Student Research Paper by Rishub Shah ([email protected])
Mentor: Dr. Franz Richter ([email protected]). Thanks for the opportunity: Dr. Michael Gollner ([email protected]).
December 2020
Abstract
Wildfires are a leading cause of property damage, injury and death throughout the world. A major way in which wildfires spread is through embers or firebrands. The transport and deposition of embers has been researched significantly, as has the subsequent combustion. However, the accumulation of embers has not. In this paper, we research the accumulation of embers on and around a typical house in the Wildland Urban Interface. We measure the accumulation of embers by varying the following variables one at a time: wind speed, ember mass flux and the roof friction factor. We find that as the wind speed increases, the ember accumulation reduces on open surfaces like the roof and the floor around the house, but it increases in enclosed areas like balconies and patios. This leads to the conclusion that roofs are at a higher risk of ignition at low wind speeds, and balconies/patios are at a higher risk of ignition at high wind speeds. It is not sufficient to fire-proof just the roof, we must also avoid combustibles on the deck and balconies.
We find that as the mass flux of embers increases, the accumulation also increases everywhere on and around the house. Finally, we find that as friction factor decreases, the accumulation also decreases, especially if the wind is blowing parallel to the roof. This might indicate the need to build roofs from smoother materials like commercial steel or concrete tiles.
Introduction
Wildfires are a big problem in the United States and the world leading to large losses of life and property. In California, wildfires have become an annual event. California wildfires caused $85 billion in 2017, $400 billion in 2018 and about $80 billion in 2019 [1]. 2020 is turning out to be the worst year so far for wildfires, with 4 million acres already burned. The 2019–2020 Australian bushfire season, scorched an estimated 46 million acres, destroyed 5900 buildings and killed 34 people [2]. This gives an idea of the scale of destruction that occurs due to wildfires and why researching various aspects of wildfire prevention is important.
A wildland urban interface (WUI) is an area where man-made infrastructure (e.g., cell towers, schools, water supply facilities, etc.) is in or adjacent to areas prone to wildfire [3]. Wildfire prevention research focuses on the WUI as opposed to uninhabited areas of forests because this has the biggest impact in preventing loss of life and property. One important aspect of how fires spread in a WUI is through embers or firebrands. An ember is a small, glowing piece of heated wood, coal or other material that remains after (or sometimes precedes) a fire. Embers can glow as hot as the fire from which they arise, and are light enough to be carried by the wind for long distances without being extinguished. Many studies show that embers are a leading cause of wildfire spread [4].
A lot of research has been carried out into the generation and transport of embers and the subsequent ignition induced by them [5]. However, there is little research done into the accumulation of firebrands on and around houses lying in the WUI. This is the topic of our research. Specifically, we are trying to study ember accumulation on and around houses in the WUI with varying wind speeds, varying ember mass and varying friction factor of the house roof. This can help us understand the risk of accumulation by these various factors and can inform how we can protect structures from igniting due to this ember accumulation. This will in turn help reduce the spread of the wildfire.
Method and Model
For this research we used Pyrosim and Fire Dynamics Simulator (FDS). Pyrosim is a software built by Thunderhead Engineering which allows users to graphically build FDS models that can simulate fire, flow and other scenarios.
For our research, we focused on a typical house in the WUI. We imported a CAD version of the house into a Pyrosim model. Figure 1 shows what the house looks like. It has many features that lend themselves to accumulation of embers: the roof, the awnings, the porch, the balconies and so on. Embers are introduced into the model from the right of the house and they flow in the negative X direction for all simulations. The details of how the embers were modeled in Pyrosim are included in Appendix 1.
Figure 1: Model of the house used for all simulations. The house has roofs, awnings, porch and balconies that lend themselves to ember accumulation. Wind blows from the right of the house in the negative X direction for all simulations.
Varying Wind Speed
For the first set of simulations, we varied the wind speed. We used velocities of 5 mps, 15 mps, 20 mps, 25 mps and 30 mps. The range of 5 mps to 30 mps is similar to that used in other studies. Quarles et al used 30 mph or about 13.4 mps in their study on embers landing on decks [6]. A review paper on the role of embers in fire spread published by Dr. Manzello and other scientists in 2020 quotes several studies that use wind speed in the range of 7-20 mps [5].
Many studies show that as wind speed increases, the average mass of the embers transported increases as well. In Manzello’s review paper “Role of firebrand combustion in large outdoor fire spread”, 2 data sets for how mass varies by speed are provided [5]. We used these 2 data sets to fit a line to derive the relationship between the mass and the embers. These graphs are shown below.
Figure 2: A graph showing the average mass of embers transported against wind speed. Data Set 1 from a review paper by Dr. Manzello and other scientists published in 2020 [5]. The data shows how the average mass of embers transported increases as wind speed increases.
Figure 3: A graph showing the average mass of embers transported against wind speed. Data Set 2 from a review paper by Dr. Manzello and other scientists published in 2020 [5]. The data shows how the average mass of embers transported increases as wind speed increases.
Using the correlations, we calculated the average mass of embers transported for each of the wind speeds in our simulations. In the simulation, we introduce 2000 embers per second into the model. The following table shows the calculation of how much mass is introduced per second for each of the velocities.
Table 1: For each wind speed calculate average mass from each of the correlations in Figure 2 and Figure 3. Take the average of the two to get average mass per ember. Multiply by the number of embers transported per second and divide by 1000 to get Mass Flux in Kg/s to use for each simulation
Wind Speed for our simulation |
Mass from Correlation 1 (gms) |
Mass from Correlation 2 (gms) |
Average Mass (gms) |
Number of embers transported per second |
Mass Flux Kg/s |
5 |
0.0106 |
0.0096 |
0.0101 |
2000 |
0.0202 |
10 |
0.0326 |
0.0506 |
0.0416 |
2000 |
0.0832 |
15 |
0.0546 |
0.0916 |
0.0731 |
2000 |
0.1462 |
20 |
0.0766 |
0.1326 |
0.1046 |
2000 |
0.2092 |
25 |
0.0986 |
0.1736 |
0.1361 |
2000 |
0.2722 |
30 |
0.1206 |
0.2146 |
0.1676 |
2000 |
0.3352 |
Each simulation runs for a total of 60 seconds, the embers are injected every 0.5 seconds for 55 seconds. At the end of the simulation the screenshot showing the mass per unit volume (MPUV) of the embers at various places in the simulation is captured as well as quantitative data showing the MPUV of the embers at various parts of the model are captured. How this was done in Pyrosim is covered in Appendix 1.
Varying Mass
For the next set of simulations, we used 10 mps and 0.0832 Kg/s mass flux as the baseline and varied the average mass of the embers. We ran the simulation for the following mass flux (Kg/s) : 0.0823 (base case), 0.0416, 0.1646, 0.3328.
As earlier each simulation ran for 60 seconds and the same data was captured – MPUV visual images and quantitative data.
Varying Friction
For the final set of simulations, we used a wind speed of 10 mps and a mass flux of 0.0832 Kg/s and varied the roughness property of the roof’s surface to 0, 0.01, 0.05, 0.1, and 1. We again measured the accumulation after running for 60 seconds, just as with other simulations.
Table 2: List of all the simulations
Speed (m/s) |
Mass (kg/s) |
Friction Factor (Roughness in m) |
|
5 |
0.0202 |
0 |
Varying Speed |
10 |
0.0832 |
0 |
|
15 |
0.1462 |
0 |
|
20 |
0.2092 |
0 |
|
25 |
0.2722 |
0 |
|
30 |
0.3352 |
0 |
|
10 |
0.0832 |
0 |
Varying Mass |
10 |
0.0416 |
0 |
|
10 |
0.1664 |
0 |
|
10 |
0.3328 |
0 |
|
10 |
0.0832 |
0 |
Varying Friction |
10 |
0.0832 |
0.01 |
|
10 |
0.0832 |
0.05 |
|
10 |
0.0832 |
0.1 |
|
10 |
0.0832 |
1 |
Results
In this section, we present the results of the simulations of varying speed, varying mass and varying friction. In the next section, we draw some conclusions based on the results of the simulations.
Varying Speed
The following 6 images show the accumulations for the 6 different velocities. Note that the scales are different for each figure.
A: 5 mps |
B: 10 mps |
C: 15 mps |
D: 20 mps |
E: 25 mps |
F: 30 mps |
Figure 4: Visualizing mass per unit volume accumulation of embers for different velocities. Note that the scale is different for each speed.
As you can see from the images in Figure 4 there is a large accumulation of embers on the roof and awnings at 5 mps wind speed. There is also a large accumulation of embers around the house. At 10 mps, the accumulation on the roof is almost gone, but the awnings and surroundings of the house continue to have accumulation. Also the accumulation around the house is reduced. At 15 mps even the awnings and surrounding accumulation has reduced. This trend continues at higher speeds and at speeds above 20 mps, there is no accumulation on the roof or awnings and very less surrounding the house. However, at higher speeds the accumulation in the areas where embers can be trapped, like the balcony, has increased significantly. This can also be observed in Table 4 and graphs in Figures 5-7.
Table 3: Ember accumulation in mass per unit volume for balcony, roof and floor by wind speed
mps |
Floor (kg/m3) |
Balcony (kg/m3) |
Roof (kg/m3) |
5 |
0.073 |
0.004 |
0.103 |
10 |
0.074 |
0.012 |
0.004 |
15 |
0.043 |
0.022 |
0.001 |
20 |
0.034 |
0.031 |
0.001 |
25 |
0.025 |
0.042 |
0.001 |
30 |
0.018 |
0.052 |
0.000 |
Figure 5: Ember accumulation on roof by wind speed. Accumulation decreases with speed becoming near zero by 10 mps. |
Figure 6: Ember accumulation in balcony by wind speed. Accumulation increases linearly with wind speed. |
Figure 7: Ember accumulation on the floor around the house by wind speed. |
Varying Mass
The 4 images in Figure 8 show the accumulation for the different masses of embers introduced into the simulations. Note that the scale is different for each figure. The accumulation pattern is almost identical for the different masses, but the mass accumulated increases proportionately to the mass introduced in the simulation. The speed is 10 mps for all simulations.
The data in Table 5 and the graphs in Figures 10-12 also show that the accumulation increases almost linearly with the increase in mass flux.
A: Mass flux = 0.0416 Kg/s |
B: Mass flux = 0.0832 Kg/s |
C: Mass flux = 0.1664 Kg/s |
D: Mass flux = 0.3328 Kg/s |
Figure 8: Visualization of ember accumulation for different masses of embers introduced for same speed 10 mps. The accumulation pattern is identical but the mass keeps increasing, the scales are different for each image.
Table 4: Ember accumulation in mass per unit volume by mass flux at a wind speed of 10 mps.
Mass Flux Kg/s |
Floor Kg/m3 |
Balcony Kg/m3 |
Roof Kg/m3 |
0.0416 |
0.0377 |
0.0060 |
0.0023 |
0.0832 |
0.0739 |
0.0120 |
0.0042 |
0.1664 |
0.1496 |
0.0215 |
0.0104 |
0.3323 |
0.3007 |
0.0464 |
0.0164 |
Figure 9: Ember accumulation on roof by mass flux. Accumulation increases linearly with mass flux. |
Figure 10: Ember accumulation in balcony by mass flux. Accumulation increases linearly with mass flux. |
Figure 11: Ember accumulation on floor around house by mass flux. |
Varying Friction
When we ran the 5 simulations to vary the friction factor on the roof in our model, we noticed no real difference in accumulation from the base case. This does not necessarily mean that friction has no impact on the accumulation, but it could be the way the model is setup minimizes the impact of roughness or there is a modeling limitation in FDS. One limitation of FDS is that you cannot build a slanted plane (like a real-life roof), so we had to build a roof as a series of small steps. Figure 12 shows this by taking a section of the original model that includes the roof. Also, the embers are blowing perpendicular to the roof. As you can imagine, this setup causes friction to have little effect on the accumulation.
Figure 12: Section of the roof shows that in FDS a slanted plane is modeled as a series of steps. Embers blowing perpendicular to the roof. This may explain the impact of friction on accumulation is minimized in FDS.
In order to allow friction to have an impact on the accumulation and minimize the impact of the modeling limitations mentioned earlier, we ran two other sets of simulations.
- Instead of blowing the embers perpendicular to the roof, we blew the embers parallel to the roof and measured the accumulation. (Figure 14)
- We created a flat extension to the roof at the top and blew the embers perpendicular to the roof. (Figure 15).
Figure 13: Embers blowing parallel to the roof. Part of additional simulations ran to further understand the impact of Roughness on accumulation. |
Figure 14: Large Flat top roof, embers blowing perpendicular to roof. Part of additional simulations run to further understand the impact of Roughness on accumulation. |
In both cases, we blow the embers for 25 seconds at 3 mps, and then continue to blow the wind without any embers for another 35 seconds for a total of 60 seconds. Then, we measured the accumulation. We repeat this for various values of Roughness. The following graphs show the results.
Figure 15: Accumulation trend – Embers blowing parallel to the roof. Accumulation decreases linearly as the roughness decreases. |
Figure 16: Accumulation trend – Large Flat roof, embers blowing perpendicular to roof. Accumulation decreases linearly as roughness decreases, but the rate of decrease is not as large as when embers are blowing parallel to the roof. |
Table 5 shows the roughness coefficients of some real-life materials that can be used to put the graphs in Figures 15 and 16 in perspective.
Table 5: Roughness coefficient of some commercial materials.
Material |
Roughness (mm) |
Ordinary wood |
5 |
Well Planed Wood |
0.18 – 0.9 |
Ordinary Concrete |
0.3 – 1 |
Commercial Steel |
0.045 – 0.09 |
PVC, Plastics |
0.0015 – 0.007 |
Conclusion and Discussion
Our research focused on studying accumulation of embers on and around a typical house in the WUI. In order to do this, we ran fifteen simulations: six for Varying Speed, four for Varying Mass Flux and five for Varying Friction Factor. Since the results for Varying Friction Factor in our model were questionable (no change in accumulation for any friction factor change), we ran 13 additional simulations with simplified models of the roof. We presented the results of these simulations both as visualizations and numerically as tables and graphs in the previous section.
From our results, we can draw the following conclusions:
- As the wind speed increases, the accumulation of embers on open surfaces like the roof or areas around the house decreases. The accumulation on the roof drops to almost zero above speeds of 10 mps. The accumulation drop on the floor is linear.
- As the wind speed increases, the accumulation of embers in closed surfaces like a balcony increases linearly.
- As the mass flux of the embers increases, the accumulated mass of embers increases in direct proportion for a given speed.
- As the friction factor of the roof decreases, the accumulation on the roof decreases when the wind is blowing parallel to the roof. When the wind is blowing perpendicular to the roof, the accumulation also decreases with decrease in friction, but the rate of decrease is smaller compared to when the wind is blowing parallel to the roof.
The results (1 and 2) seem to show that the wind speed changes the areas of risk in the house. The roof might be higher risk at lower speeds, but the balcony and other closed areas become higher risks at higher speeds. It is likely that both the balcony and the roof are areas with high risk of ignition. Therefore, in addition to making our roofs fire-proof, we also need to avoid combustibles in closed areas, for instance, by avoiding having a sofa on a deck.
The friction factor has an impact on the accumulation, the smoother the roof the less the accumulation. This is especially true for wind blowing parallel to the roof. This might indicate the need to build roofs from smoother materials like commercial steel or concrete tiles. You\’ll still likely need roof repairs in Auckland but you won\’t have to spend more money to replace the whole roof.
Houses are not built in isolation at the WUI, so as a next research topic, we should investigate patterns of accumulation in a community of houses.
Wildfires cause billions of dollars in damage and lead to massive loss of life as mentioned at the beginning of the paper. Through our research, we hope to make at least some contribution in helping reduce the damage and preventing loss of life.
Appendix 1: Modeling Embers in Pyrosim
The first four sections of this appendix have been borrowed heavily from “Modeling Embers in FDS” Tutorial by Jack Wallis [7]. Wallis uses an Ember Supply to insert the embers, whereas this study uses an Ember Cloud backed by an Air Supply. This allows us to stop inserting the embers 5 seconds before the end of the simulation while maintaining the Air Supply to the end. With an Ember Supply, stopping the embers also stops the wind.
Meshing and Vents
Set up a simple mesh in Pyrosim (or FDS), define a vent on one wall and an exhaust (vent with OPEN condition) on the other. Follow this tutorial: https://support.thunderheadeng.com/files/AirMovementEx.pdf
Ember Material
Define a new material, ‘EMBER_MATL’. Density may be set arbitrarily to facilitate ember transport via wind, or somewhat realistically around 60-100 kilograms per cubic meter. Different thermal properties may be set, but for this example we will leave the rest of the values as defaults.
Figure 17: Defining Ember Material in Pyrosim
Ember Surface
Now define a new surface, ‘EMBER_SURF’ using the surface type ‘Layered.’ Under ‘Geometry’ select ‘Spherical’ and set the inner radius to 0. Under ‘Material Layers,’ input some radius for the particles under ‘Thickness’ and input ‘1.0 EMBER_MATL’ under ‘Material Composition.’ For this example we will use particles with a radius of 10 mm. Finally under ‘Surface Props’ set the ‘Initial Internal Temperature’ to some value for the temperature of the embers. For this example we will use 750°C.
Figure 18:Defining Ember Surface in Pyrosim.
Ember Particles
Now define a new solid particle, ‘EMBER_PART’ and select ‘EMBER_SURF’ as
its surface. Select ‘Particles Can Move.’ Particle age parameters can be changed in the ‘Injection’ Tab but we will leave as default.
Figure 19: Ember Particles in Pyrosim
Ember Particle Cloud
Define a new Particle Cloud in Pyrosim. Place the particle cloud near the open vent in your mesh, just inside the mesh. In a later step you will place an air supply behind the particle cloud to blow the embers in. Choose ‘EMBER_PART’ for the particle dropdown. For Droplet Counter select Constant and enter 1000. For insertion select Insert Periodically with an insertion interval of 0.5 s and mass of 1 kg/m-cube. We also deactivate the cloud 5 seconds before the end of the simulation – which in my case was 55 seconds.
Figure 20: Ember Particle Cloud in Pyrosim
Air Supply and Blower
Define an Air Supply as described in https://support.thunderheadeng.com/files/AirMovementEx.pdf. Set the speed to whatever speed you want the embers to be blown at.
Finally define an obstruction called “blower” right behind the “Ember Cloud” and use the Air Supply as the Surface for this obstruction.
Figure 21: Air Supply in Pyrosim
Figure 22: Blower Surface behind the Ember Particle Cloud. General Tab.
Figure 23: Blower Surface behind Ember Particle Cloud. Surfaces Tab.
Appendix 2: Visualizing and Measuring the Mass per Unit Volume (MPUV) of the Embers in Pyrosim
Visualizing Accumulation of Embers
In Pyrosim under the Output menu select Plot3D. This shows a dialog like the one below. Simply select the “[Particle: EMBER_PART] Mass Per Unit Volume” quantity. You can select others if you like but they are not needed. Also select an output interval that you want the plot3d data to be written out. When you run the simulation results in the Pyrosim Results Viewer you will be able to enable the Plot3D out that you selected for visualization. All the pictures of the simulation results shown in this paper use this technique to visualize the MPUV accumulation.
Figure 24: Enabling MPUV visualization in Pyrosim
Measuring the MPUV of the Embers
To measure the MPUV of the embers we have to add DEVICEs to the Pyrosim model. First you must decide the bounding box for which you want to measure the MPUV. In our simulations we measured the MPUV in a bounding box around the top roof and around the balcony. This was done to get an idea of how the accumulation varies on open surfaces as well as closed surfaces. You can find the coordinates of the bounding box by selecting the objects you want to bound and looking at the properties dialog. You might have to use some manual calculations.
Once you have the bounding box, select Devices – New Gas Phase Device. In the dialog that pops up, in the properties tab provide the name for your device. Select [Particle:EMBER_PART] Mass per Unit Volume in the Quantity dropdown.
Figure 25: Specifying Device to output the MPUV into a CSV file. Properties tab.
Switch to the Advanced Tab. Enter XB as the name and the bounding box coordinates in the Value. Also enter SPATIAL_STATISTIC as name in another row and ‘VOLUME_INTEGRAL’ as value. See example in the dialog below. Now when the simulation runs you will see a *devc* file in the output directory which will have values for the device you just created.
Figure 26: Specifying a device to output MPUV values into a CSV file. Advanced Tab.
Bibliography
[1] J. Roach, “The 2019 California wildfires caused less damage than the last two devastating seasons | AccuWeather.” https://www.accuweather.com/en/business/the-2019-california-wildfires-caused-less-damage-than-the-last-two-devastating-seasons/643455 (accessed Oct. 02, 2020).
[2] S. Fowler, “Six Months After Australia’s Wildfires, Recovery Continues.” https://www.directrelief.org/2020/06/six-months-after-australias-wildfires-recovery-continues/ (accessed Oct. 02, 2020).
[3] Fema.org, “Wildland urban interface (WUI).” https://www.usfa.fema.gov/wui/ (accessed Oct. 03, 2020).
[4] Frontlinewildfire, “Most Wildfire Damage Caused by Embers – Frontline.” https://www.frontlinewildfire.com/why-embers-cause-the-most-damage-in-wildfires/ (accessed Oct. 02, 2020).
[5] S. L. Manzello, S. Suzuki, M. J. Gollner, and A. C. Fernandez-Pello, “Role of firebrand combustion in large outdoor fire spread,” Prog. Energy Combust. Sci., vol. 76, p. 100801, 2020, doi: 10.1016/j.pecs.2019.100801.
[6] S. Quarles and C. Standohar-Alfano, “Ignition Potential of Decks Subjected to an Ember Exposure,” no. May, p. 35, 2018, [Online]. Available: http://disastersafety.org/wp-content/uploads/2017/10/Deck-Ember-Testing-Report-2017_IBHS.pdf.
[7] J. Wallis, “Modeling Embers in FDS,” p. 20.
About the author
Rishub is a Sophomore at Mission San Jose High School, in Fremont, CA. Rishub is passionate about coding and technology. He has been learning Python since fifth grade and also teaches coding to kids. In his spare time he likes to work on Robotics and CAD projects. He has participated in local and state level archery competitions and plays junior varsity water polo for his school.