasonable to assume the firebrand densities are similar. The density of firebrands from OSB material was estimated to 60 kg/m³ for this study. This estimation assumes the non-combusted OSB density would be reduced by 10 times, which is within the range of other firebrands from various woody materials [7]. Therefore, the largest slope indicates firebrands with the largest average thickness, suggesting that the thickest firebrands were produced at angles of 45° for applied wind speeds of 8 m/s.

Figure 6.

Effect of wind speeds on characteristics of firebrands generated from mock-up assemblies.

3.3. Projected Area and Thickness

To verify these assumptions, the maximum thickness of firebrands was measured and plotted versus mass and is shown in Fig. 7 and Table 1. For ease in visualizing the data, a curve-fit was applied for Fig. 7 and is y = a log(x) + b, where y is the maximum thickness of firebrands, d_max, and x is the mass of firebrands, m_F. As seen in Table 1 and Fig. 8, the relationship is not strong as demonstrated in the R² values but the increase firebrand thickness is apparent at 8 m/s. The influence of angle is observed at 6 m/s and 8 m/s.

Figure 7.

The maximum thickness as a function of firebrand mass collected under 6 m/s (top) and 8 m/s (bottom).

Table 1.

Numerical values obtained from curve-fitting data in Fig. 7.

Wind Speed	Angle	a	b	R2
6 m/s	25°	1.81	5.56	0.461
	45°	2.34	6.63	0.549
	65°	1.85	5.70	0.470
8 m/s	25°	1.78	5.78	0.336
	45°	3.91	9.31	0.803
	65°	3.04	7.54	0.669

Figure 8.

Values obtained from curve-fits applied to the data in Fig. 7.

3.4. Area Density and Thickness

The area density of firebrands, m_F/A_proj, was also determined as a function of the maximum thickness. These results are plotted in Fig. 9. Fig. 10 shows the relationship between the angle and the value of the slope in Fig. 9 (see Eq. (3)). The error bars are calculated based on R². Again, assuming the density of firebrands is the same for all conditions, 60 kg/m³, the ratio is plotted in Fig. 10 which shows the ratio of average thickness of a firebrand versus the maximum thickness of a firebrand. In average, the maximum thickness is 3 times greater than the average thickness. This relationship is quite useful when modeling firebrand transport, as one of the key elements in modeling firebrands is how to define the firebrand shape or effective size.

Figure 9.

The area density and the maximum thickness of firebrands collected under 6 m/s (top) and 8 m/s (bottom)

Figure 10.

The value of the slope obtained from Fig. 9 (top) and ratio was taken (bottom).

4. Summary

A simplified experimental protocol has been developed that allows for the study of firebrand generation processes from various materials exposed to an applied wind field. Experiments were performed at 6 m/s and 8 m/s with three angles considered, 25°, 45° and 65°. At both wind speeds investigated in this study, the angle of orientation influenced the thickness of the released firebrands. The influence was more apparent at 8 m/s than at 6 m/s.

While current PV installations fitted to roofs are predicated on the best angle to collect solar energy, it is not considered on how the installation angle may influence the production of firebrands in the event of large outdoor fire outbreaks. As has been identified in the literature, the vast number of PV panel assemblies in existence make it intractable to experiment on every type of material used in various countries. Since it has been suggested that a surrogate for PV panel experimental is needed, the authors, as a first step, undertook experiments using common, combustible building to investigate firebrand production.

In this study, the installation angle had an influence on the characteristics of firebrands that are generated using surrogate combustible building materials. As part of future work, experiments using more costly materials that may be more representative of the combustible materials found in PV panel assemblies should be undertaken, as the results in this study indicated angle of orientation is important in firebrand production.