Proposal of Standard Wildfire Curves for the Design Protection of Dwellings against Wildland Fire

Cantor, P.; Arruda, M. R. T.; Firmo, J.; Branco, F.

doi:10.1061/(ASCE)HZ.2153-5515.0000706

Open access

Technical Papers

May 12, 2022

Proposal of Standard Wildfire Curves for the Design Protection of Dwellings against Wildland Fire

Authors: P. Cantor https://orcid.org/0000-0002-5293-1371 [email protected], M. R. T. Arruda https://orcid.org/0000-0002-4140-2204 [email protected], J. Firmo [email protected], and F. Branco [email protected]Author Affiliations

Publication: Journal of Hazardous, Toxic, and Radioactive Waste

Volume 26, Issue 3

https://doi.org/10.1061/(ASCE)HZ.2153-5515.0000706

PDF

Abstract

This paper presents research that proposes a new standard fire curve to simulate exterior wildfire due to burning trees and bushes. An extensive inquiry was performed in the existing scientific literature, in the search of measured temperature versus time wildfire curves by other authors, in trees and bushes using in situ thermocouples. The gathered wildfire curves for both trees and bushes are plotted together, and envelope upper-bond wildfire design curves are proposed. The objective of these new proposed fire curves is for them to be used in the design of the protection of dwellings located in the urban wildland interface against the action of a natural wildfire. To access the efficiency of these proposed fire curves, a thermal numerical analysis is performed using the commercial finite-element software Abaqus, in which a wood specimen is subjected to real wildfire and compared with the proposed wildfire design temperature versus time curves.

Introduction

The safety of dwellings (and their residents) against a wildfire in the wildland–urban interface (WUI) is dependent on the quick action of firefighters, either on the urban settlement or in the forest field (Glickman and Babbitt 2001; Radeloff et al. 2005). The individual safety of the dwellings regarding passive protection (fireproof) against the action of a wildfire is rarely taken into account (Tenreiro 2018).

Modern society has generally considered two approaches to deal with the unavoidable existence of wildland fire (Davis 1979). The first approach is forest management, and the second approach is the suppression of wildfire. Despite the fact that both management and suppression tactics have existed for a long time, the ideal means of implementing these strategies, particularly in terms of management, has yet to be achieved (Stephens and Ruth 2005).

Even with all these problems and limitations, few countries possess any design or fireproof construction guidelines against wildfire action. This is mainly due to the nonexistence in codes of a standard wildfire temperature versus time curves, to provide fire resistance design in the exterior façade of a dwelling. Several engineers and architects have proposed some solutions for new construction guidelines in Australia (Weir and Davidson 2011) and the United States (Wright 2019). However, these solutions are based on fire-resistant wood and protected steel structures. Some general guidelines exist to allow nonflammable proposals to prevent ignition and destruction of buildings near the WUI (Ramsay and Rudolph 2003), but they are not 100% fireproof standard solutions.

The authors of this work also proposed new fireproof construction guidelines for dwellings against wildfire (Arruda et al. 2021), but these were based on an interior 60 min standard fire curve, which is more severe than exterior wildfire; for this reason, these may not be economically competitive. Thereby, a standard wildfire curve is needed to promote a more efficient fireproof construction guideline, in the future to encourage a competitive/economically structural solution during the design phase.

Objectives

The main objective of this work is to propose standard wildfire curves to be used in the design protection of dwellings located in the wildland–urban interface (WUI) against the hazard of a wildfire. To accomplish this, an extensive search was performed in the scientific literature concerning wildfire temperature versus time curves measured by other authors in trees and bushes using in situ thermocouples. These temperatures versus time curves were then plotted together and analyzed, resulting in two design wildfire curves proposed for fires in trees and bushes, using the equations found in the work of (Blagojević and Pešić 2011). These curves simulate the direct wildfire contact with the dwelling that will be used to analyze the fire resistance of the chosen construction materials.

Numerical thermal analysis is used to validate the proposed design temperature curves, in which the curves are compared with experimentally measured temperature curves and with some existing standard exterior fire curves. The commercial finite element software Abaqus 2018 standard is used to perform the thermal analysis.

Research Significance

According to extensive research in the scientific community, there is no current standard proposed wildfire temperature versus time curve to be used in the fire resistance design of dwellings. This work intends to fill this knowledge gap to promote wildfire safety protection for dwellings located in the WUI and, therefore, to endorse a future economically competitive fireproof construction guideline.

Existing Fire Curves

In this section, existing standard fire curves are presented and later compared with existing measured wildfire curves in situ, in order to understand if these are appropriate to design wildfire protection. The scientific interpretation will be based mainly on the temperature versus time curve during the time length of a wildfire.

Measured Wildfire Curves in situ

Several researchers have provided wildfire temperature versus time curves for trees [Fig. 1(a)] and bushes [Fig. 1(b)], measured in situ with thermocouples, which provide different behaviors. For example, fire in trees presents a maximum high temperature (1,100°C to 800°C) but a lower time duration (2 to 4 min) [Fig. 2(a)]. On the other hand, fire in bushes displayed a lower high maximum temperature (800°C to 500°C) but a wider time duration (5 to 10 min) Fig. 2(b). These fire curves were highly dependent on the type of fuel and atmospheric conditions, and can vary in surface temperature, emissivity, relative humidity, solar radiation, and three critical factors for fire propagation, precipitation speed, wind direction, and geography. Therefore, for design purposes, the most severe case must be adopted, even though these conditions might not be present near the dwelling.

Fig. 1. Types of wildfire in the field: (a) fire in trees; and (b) fire in bushes.
(Images courtesy of CERIS.)

Fig. 2. Temperature versus time curves for wildfire: (a) fire curves in trees; and (b) fire curves in bushes. The subdivisions in (b) are related to the number of temperature curves presented by the authors in the analysis.

There was also a small number of studies related to firebrand temperature versus time curves (Thomas et al. 2018) with even lower maximum temperature (300°C to 400°C), but these are presently of small scale, and more studies are necessary to be used in standard wildfire curves.

One study fully detailed the maximum temperature found/measured in a wildfire (Wotton et al. 2012), in which the fire propagates in dense, high, dry eucalyptus trees. Although the temperature was high, the fire’s duration was extremely short, and the thermocouples were located in the branch and trunk, whereas in reality the fire’s temperature field is never close to the dwelling. Therefore, despite these high mean temperature values, it is possible to use a conservative maximum temperature for design purposes.

The second study (Chetehouna et al. 2008) considers flame fronts generated in a laboratory fire tunnel using Quercus coccifera shrubs as vegetal fuel, simulating bush fire. In this case, the maximum temperature was lower than the previous study, but the time duration was wider, as discussed previously. A similar study was performed (Silvani and Morandini 2009) in which in situ four-fire spread experiments were conducted across various vegetative fuels, ranging from pine needle bed to shrub, to simulate bush fire.

A mixture of small trees and bushes was experimented with, in which a fire spread experiment was conducted in the field under wind-blown conditions (Morandini et al. 2006). The fuel consisted of tall and dense Mediterranean shrub vegetation, promoting different maximum temperatures and time duration. The temperature versus time of the prescribed fire was studied (Kennard et al. 2005), in which a longleaf pine forest was monitored, with most conclusions valid for bush fire or small trees. Another study regarding wildfire temperature in nondense mixed small tree types is presented by Santoni et al. (2006), with the same conclusions as Wotton et al. (2012), but with an even smaller time fire duration. This study was significant because the in situ conditions were selected in this area to simultaneously take advantage of some of those conditions for which high-intensity wildfires occurred.

Standard Fire Curves

The most popularly used temperature versus time curve is the nominal ISO 834 standard fire curve (1), which represents a fire inside a building compartment (Fig. 3). The curve represents the fire in its “post-flashover” phase, that is, in its phase of full development. However, this curve is used in general purpose design for structural elements EN1991-1-1 (CEN 2002b), and it does not simulate all the fire phases (growth, flashover, and cooling phase); therefore, it is more severe than a real fire. For this reason, some authors have suggested using performance-based design curves (Hadjisophocleous and Zalok 2008) to produce a more competitive fire design solution:

\begin{matrix} T [^{\circ} C] = T_{0} + 345 \times \log_{10} (8 \times t [\min] + 1) \end{matrix}

(1)

Fig. 3. Internal and external nominal ISO curves for fire design in buildings.

Although these curves are for internal fires, it is already possible to use an external fire exposure curve, even though it is not related to a wildfire. This one tends to simulate a fire that occurs in a compartment at the perimeter of the building, in which flames may protrude through openings in the façade [Eq. (2)]. When the exterior of the building should provide fire resistance to prevent the fire from coming back into the building on the next floor, the external fire curve is used EN1991-1-1, (CEN 2002b). This curve considers that the flames are cooled by the presence of air outside the building; therefore, its maximum temperature is below the ISO-834:

\begin{aligned} \begin{matrix} T [^{\circ} C] = T_{0} + 660 \times [1 - 0.686 \times \exp (- 0.32 \times t [\min]) \end{matrix} \\ - 0.313 \times \exp (- 3.8 \times t [\min])] \end{aligned}

(2)

As observed in the previous section, it is important to realize that a wildfire possesses a maximum temperature of 800°C to 900°C for a 3 to 10-min period, depending on wind, fuel, and topography (Pyne et al. 1996). Therefore, even if a wildland fire endures for 30 min, its maximum temperature is clearly below 950°C, usually associated with 60 min of resistance. For this reason, the cited nominal temperature versus time fire curves are not suitable for an economical design protecting a dwelling against the action of a wildland fire.

Proposed Wildfire Curves

Performance-Based Wildfire Curve

Contrary to indoor fire, exterior wildfire presents a different behavior; therefore, the authors proposed a performance-based design curve for wildfires, opposing classical nominal curves. Several curves are presented in the literature, which considers three phases of a fire growth, flashover, and cooling. One of these curves is the one presented by Blagojević and Pešić (2011), in which these three phases are well represented for wildfires [Eq. (3)]. This one presents the best fit when comparing other studies’ measured in situ temperature versus time wildfire curves. Therefore, this temperature function is used in this work (Fig. 4):

\begin{matrix} T [^{\circ} C] = T_{\max} {[\frac{t [s]}{t_{max} [s]} \exp (1 - \frac{t [s]}{t_{\max} [s]})]}^{c} \end{matrix}

(3)

Fig. 4. Proposed fire-performance-based curved for wildfire.
(Adapted from Blagojević and Pešić 2011.)

where parameter T_max = maximum peak temperature; t_max = time instant when it happens; and c = dimensionless parameter that measures the growth and decrease of temperature. If c = 1, then the cooling is near-linear degradation, but if c > than 1, then the cooling presents a rapid exponential decrease.

Two different curves were generated: the first is for fire in trees with a higher temperature amplitude but a shorter duration; the second is for fire in the bush with a lower temperature amplitude but a longer duration. The authors chose to use a design fire curve instead of an upper-bond of all measured experimental fire temperatures. This is because the results measured by other authors present some variability due to atmospheric and geographic conditions. Therefore, some of these confounders or variables are indirectly accounted in the design wildfire curve.

The first two different T_max were computed using the average maximum temperature

T_{m}^{\max}

[Eq. (5)] of all other authors referenced before, for fires in trees and bushes. Second, both curves were assembled using the 95% characteristic values in Annex D “Design Assisted by Testing” in EN1990-1-1 (CEN 2002a) for the maximum temperature

T_{m}^{\max}

. The maximum temperature values were tested for normal distribution using the Kolmogorov–Smirnov test. The values of k_n (that affects the size of the sample) in Annex D (Table D.1) of EN1990-1-1 (CEN 2002a) were used with standard deviation coefficient s_x in Table 1 to calculate

T_{k, 95 %}^{\max} [^{\circ} C]

[Eq. (4)]. No safety coefficient is used for the maximum temperatures since it is an accidental action:

\begin{matrix} T_{k, 95 %}^{\max} = T_{m}^{\max} + k_{n} \times s_{x} \end{matrix}

(4)

with

T_{m}^{\max} = \sum_{i = 1}^{n} \frac{T_{\max}^{i}}{n} s_{x}^{2} = \sum_{i = 1}^{n} \frac{{(T_{\max}^{i} - T_{m}^{\max})}^{2}}{n - 1}

(5)

Table 1. Characteristic 95% temperatures for fire in trees and bushes

Parameters	$T_{m}^{\max} [^{\circ} C]$	s_x[°C]	k_n[−]	$T_{k, 95 %}^{\max} [^{\circ} C]$
Trees	940.81	154.37	1.83	1,223.30
Bushes	675.11	118.94	1.72	879.69

However, for the case of the peak time frame t_max and indirect duration parameter c for the cooling phase, due to the scatter of experimental results of wildfire curves measured in situ, the chosen values are the most severe for design purposes and the ones in the range of an upper-bond curve. Therefore, the adopted values are presented in Table 2, for both fires in trees and bushes.

Table 2. Adopted parameters for the design wildfire curves for trees and bushes

Parameters	T_max[°C]	t_max[sec]	c[−]
Trees	1,223	50	1.5
Bushes	880	150	1.0

Both curves are plotted in Fig. 5, and compared with real measured wildfire curves, in which, as predicted, both are an envelope fire curve. Therefore, it can be concluded by observing Fig. 5 that the tree-fire design fire curve is capable of acquiring the first swift temperature growth, but then the bush-fire design is good for attaining the longer length duration of a wildfire due to burning bushes.

Fig. 5. Proposed fire-performance-based curved for wildfire: (a) fire in trees; and (b) fire in bushes.

When designing the protection of dwelling against wildfire, it is possible to not know in advance if trees or bushes surround it; in that case, it is wise to use the combined input of both fire design fire curves. Also, it is impossible to predict whether the fire has started in bushes or trees during the design phase. Therefore, the design fire curves in Fig. 5 may present some delay that may change the thermal response due to variations of material thermal properties. Therefore, as an alternative a combined envelope wildfire curve is proposed next to solve this problem, using the combined maximum temperature data (T_max) of fires in trees and bushes.

Therefore, this work also studied the possibility of using a single wildfire design curve, with the average values and characteristic values of the maximum temperatures of trees and bushes at the same time. This will provide a new fire curve, with the characteristic parameters given in Table 3.

Table 3. Adopted parameters for the design wildfire curve combined envelope for trees and bushes simultaneously

Parameters	T_max[^°C]	t_max[sec]	c[ − ]
Trees + bushes	1,050	150	1.0

Fig. 6 depicts the new combined envelope wildfire design curve, which captures almost all maximum wildfire temperatures. Some minor values of fires in trees are below the characteristic envelope temperature, but their duration is quite short. Therefore, for a general design, this envelope wildfire curve is still suitable, and this curve will be later compared with the previous ones in the thermal numerical section.

Fig. 6. Proposed combined envelope fire-performance-based curved for wildfire.

Nominal Fire Curves versus Proposed Wildfire Design Curve

In Fig. 7, both curves for trees and bushes are compared with standard ISO curves, in which it is clear that the proposed curves present higher temperatures for shorter time instances; therefore, it can be concluded that standard ISO curves may not be suitable to be used in the design protection of dwellings against wildfire. In any case, in the following section, ISO standard curves will also be used and compared in the analysis with the proposed wildfire design curves.

Fig. 7. Comparing the proposed fire-performance-based curved for wildfire, with ISO standard fire curves.

Limitations of the Proposed Wildfire Curves

It is important to realize that the proposed wildfire curve is only valid if the dwellings are near trees and bushes, which is the most severe design case. In the case of a dwelling isolated in a field without any vegetation in the proximity, the most severe effect is the firebrand projection that may accumulate near the perimeter of the dwelling. These curves are not able to achieve the thermal effects of the firebrand projection; thus, this effect is beyond the scope of this study, but it will be investigated in the future.

Thermal Numerical Assessment

Description of the Numerical Model

In this section, the measured fire in situ temperature versus time curves are numerical tested using a wood specimen of 100 × 100 × 5 mm³, in order to reduce the computational costs. Wood was chosen, since it possesses an average thermal ignition temperature of around 288°C (Schaffer 1984), after which it abruptly changes its thermal properties, promoting higher temperature transmission.

To test the design temperature curves efficiency, these are directly compared with the measured fire in situ, concerning internal material maximum temperatures, and compared with the wood ignition temperature. In addition, the classical ISO standard fire curves, described in Eqs. (1) and (2), are also used.

The wood material adopts the thermal properties, density, conduction, and specific heat EN 1995-1-2 CEN (2004). Although some authors may suggest some variations of the convection coefficients (López et al. 2013), the values of EN1991-1-2-2:1995 (CEN 2002c) were adopted to follow standard design codes. The radiation coefficient adopted in the work of for wood (Gernay 2021) was used. The commercial finite-element software Abaqus standard model was used for the numerical thermal analysis. Since all thermal properties vary with temperature, a nonlinear thermal analysis is performed using the incremental Newton–Raphson iterative method.

To simulate fire contact with the wood surface, it is necessary to simulate the flux from convection and radiation that interact with the interface. To simulate these phenomena (convection and radiation heat flux), two film conditions are applied (in the lateral area) as presented in Fig. 8(a) and using adiabatic surfaces for the rest of the boundary 288 conditions (in the volume boundaries) Fig. 8(a). Since the temperature variation in wildfire is fast, 20 DC3D8 eight-node bilinear heat-transfer brick finite-element with full integration was used [Fig. 8(b)], to correctly acquire the temperature field distribution, due to strong flux gradient [Fig. 8(c)]. A plane 3D analysis was chosen to reduce the computational costs, since the use of a full cube provided no extra improvement in the accuracy of the solution.

Fig. 8. Convection and radiation heat flux: (a) thermal film condition; (b) finite-element mesh; and (c) temperature field distribution.

Numerical Results

Fig. 9 presents the numerical thermal responses of all the wildfire curves, in which it is possible to observe that all temperature curves presented values higher than the wood ignition temperature. The temperature values were measured in the closest integration Gauss point to the fire contact surface. This was performed to improve accuracy due variation of the internal thermal proprieties described previously (density + conduction + specific heat).

Fig. 9. Thermal numerical response of experimentally measured in situ wildfire curves, and the proposed wildfire design temperature versus time curves. Temperatures measured in the integration Gauss point nearest to the fire contact surface: (a) design fire curves in trees and bushes; and (b) combined design wildfire envelope.

To observe the advantage of the proposed wildfire fire curves, and the standard ISO internal and external fire curves, the thermal numerical output is also presented in Fig. 10. It is clear that the use of ISO fire curves may underestimate the effect of wildfire, in terms of maximum temperature, if the effect is designed for less than 10 min, which is the maximum duration time of a wildfire. To use the standard ISO fire curves, a minimum duration of 30 min is necessary to obtain a temperature above 800°C. Even if the external standard fire curve is used, the maximum temperature is around 600°C for 30 min, which is clearly below the proposed design wildfire curve.

Fig. 10. Thermal numerical response of the proposed wildfire design and standard ISO temperature versus time curves. Temperature values were measured in the integration Gauss point nearest to the fire contact surface.

Conclusions and Further Developments

In this work, two wildfire temperature versus time curves were proposed, to be used in future design protection of dwellings against wildland fire. The curves are upper-bound characteristic temperatures of in situ measured wildfire curves for fire in trees and bushes. It is possible to conclude, using a numerical assessment in a specimen that:

•

The proposed wildfire curves, clearly present an upper bound of temperature in the materials temperature distribution, when compared with the real ones measured in situ, for both trees and bushes.

•

The existing standard ISO curves for interior and exterior fires in buildings are not suitable for the design protection of dwellings against wildfire, due to its initial temperature values being quite small for a short duration. Both ISO curves, after only 30 min, are able to replicate the wildfire temperatures and are therefore more severe for the design purposes.

•

Contrary to classical standard ISO curves, in a wildfire the material temperature heats up very quickly; therefore, in terms of fire safety, some of the safety assumptions for short durations may not be accurate.

•

For design protection purposes against wildfire, due to unknown presence of trees or bushes in the vicinity of a future dwellings, it is recommended to use the combined design wildfire curve of both trees and bushes, since these produced the maximum field temperatures inside the specimen during the numerical assessment for a longer duration.

The most severe combination of the proposed design wildfire curves may be affected by the nonlinearity of the thermal properties related to temperature evolution. This is due to some dependency on the tested materials’ thermal properties during the temperature field propagation. Therefore, a parametric experimental and numerical campaign is necessary to assess the most severe load fire combinations in the future, considering the type of material and the origin of the wildfire being on bushes or trees.

For future research work, due to the lack of information in the scientific community, the authors will also study the temperature versus time curve of firebrands. Furthermore, the authors think the action of the firebrand may also be an important aspect in the design of fireproof construction solutions, since, according the reports of the great Portuguese wildfires of 2017 (Guerreiro et al. 2018; Viegas et al. 2017), these may be one of the responsible parts of fire ignition in dwellings.

Data Availability Statement

No data, models, or code were generated or used during the study.

Acknowledgments

The authors would like to acknowledge FCT, National funding agency for science, research and technology, Portugal (Research Project “New Fireproof Dwellings for Wildfire” PTDC/ECI-CON/2240/2020) and CERIS for the financial support.

References

Arruda, M. R. T., T. Tenreiro, and F. Branco. 2021. “Rethinking how to protect dwellings against wildfires.” J. Perform. Constr. Facil 35 (6): 06021004. https://doi.org/10.1061/(ASCE)CF.1943-5509.0001643.

Abstract

Introduction

Objectives

Research Significance

Existing Fire Curves

Measured Wildfire Curves in situ

Standard Fire Curves

Proposed Wildfire Curves

Performance-Based Wildfire Curve

Nominal Fire Curves versus Proposed Wildfire Design Curve

Limitations of the Proposed Wildfire Curves

Thermal Numerical Assessment

Description of the Numerical Model

Numerical Results

Conclusions and Further Developments

Data Availability Statement

Acknowledgments

References

Information

Published In

Copyright

History

Authors

Affiliations

Metrics

Citations

Download citation

Cited by

Figures

Other

Share

Copy the content Link

Share with email

Share

Request Username

Create a new account

Change Password

Password Changed Successfully

Verify Phone

Congrats!