Fire hazard of compressed straw as an insulation material for wooden structures

The construction sector continues to adapt to the challenges posed by climate change. Architects and engineers aim to build sustainable, energy, resource, and cost‐efficient structures by increasingly using bio‐based building materials. However, fire safety has always been a significant concern for timber building construction internationally. The objective of the study presented in this article is to document fire hazards of compressed straw when used as thermal and acoustic insulation within wood‐framed building assemblies. Three densities of compressed straw (75, 125, and 175 kg/m3) were selected and their combustion and thermal responses were evaluated at various scales, in attempt to define the optimal density considering various factors. The performance of the straw was also compared with commercially available insulation materials and then tested under exposure to severe heating in medium‐scale wood‐framed assemblies to evaluate the impacts of the straw as compared with a noncombustible insulation. The compressed straw with a density of 75 kg/m3 was found to have the best behavior with respect to both reactions to fire and insulation properties. The results suggest that compressed may have similar or better behavior under the heating conditions investigated when compared to a commercially available combustible insulation material. The use of this material as a primary insulation in a buildings is considered manageable by thoughtful design, construction, and building use without unduly increasing risks associated with fire.


| INTRODUCTION
Sustainable development is a significant challenge in modern times.
The building sector is responsible for about 40% of the global annual natural resource consumption and contributes up to 30% of all greenhouse gas emissions. 1,2 In recent decades, raw material consumption has increased significantly as has the trend of buildings' energy consumption. 1,3 Stakeholders in the building sector increasingly strive to build sustainable, energy, resources, and cost-efficient structures by using bio-based building materials. Their excellent environmental performance, as well as their physical properties, are the basis of more sustainable designs.
Wood-frame systems insulated with bio-based material represent an effective strategy to create biogenic carbon sinks at comparatively low-cost. [4][5][6] In recent years, interest has increased in these low embodied carbon and local building materials. 7 Using cereal crop byproducts in buildings is an attractive alternative to using conventional insulation materials. The revaluation of existing natural resources from accumulated agricultural waste does not require significant manufacturing costs, and generates almost no construction waste. 8 These beneficial aspects encourage their use in building insulation.
However, fire safety has always been a concern to combustible bio-based building materials. As naturally growing materials, timber structures and bio-based insulation materials consist mainly of organic compounds and are combustible. Whilst currently compressed straw insulation is currently only occasionally used in buildings, wood is more commonly used and specific fire protection regulations for timber are available. 9,10 A paucity of information regarding the fire hazard of compressed straw insulation, however, leads to its limited use. 11 Some limited evidence exists from fire resistance tests performed in the recent years that non-load-bearing walls insulated with straw bales can achieve 2-hours of fire resistance when tested under standard fire exposure according to ASTM E-119 or ISO-834. 12,13 It is well known that combustion processes typically require sources of heat, fuel, and oxygen. By compressing straw it is possible to dramatically decrease the ability of oxygen within a straw insulation product. 13,14 However, the fire behavior of this material may change under different conditions, such as with varying density, the type of cover material (ie, gypsum board, plaster, or stucco), the use of furring strips (metallic resilient channels).
Fire testing of construction materials has traditionally relied on the use of a standard reaction to fire test, for materials and construction products, and fire resistance test (ie, furnace test), for structural elements and assemblies, to ensure compliance and to unify all fire tests under one single standard for the purposes of comparison. [15][16][17] Reaction to fire is related to the flammability of materials and construction products, that is, combustibility, ignition, heat release, spread of flame, and evolution of smoke and toxic gases. 18 Fire resistance is related to the ability of a building element to resist a severe fire, defined as the time for which the element meets mechanical resistance (R), integrity criteria (E), and insulation criteria (I) when exposed to a standard test fire. 19 Such standard tests also aimed to provide limited information on the structural response to fire. Originally developed in the early 1900s, the furnace test has remained largely unchanged. [20][21][22] Despite some technical improvements, numerous problems remain until now such as high operating costs, poor repeatability, unrealistic and/or inappropriate boundary conditions, and poor statistical confidence. 23 Novel methods have emerged to undertake tests to study the performance of materials and assemblies to severe heating by controlling a fire experiment using an incident radiant heat flux rather than the gas phase temperature at the exposed surface. Using an incident radiant heat flux for fire science experiments is not, however, a recent concept; it has been widely used over the past 50 years in a variety of studies. [24][25][26] Several commercial fire testing apparatus using a radiant heat source are widely available, such as the Fire Propagation Apparatus 27,28 or Cone Calorimeter, 18,29,30 and these are routinely used for small-scale tests with a controlled incident radiant heat flux. Several authors have also suggested using a time-history of incident radiant heat flux, rather than a prescribed time-history of temperature when describing a fire. [31][32][33][34] The main scientific advantage of testing via heat flux rather than using gas phase temperature is associated with the capacity to control thermal boundary conditions directly at the exposed surface of a test specimen. Other advantages include the ability to impose a range of thermal exposures, outstanding repeatability, and operation at low economic and temporal costs. 23 This is the approach taken in the current research.
The research presented in this article aimed to partially document some of the fire hazards associated with compressed straw when used as a primary insulation material within wood-framed assemblies.
Specifically, the study evaluated the thermal response of three different densities of compressed straw insulation. The optimum density, based on the results of those tests, was selected and compared against test results for commercially available combustible insulation material that was previously tested using similar methodologies by Dagenais et al. 35 The selected density was installed within a woodframed assembly to evaluate the comparative performance of compressed straw under severe heating from one side, as compared with an assembly traditionally insulated with mineral wool. This was done using a Heat-Transfer Rate Inducing System (H-TRIS) apparatus. 36 The assembly method used in this study was intended to some aspect of reaction to fire of the insulation materials in a wood-frame assembly.

| Cone calorimeter testing
Six products were selected and evaluated following the cone calorimeter test method ISO 5660-1. 18 The apparatus used was the dual cone calorimeter produced Fire Testing Technology Ltd. The specimens were exposed to a heat flux of 50 kW/m 2 , when positioned horizontally 25 mm below the conical heat source. To perform the cone calorimeter tests, the straw was restrained within a 25 mm thickness using a metallic wire mesh (6.35 mm mesh) ( Figure 1). No adhesive was used to secure the straw within the wire mesh. Additional testing was conducted on the selected straw density (75 kg/m 3 ) with decreasing heat fluxes, that is, 35, 20, 15, and 10 kW/m 2 , to calculate the critical heat flux for piloted ignition ( _ q 00 cr ) and the ignition temperature (T ig ) according to Grenier and Janssens method. 37 All cone calorimeter tests were conducted at FPInnovations' Materials Evaluation Laboratory in Quebec City, Canada, including those of Dagenais et al. 35 Prior to testing, all specimens were conditioned to a constant mass at FPInnovations' facilities at a temperature of 23 C ± 2 C and a relative humidity of 50% ± 5%.

| Thermal conductivity
The thermal conductivities of the three compressed straw specimens were evaluated with a heat flow meter according to ISO 8301 38 with a temperature range of 15 C to 40 C. The apparatus used was a FOX 314 produced by TA Instruments. The straw was restrained in a wooden box of 305 × 305 × 102 mm, built using 19.1 mm thick standard CSA O121 Douglas Fir Plywood. Since the straw was restrained within a wooden box, the thermal conductivity of one layer of plywood (same as that used for the box) was evaluated to subtract its thermal contribution according to the method described by Drysdale 2011. 39 The thermal conductivity tests were conducted at Laval University, Canada. Prior to testing, all specimens were conditioned to a constant mass at University Laval's facility at a temperature of 21 C ± 1 C and a relative humidity of 60% ± 5%. The thermal conductivities of the combustible commercial insulation materials, that is, Multitherm, Thermowall-gf, and XPS, were taken from the manufacturers' websites. [40][41][42]

| Heat-Transfer Rate Inducing System
The H-TRIS apparatus, originally developed at the University of Edinburgh and shown in Figure 2, was used for evaluating the heat transfer within the wood-framed assemblies. Rather than using a standard fire test curve in a furnace such as CAN/ULC S101, 22 the H-TRIS apparatus allows direct and independent control of the thermal boundary condition by controlling the time-history of the incident radiant heat flux impinging at the exposed surface of a test specimen. 36 The H-TRIS apparatus consists of four propane-fired radiant panels mounted to a metal frame to form a 200 × 400 mm 2 radiant array. The panels are installed on a linear motion system, which allows the incident radiant heat flux to be varied by changing the distance F I G U R E 1 Compressed straw specimen restrained by a metallic wire mesh [Colour figure can be viewed at wileyonlinelibrary.com] between the radiant panels and the exposed surface of the test specimen.
A time-history of incident radiant heat flux corresponding to the standard fire curve ISO-834 43 was applied to the exposed surface of the specimens for a test duration of 60 minutes. The specified heat flux time-history is shown in Figure 3.

| Wood-frame assemblies
Two different wood-framed assemblies were built for the mediumscale tests in H-TRIS. The only parameter that varied between these was the cavity insulation. Manually compressed straw at 75 kg/m 3 was used in all cavities, whereas mineral wool was placed on the exposed side. For the wood-framed assemblies, 302 mm depth I-joists manufactured by Nordic Engineered Wood were used. These were spaced at 406 mm on center and made with 63.5 mm wide and 38.1 mm thick finger-jointed black spruce lumber flanges and a 9.5 mm thick oriented strand board (OSB) web panel. The exposed side was covered by one layer of 12.7 mm Fireguard C (Type C) gypsum board and two horizontal 12.7 mm deep resilient metal channels spaced at 600 mm on center. The unexposed side was covered by one layer of 19.1 mm of standard CSA O121 Douglas Fir Plywood.
Metallic wire mesh was used to restrain the compressed straw in the cavity on the side of the resilient channel and to maintain the air gap.
The height of the assemblies was 762 mm. Plywood was used to seal top and bottom end. These tests were conducted at the University of Edinburgh, Scotland. Prior to testing, all specimens were conditioned to constant mass at the University of Edinburgh's facilities at a temperature of 20 C ± 2 C and a relative humidity of 40% ± 5%.

| Instrumentation
Between 15 and 21 high temperature glass insulated Type K thermocouples were positioned throughout the assemblies. The locations of the thermocouples for the test assembly with compressed straw are shown in Figure 4. The thermocouple locations were as follows: 1 Two thermocouples at the unexposed surface of the gypsum board.
One in the center of the assembly and one at mid-height and midway between two joists.
2 Four thermocouples on the middle joist of the assembly. Two on the exposed flange at 20 mm from the exposed edge, one on each side of the flange; two at mid-depth of the joist, one on each side of the OSB web panel.
3 There were two rows of thermocouples within the insulation. One at midway of the central and side joists, one at quarter distance   Furthermore, its lower thermal conductivity makes it a more effective insulation material when compared to the other densities. As such, a density of 75 kg/m 3 was selected for further tests and comparisons.
A summary of the test results of the selected density of compressed straw (75 kg/m 3 ) and a number of commercially available combustible insulation materials is presented in Table 2, calculated over a 180 seconds burning duration; after the ignition of all specimens. The analysis time was selected due to the shortest burning duration of the XPS. All the results were compared to the 75 kg/m 3 density, in parentheses, as a benchmark. Figure 5 shows the heat release rates of these insulations.
A significantly lower average peak of heat release rate was observed for compressed straw than for Multitherm, Thermowall-gf or XPS. The XPS peak HRR was more than three times higher than that of the compressed straw. The compressed straw, as well as Multitherm and Thermowall-gf, burned more slowly throughout the tests whereas XPS burned vigorously after melting. XPS released heat 16 seconds later than the others specimens due to the melting time.
These results confirm that compressed straw would have a lower contribution during a fire than XPS. The results also suggest that compressed straw, if used as an insulation material, would be less prone to contribute in the early stages of a building fire when compared to the Multitherm, Thermowall-gf, or XPS. The thermal conductivity of the compressed straw was, on the other hand, higher than all other materials. Figure 6 shows the ignition times of 75 kg/m 3 density straw exposed to different heat fluxes. These results allow calculating the critical heat flux for piloted ignition ( _ q 00 cr ) and the ignition temperature (T ig ) to be calculated, which are determined as 7.5 kW/m 2 and 240 C, respectively.  less spectacular. Some steam and smoke appeared slightly earlier than in the compressed straw test, but the intensity remained lower. Small flames were visually noticed after 44 minutes at the junction of the gypsum board and the top plywood. The intensity of the flaming was not seen to increase significantly.

| Average temperatures on the unexposed side of the gypsum board
Two thermocouples were located on the unexposed surface of the gypsum board, as detailed in Section 2.6. The average temperatures measured between these two positions are shown in Figure 7.
The average temperature profiles followed the same trends in the early stages of the test, and then diverged due to the insulation configurations: 1 An initial temperature rise with temperatures at the back face reaching 88 C to 90 C at about 3.5 minutes.

| Average temperatures in the insulation materials
The average internal temperature profiles in the compressed straw and mineral wool insulation are shown in Figures 8 and 9, respectively. For both insulation materials, there was an increase in the temperature starting at approximately 3 minutes for the nearest thermocouples to the exposed surface. In the compressed straw, this was characterized by a short rise of temperature for the first thermocouples reaching 70 C at 5 minutes and then a more gradual In the mineral wool, the temperature increased about the same time as it did in the compressed straw, but conversely, the temperature rise was almost uniform and followed the same trend depending on the depth of the thermocouple. Since the mineral wool is a noncombustible material, the temperature profiles followed the same trend. The slope became linear around 45 minutes when the measured temperature was 630 C.
For the temperatures measured on the unexposed face of the assemblies, no significant temperature rise was expected due to the F I G U R E 9 Average temperature profiles in the mineral wool thickness of the wall and the comparatively short test time. The temperature rise was little to none during the tests, as shown in Figure 11. At about 20 minutes, a small gap was created between the assembly with compressed straw and the one with mineral wool.
Since the mineral wool did not fill all the cavity (only 89 mm in thickness), the heat transfer was slightly faster in the air cavity due to the air convection. However, this temperature gap tended to decrease by 40 minutes due to a temperature increase in the assembly with compressed straw. As suggested from the previous graphs, when the compressed straw began to combust, the temperature increased in the material.

| Average temperatures on the I-joist
Four thermocouples were located on the middle I-joist within the assemblies. Two were placed on the exposed flange at 20 mm from the exposed edge and two at mid-depth of the joist web panel. The respectively. Despite the combustible nature of the compressed straw, the delay to cause combustion of the timber was less important than expected when compared to mineral wool. If the compressed straw ignition temperature was the same that the one calculated in Section 3.1, the ignition temperature would have been reached 6.5 minutes (25%) faster in the assembly the thermal conductivity of which was 26% higher than in the assembly insulated with mineral wool (thermal conductivity of 0.036 W/mK for Comfortbatt). It would most likely suggest that, in a wooden structure, the fire risk of compressed straw is not necessarily proportional to its fire load, but also to the heat transfer occurring due to a greater thermal conductivity.
Insulation is used as a protective material to enhance the thermal and acoustic performances of a structure/assembly at ambient temperature. However, researchers [44][45][46] have stated that cavity insulation reduces the fire performance of a load-bearing structure. In cavityinsulated wall or floor, the insulation acts as a heat barrier and thereby heating the flanges faster, regardless of the combustibility of the insulation. Figure

| CONCLUSION
This article aimed to partially document some of the fire hazards associated with compressed straw when used as a primary insulation material within wood-framed assemblies. This was done by evaluating the thermal responses of three different densities of compressed straw insulation. The optimum density, based on the results of the tests performed, was selected and compared against test results for one particular commercially available combustible insulation material; this was previously tested using similar methodologies by Dagenais et al. 35 The selected density was installed within a wood-framed assembly to evaluate the comparative performance of compressed straw under severe heating from one side, as compared with an assembly traditionally insulated with mineral wool. This was done using a H-TRIS apparatus. 36 The assembly method used in this study was intended to evaluate some aspect of reaction to fire of the insulation materials in a wood-frame assembly.
According to the cone calorimeter test results, the lowest density of compressed straw (75 kg/m3) emitted less energy for the peak and heat release rate, when burning compared to the other densities (125 and 175 kg/m3) and compared to the commercially available combustible insulation materials. The heat release rate was lower than the bio-based insulation materials (Multitherm and Thermowall-gf) and was much lower than that of the XPS, suggesting that the compressed straw is less prone to contribute energy in the early stages of a building fire under conditions similar to those tested herein.
The medium-scale reaction to fire experiments on wood-frame assemblies insulated with compressed straw and mineral wool highlighted the impact of the fire load from the bio-based insulation material compared with a noncombustible cavity insulation. These tests used the H-TRIS test method and apparatus. All of the assemblies were constructed in a similar manner, with the exception that the cavity filling material was different between specimens. The test duration was 60 minutes using a calibrated time-history of incident radiant heat flux which was intended to be similar to a standard fire curve according to  Temperature measurements were made throughout the test assemblies. The data were used to investigate the thermal conditions within the test assemblies. The results indicated that one layer of 12.7 mm Type C gypsum board provided approximately 18 minutes encapsulation time (delay) before the effect became less important.
The measured temperatures showed that the compressed straw had a similar temperature profiles to mineral wool in the early stages of the tests. However, combustion of the compressed straw influenced the heat transfer and accelerated the ignition of the wood, which in turn accelerated combustion of the straw. Using furring strips (metallic resilient channels) may have contributed to this behavior due to a free air flow feeding oxygen into the assembly to sustain combustion.
In light of these results, and under the conditions tested, compressed straw appears to pose a lesser risk in terms of reaction to fire as compared to conventional combustible insulation materials which are currently being used in practice.
Further research must be performed on different assembly configurations to further investigate delaying the ignition of the straw (ie, additional gypsum board, avoiding loose straw, removal of the air gap) before confident conclusions for application in real buildings can be given. Like many other commercially available combustible insulation materials, the fire hazard of compressed straw appears to be manageable by careful design and construction. It is expected that continued research in this area will allow confident guidance for building designers to be proposed.