A Comparative Study of Fire Code Classifications of Building Materials

Abstract

Whether noncombustible or combustible construction is used, the presence of combustible materials is likely to be used for various reasons, such as interior finishes, flooring, and insulation. Consequently, how regulations consider the degree of combustibility in their fire classifications will influence the level of fire safety provided in these buildings and the exchanges between all actors in the construction sector. In North America, the regulation of combustibility is primarily governed by surface flame spread assessed through the Steiner tunnel test. While there is a growing prevalence of calorimetric methods globally, their incorporation into North American building codes remains notably limited. Based on ISO 5660-1 cone calorimeter test results of twenty commercially available North American building materials, a comparative study was conducted between the Canadian flame spread classification and the classifications in Japan, New Zealand and the European Union (Euroclass). The tests and their limitations are described herein, as well as the conceptual frameworks. The results suggest that as materials’ combustibility levels increase, the level of agreement between classifications decreases and remains binary. The choice between the material and system scales is crucial for determining the effective development and implementation of regulations.

1. Introduction

The construction industry covers various actors with different profiles, including architects, designers, manufacturers, consumers, and regulators. The latter regulate the entire field of building materials to ensure compliance with safety, health, accessibility, building protection against fire, structural damage, and the environment [1]. To better manage the fire impact and to facilitate international commercial trade, a simple and comprehensive common language between all partners in the construction sector is necessary for proper fire risk management. This language is the fire classification of building materials.

The rise of eco-friendly building practices such as timber construction, particularly in high-rise buildings, has highlighted the importance of fire risk assessments and building safety management [2,3]. As a result, regulators are pushing toward performance-based regulations of the reaction to fire to address these concerns [4,5,6,7].

Global differences in practices and regulations across countries present a significant obstacle to achieving harmonization [8,9]. This is primarily because each nation has its own guidelines and protocols [10], which lead to inconsistencies and divergences in approaches. Moreover, it is essential to recognize the limitations and shortcomings of current fire behaviour classification systems [11,12]. These classifications focus on specific aspects of fire behaviour, such as surface flame spread and from time to flashover [2,3]. While these aspects are undoubtedly significant and worthy of consideration, adopting a more holistic approach that considers the multifaceted nature of fire behaviour is imperative.

There are still uncertainties regarding the lack of performance criteria or agreements, the fire scenario to be considered, and the limitations of predictive fire models, which are not well documented or widely understood [13,14,15]. For materials with a low degree of contributions to fire incidents, which are typically known as noncombustible materials due to their absence of ignition (or not easily igniting), there is some divergence in terms of conceptual understanding and classification criteria among different countries [10,16]. However, it is noteworthy that there is a certain degree of agreement or consensus regarding the chosen fire test and its implementation by these countries to evaluate the reaction to fire of such materials [16].

A substantial portion of construction materials used as insulation, interior finishes, and structural materials is typically considered combustibles, meaning that they may contribute significantly or moderately to a fire. Lately, there is an increasing trend among regulations to incorporate calorimetric tests in evaluating the fire reaction of combustible materials [10,12,17], albeit not quite recognized in North American regulations. Even though there are considerable similarities between North American standards (e.g., ASTM) and international (ISO) reaction to fire tests, it is noteworthy that the Steiner tunnel is exclusively used as the primary reaction to fire tests in both Canada and the United States of America [18]. This fire test is the main topic of discussion herein.

This study aims to analyse and compare current practices for fire classification, specifically between calorimetric and surface flame spread test methods, and to assess the advantages and disadvantages of these practices. For this purpose, an experimental study was conducted on 20 commercially available construction materials with different uses (structural, insulation, etc.) and fire behaviours (thermoplastic, timber, etc.). Materials were chosen based on their flame spread ratings and compliance with European Union regulations. Only cone calorimeter tests were conducted and discussed hereafter.

Regarding the surface flame spread method, the choice was made based on the North American classifications where it is widely employed, particularly the Canadian classification, which is similar (but not equivalent) to the US classification and widely documented [19]. As for calorimetric methods, the choice was made based on the European classification for the single burning item test [20], the Japanese performance classifications using the cone calorimeter, and the New Zealand classification, which is equivalent to the Australian classification [10], also using the cone calorimeter as an alternative to the room corner test.

2. Overview of the Fire Classifications

A fire is an exothermic chemical reaction between a fuel and an oxidizer (oxygen) under the effect of a heat source, typically known on a material scale as the “fire triangle”. On a system scale, a similar relationship can be made where as a fire evolves, the dynamics of each element of the fire triangle gradually evolve towards a more complex and dynamic interaction than at the material scale. This evolution leads to the establishment of a “dynamic fire triangle” characterized by behaviours and characteristics controlled by the ventilation, the fuel, and the boundary conditions of the system (radiation view factor, material distribution, etc.).

Given the complexity and panoply of possible configurations in a building, fire dynamic management would be complicated to generalize in a building code for ventilation conditions [21] (openings, etc.) and for system effects (building configurations). Hence, regulations are preferred to manage and classify the combustibility of materials in an attempt to control the combustion process. Building materials represent the building components and play a significant role in the construction sector. It is why the control of a material’s combustion process needs to be considered earlier in the design phase of a project.

According to the NFPA 550 [22] fire safety concept tree, combustibility can be evaluated by considering the properties and quantities of fuel and its distribution within the building. The amount and distribution of material in construction will depend on its reaction to fire properties. A material that does not react when exposed to fire (noncombustible) will be favoured over a material that does react when exposed to fire (combustible). However, combustible materials include many materials that respond differently to fire, resulting in varying fire risks. The same applies to noncombustible materials, as, although they may not contribute to fire, their behaviour can be quite different in fire conditions, and their associated deemed level of safety will be questionable. Hence, there is a need to consider the fire behaviour of each material to better control fuel based on a rational approach while not being too restrictive.

A classification system is based on a set of standardized test methods and performance criteria that enable the quantification and classification of a wide range of materials. Such a system loses its validity if materials cannot be classified using the same test method conditions and performance criteria. In such cases where the evaluation method for materials differs, it becomes challenging to establish a common agreement, thus making meaningful comparisons impossible. Therefore, it is essential to have a standardized evaluation framework that ensures consistency across all elements being compared. By implementing a uniform evaluation method, one can effectively assess and compare different elements while enabling them to draw meaningful conclusions and make informed decisions.

As a result, the choice of the measurement method will directly influence the resulting classification and associated fire risk. This study looks at the current state of combustibility testing and classification in specific international and Canadian regulations.

Over the last few decades, scientific knowledge of fire dynamics has constantly evolved, as have the methods used to characterize the reaction of materials to fire [21]. A characterization test aims to describe or measure a particular dynamic of a physical phenomenon (e.g., Young’s modulus in a bending test). Materials’ fire characterization tests follow the same path. A fire test should enable the measurement of a quantity related to fire dynamics. As shown in Figure 1, fire is an exothermic reaction between an oxidant and a fuel that produces heat and combustion products such as combustion gases and ashes.

As a result, there are three possible ways of characterizing a material’s fire reaction: depending on the degradation and consumption of reactants, energy production and combustion products, or a combination of these. An example application is the characterization of the heat release rate (HRR), which represents one of the leading fire characteristics in fire dynamics [23]. Three methods are known for measuring the rate of heat release: (I) mass loss rate analysis, (II) gas analysis (oxygen consumption, production of combustion gases, etc.), and (III) temperature rise analysis.

The method based on oxygen consumption is the current benchmark [24]. The principles of the method date back to Thornton’s work in 1917 [25] before being generalized by Huggett in 1980 [26]. To date, various standards have used oxygen consumption as the principle for determining heat release rates, such as ISO 5660-1 [27] (cone calorimeter), ISO 9705 [28] (room corner), and EN 13823 [20] (single burning item).

Three oxygen consumption tests that are widely used around the world are ISO 9705 (also available in an equivalent version of ASTM E2257 [29]), known as the room corner test; EN 13823, known as the single burning item (SBI) test; and ISO 5660-1, known as the cone calorimeter. The corresponding classifications that will be considered are the New Zealand classification for the ISO 9705 test and its equivalence using ISO 5660-1, the European classification from EN 13823, and the Japanese classification using ISO 5660-1.

2.1. Canadian Fire Classification

The National Building Code of Canada (NBCC) quite restrictively divides construction materials into two categories, noncombustible and combustible, using the vertical tube furnace test following the standard test method CAN/ULC-S114 [30]. The vertical tube furnace is a fire test used to determine a material’s fire behaviour. The test exposes the six faces of a small rectangular specimen to a temperature of 750 °C and measures the resulting temperature increase ($∆$T). Additionally, the test measures the material’s mass before and after the test ($∆$m) and the ignition time. In the United States, the ASTM E136 [31] apparatus is used to determine the noncombustibility of materials and is similar to the CAN/ULC-S114 apparatus. However, ASTM E136 and CAN/ULC-S114 methods differ in temperature measurement methodology and acceptance criteria.

The cone calorimeter allows measurements of the heat release rate (HRR) by the oxygen consumption of a specimen exposed horizontally to a constant radiant heat flux. The test provides access to additional quantities, such as the times to ignition and flame out, mass loss rate, effective heat of combustion, smoke production, surface extinction area (SEA), and the material’s total heat released (THR). Although ISO 5660-1 was used to develop CAN/ULC-S135 [32], there are major differences between the two test methods and they cannot be used interchangeably.

Furthermore, the CAN/ULC-S135 standard is only intended for testing materials that exhibit low levels of combustibility subject to a fixed heat flux of 50 kW/m² and based on a specific set of acceptance criteria, shown in

Table 1. Canadian fire classification criteria.

Class	CAN/ULC-S114	CAN/ULC-S135	CAN/ULC-S102/S102.2
Noncombustible	$∆$T ≤ 36 °C $∆$m ≤ 20% Duration of 15 min (1)	THR ≤ 3 MJ/m2 TSEA ≤ 1 m2 Duration of 15 min (2)	-
Combustible	-	-	Flame spread rating (3) Smoke production classification (3)

⁽¹⁾ No ignition is recorded during the 14 min and 30 s of the test. ⁽²⁾ The test duration is extended until there is no further release of heat or smoke. ⁽³⁾ The values are specified in the NBCC [1] according to the location, distribution, and amount of exposed materials.

, of THR and the total smoke extinction area (TSEA). It is noted that NBCC does not mention that materials meeting the acceptance criteria are classified as noncombustible. It only states that such materials can be used in noncombustible construction.

A similar comparison can be made for the United States, which has established standards for the generic version of the cone calorimeter, ASTM E1354 [33], and low-combustibility materials, ASTM E2965 [34]. However, these methods are still not widely used in the United States codes for classifying building materials. Combustible materials are further classified based on surface flame spread characteristics using the Steiner tunnel test per CAN/ULC-S102 [35] in Canada or ASTM E84 [36] in the United States of America.

Surface flame spread refers to the phenomenon in which the flame, located near a pyrolysis zone, advances toward the surface of a solid material that acts as a source of fuel [37]. This spread is a crucial element, regardless of whether the fire is a large-scale or a small fire that ignites within a confined space, such as a room. The spread of the fire across the surface is contingent upon factors such as the orientation and direction of the flame spread, the thickness and thermal inertia of the material, the sample’s geometry, and the influence of the environment [21].

The flame spread assessment procedure involves a series of tests to closely observe the progress of the flame front and pyrolysis zone or uses quantitative techniques such as measuring the temperature evolution along the sample [38].

The test apparatus used in North America, shown in Figure 2, consists of a duct of 7.6 m in length and 0.45 m × 0.3 m in width and height, respectively. Made of refractory bricks, the specimen is suspended from the ceiling in the case of the CAN/ULC-S102 standard [35] (as in the case of the ASTM E84 [36], UL 723 [39], and NFPA 255 [40] tests) and held on the floor in the case of the CAN/ULC-S102.2 version [41]. The thermal exposure is generated by a gas burner placed on one end of the duct, which produces a 90 kW flame for 10 min for the Canadian version and 88 kW for ASTM E84.

The test results are dimensionless and called differently. For CAN/ULC-S102, the flame spread rating (FSR) and the smoke production classification (SDC) are reported, while the flame spread index (FSI) and smoke production index (SDI) are reported for ASTM E84.

Flame spread ratings (FSR) provided by the Steiner tunnel test method should not be interpreted as the flame spread rate, ${\mathrm{V}}_{\mathrm{p}}$, of a given material, as shown in Figure 3. The flame spread rate represents the speed at which a flame travels across the surface of a material. The measurement of the flame spread rate is typically expressed in units of length per time, such as millimeters per second. The flame spread rate is a property intrinsic to the material itself, and is influenced by factors such as combustibility, density, and thickness.

However, the flame spread rating is a numerical value, with no unit, assigned to a material to compare its surface flammability to that of a control material (red oak). A flame spread value (FSV) is calculated for each test based on the area (A_T) under the curve of the flame front displacement (x_p) as a function of time (t), as shown in Figure 3b. According to CAN/ULC S102, the average of at least three FSVs, rounded to the nearest multiple of five points, determines the FSR.

The Steiner tunnel is not a gas analysis test, but a surface flame spread test. The operator monitors the surface flame spread through visual observation and by a thermocouple placed at the tunnel’s end. The tunnel test provides a valid comparative analysis between wood species and considers thermal inertia. However, extending this test to materials other than wood-based materials and fire behaviours requires further discussion.

The analysis and comparison in this study will consider the criteria outlined in the Canadian classification.

2.2. Japanese Fire Classification

The Building Standard Law (BSL) dictates the Japanese fire safety regulations. Japan can be considered the first country in the Asia–Pacific region to adopt a performance-based fire engineering design for buildings. Since 1983, it has been developed to include a partial performance-based approach and a complete approach from 1998 onwards. Regarding material classification, Japanese regulations use two methods to propose three classes of noncombustibility [42] (noncombustible, quasi-noncombustible, and fire-retardant materials).

Table 2. Japanese fire classification criteria.

Class	Method 1	Method 2
	ISO 5660-1 (1)	ISO 1182	ISO 17431
Noncombustible	THR ≤ 8 MJ m−2 HRRmax ≤ 200 kW m−2 (2) Duration of 20 min	$∆$T ≤ 30 K (3) $∆$m ≤ 50%	-
Quasi-noncombustible	THR ≤ 8 MJ m−2 HRRmax ≤ 200 kW m−2 (2) Duration of 10 min	-	THR ≤ 50 MJ (4) HRRmax ≤ 140 kW (2)(4) Duration of 10 min
Fire-retardant materials	THR ≤ 8 MJ m−2 HRRmax ≤ 200 kW m−2 (2) Duration of 5 min	-	THR ≤ 40 MJ (4) HRRmax ≤ 140 kW m−2 (2)(4) Duration of 5 min

⁽¹⁾ In addition to the numerical criteria, the specimen should not develop any cracks that would permit fire to spread. ⁽²⁾ An excess of less than 10 s is permissible. ⁽³⁾ The rise refers to the difference between the maximum temperature for 20 min and the final temperature at the end of the test. ⁽⁴⁾ The burner is included.

summarizes the methods and acceptance criteria for each method.

The first method is a gas analysis method and uses the cone calorimeter test per ISO 5660-1, as shown in the Figure 4. The acceptance criteria for the first method are summarized in Table 2 and they are based on the total heat release (THR) and the peak heat release rate (HRR_max) of a material when exposed to a radiant heat flux of 50 kW/m² during different periods.

The second method is based on the ISO 1182 [43] noncombustibility test for noncombustible materials and the reduced-scale ISO 17431 [44] compartment test for quasi-combustible and fire-retardant materials. The ISO 1182 is a vertical tube furnace test but differs from the ASTM E136 and CAN/ULC-S114 methods [16]. ISO 17431 is a reduced-scale representation of the ISO 9705 [28] room corner fire test and follows a methodology similar to the cone calorimeter for measuring the heat release rate. With method 1, materials will be classified into one of three groups according to the acceptance criteria in Table 2 based on temperature rise ($∆$T) and mass loss ($∆$m) from the ISO 1182 and the THR and HRR_max from the ISO 17431.

Given the wider use and recognition of the ISO 5660-1 cone calorimeter test method [12], only method 1 will be considered hereafter.

2.3. European Fire Classification

In February 2000, the members of the European Union (EU) decided to adopt a harmonized fire safety classification system for construction products called “Euroclass” in an attempt to facilitate commercial trade among the EU countries [45].

The European classification, EN 13501-1 [46], includes various test methods for classifying construction products according to their fire performance. The materials are classified based on their contribution to fire dynamics, smoke emission levels (S1, S2, and S3), and the level of falling particles or flaming droplets (d1, d2, and d3).

Table 3. Euroclass fire classification for construction products, except floor coverings [45].

Class	Test Standard	Criteria	Additional
A1: Noncombustible No contribution	EN ISO 1182 (1) and	$∆$T ≤ 30 °C; and $∆$m ≤ 50%; and tf = 0 s (no sustained flaming)	—
	EN ISO 1716	PCS ≤ 2.0 MJkg−1 (1); and PCS ≤ 2.0 MJkg−1 (2) (2a); and PCS ≤ 1.4 MJm−2 (3); and PCS ≤ 2.0 MJkg−1 (4)	—
A2: Noncombustible Limited contribution	EN ISO 1182 (1) or	$∆$T ≤ 50 °C; and $∆$m ≤ 50%; and tf = 20 s	—
	EN ISO 1716 and	PCS ≤ 3.0 MJkg−1 (1); and PCS ≤ 4.0 MJkg−1 (2); and PCS ≤ 4.0 MJm−2 (3); and PCS ≤ 3.0 MJkg−1 (4)	Smoke production (5); and flaming droplets/particles (6)
	EN 13823 (SBI)	FIGRA ≤ 120 Ws−1; and LFS < edge of specimen; and THR600s ≤ 7.5 MJ
B: Combustible Limited contribution	EN 13823 (SBI) and	FIGRA ≤ 120 Ws−1; and LFS < edge of the specimen; and THR600s ≤ 7.5 MJ	Smoke production (5); and flaming droplets/particles (6)
	EN ISO 11925-2 (8) Exposure = 30 s	Fs ≤ 150 mm within 60 s
C: Combustible Low contribution	EN 13823 (SBI) and	FIGRA ≤ 250 Ws−1; and LFS < edge of specimen; and THR600s ≤ 15 MJ
	EN ISO 11925-2 (8) Exposure = 30 s	Fs ≤ 150 mm within 60 s
D: Combustible Medium contribution	EN 13823 (SBI) and	FIGRA ≤ 750 Ws−1
	EN ISO 11925-2 (8) Exposure = 30 s	Fs ≤ 150 mm within 60 s
E: Combustible High contribution	EN ISO 11925-2 (8) Exposure = 15 s	Fs ≤ 150 mm within 20 s	Flaming droplets/particles (7)
F: Combustible Very high contribution or unclassified	EN ISO 11925-2 (8) Exposure = 15 s	Fs > 150 mm within 20 s

⁽¹⁾ For homogeneous products and substantial components of nonhomogeneous products. ⁽²⁾ For any external non-substantial component of nonhomogeneous products. ^(2a) Alternatively, any external non-substantial component having a PCS ≤ 2.0 MJm⁻², provided that the product satisfies the following criteria of EN 13823 (SBI): FIGRA ≤ 20 Ws⁻¹; and LFS < edge of specimen; and THR_600s ≤ 4.0 MJ and s1 and d0. ⁽³⁾ For any internal non-substantial component of nonhomogeneous products. (⁴) For the product as a whole. ⁽⁵⁾ s1 = SMOGRA ≤ 30 m² s⁻² and TSP_600s ≤ 50 m²; s2 = SMOGRA ≤ 180 m² s⁻² and TSP_600s ≤ 200 m²; s3 = not s1 or s2. ⁽⁶⁾ d0 = No flaming droplets/particles in EN 13823 (SBI) within 600 s; d1 = no flaming droplets/particles persisting longer than 10 s in EN 13823 within 600 s; d2 = not d0, not d1; ignition of the paper in EN ISO 11925-2 results in a d2 classification. ⁽⁷⁾ Pass = no ignition of the paper (no classification); fail = ignition of the paper (d2 classification). ⁽⁸⁾ Under conditions of surface flame attack and, if appropriate to the end-use application of the product, edge flame attack.

summarizes the various test methods for classifying materials, excluding flooring, linear pipe thermal insulation products, and electric cables. The EU classification subdivides materials into seven classes according to their contributions to fire: A1 and A2 for noncombustible materials; B, C, and D for ranges from very limited to medium fire contributions; and E and F for highly combustible materials.

The noncombustible classes A1 and A2 are based on the temperature rise ($∆$T), the mass loss ($∆$m) according to EN ISO 1182 [43] and on the gross calorific potential (PCS) according to the EN ISO 1716 calorimetric bomb test [47]. EN ISO 1716 is a test that characterizes the fire behaviour of a material on a material scale. During this test, a specimen of a given mass is ignited in an oxygen-filled environment under standardized conditions inside a bomb calorimeter of a fixed volume. The PCS is then calculated based on the observed temperature rise, which includes the heat loss by the latent heat of vaporization of water [47]. The rise in temperature determines calorimetry in the case of ISO 1716, whereas in the case of the cone calorimeter, it is determined by a gas analysis.

For the materials limited to the medium combustible class (B, C and D), their classification is based on a combination of EN 11925-2 [48] and ISO 13823 standards [20]. The ISO 11925-2 test evaluates a material’s flammability and ability to spread fire over a short 150 mm distance. The test setup consists of a combustion chamber, a specimen holder, and a small burner inclined at 45 degrees to serve as the ignition source [48]. ISO 11925-2 does not measure physical quantities. It is only based on binary visual observations of ignition and achieving the 150 mm flame spread (Fs) criterion during an exposure time of 30 s for low and medium combustible materials, and an exposure of 15 s for highly combustible materials. Although no physical quantities are measured, the test is mandatory for the European classification of combustible materials.

The EN 13823 test method evaluates a material’s behaviour in an intermediate-scale fire scenario. The scenario simulates a single burning item (SBI) in a corner of a room formed by two vertical wings, called the small wing and the large wing, covered by the test material. The ignition source consists of a propane burner providing a two-stage heat output of 30 kW and 90 kW [20].

A calorimetric hood collects combustion gases to measure oxygen consumption and smoke production. This test method’s classification considers the fire growth rate index (FIGRA), the lateral flame spread (LFS) along the large wing of the test specimen, and the total heat release (THR). The EN 13823 standard can be used alongside the ISO 1716 standard to classify materials as class A2.

During an EN 13823 test, the smoke level is determined based on the smoke growth rate index (SMOGRA) and total smoke production (TSP). SMOGRA is the quotient of the highest smoke production rate from the specimen and the time of its occurrence. TSP measures the total smoke production from the specimen within the first 600 s of exposure to the main propane burner flames. The occurrence of flaming particles or droplets is monitored during the first 600 s of exposure. Only droplets/particles that fall within 0.3 m of the angle line between the wings of the test specimen are recorded. As in the case of the LFS, the fall of flaming droplets and particles is visually observed. The classification of high combustibility classes E and F is based exclusively on the flame spread (Fs) following ISO 11925-2 [48].

The SBI test is substituted for flooring materials with the EN ISO 9239-1 [49] radiant panel test. This test assesses a floor covering’s ability to withstand flames and radiant heat by determining the spread of flames, smoke generation, and the minimum radiant heat required to sustain combustion. The test involves placing a specimen horizontally and exposing it to a defined heat flux emitted by a 30° inclined gas-fired radiant panel.

The experiment involves igniting a pilot flame at the sample’s hottest end and recording any flame front’s development and progression. Smoke production is monitored by measuring light transmission in the exhaust stack. One of the tested specimens is cut in a particular direction, while the second is cut in the perpendicular direction to the first. The lowest performer from the two orientations is repeated twice. Flame spread is visually measured, and heat flux is measured using Schimdt–Boelter sensors. The critical heat flux (CHF) is defined as the radiant flux at which the flame extinguishes or the radiant flux after a 30 min test period, whichever is lower.

According to EN 13501-1 [46], materials tested using ISO 9239-1 for floor applications can be classified as A2fl and Bfl for a critical heat flux of 8.0 kW/m² and 4.5 kW/m², respectively, and 3.0 kW/m² for Cfl and Dfl.

2.4. New Zealand Fire Classification

Since 1992 [50], a performance-based building code has determined fire regulations in New Zealand. Acceptable solutions in C/AS2 [51] are still provided for buildings except those in risk group SH (i.e., buildings with sleeping (residential) and outbuildings [52]). The part dealing with the classification, criteria, and the tests adopted is presented in the Verification Method: Framework for Fire Safety Design C/VM2 [53]. This verification guide for the design of buildings is intended to be used by qualified fire safety engineers to demonstrate compliance with the New Zealand Building Code (NZBC).

The methodology for classifying materials was derived from the recommendations of the European Reaction to Fire Classification (EUREFIC) project. The latter aimed to develop assessment methods based on the ISO 9705 room corner test, ISO 5660 cone calorimeter test, and those of the Fire Code Reform Centre [54], subsequently adopted in New Zealand [55]. According to the EUREFIC methodology, the flashover time was taken as the time required for the total heat release rate of the burner and the material was tested to reach 1 MW in ISO 9705.

The first classification method recommends that materials be grouped into one of the four categories shown in

Table 4. NZBC fire classification according to ISO 9705 [53].

Group Number	Exposition	Criteria
1	100 kW for 10 min, then 300 kW for 10 min	THR < 1 MW
1-S	100 kW for 10 min, then 300 kW for 10 min	THR < 1 MW SPR < 5 m2/s, from 0 to 20 min
2	100 kW for 10 min	THR < 1 MW
2-S	100 kW for 10 min	THR < 1 MW
3	100 kW for 2 min	SPR < 5 m2/s, from 0 to 20 min
4	100 kW for 2 min	THR < 1 MW

based on the ISO 9705 flashover time. According to the average smoke production rate, two additional group numbers, 1-S and 2-S, are derived from groups 1 and 2. The ISO 9705 consists of a room with inner dimensions of 3.6 m × 2.4 m × 2.4 m and a 2 m × 0.8 m door. The fire source is a burner located at the back corner, generating an intensity in two subsequent steps of 100 kW and 300 kW. A hood outside the open door collects the heat release rate based on oxygen calorimetry and average smoke production rate (SPR). In a typical room corner test, the back wall, the two side walls, and the ceiling are mounted with the test material.

While the test provides valuable insights into the fire growth characteristics of a given material, the results are obtained from a specific fire exposure. They should not be extrapolated to be representative of other exposures.

The second classification method is based on the cone calorimeter test as per ISO 5660 [27] at a constant radiant heat flux level of 50 kW/m². An empirical correlation derived from cone calorimeter test results was developed by Kokkala et al. [56] and is used to predict the ISO 9705 test results. This correlation is referenced in the New Zealand Building Code [53].

Based on cone calorimeter test results, the Kokkala et al. method requires calculating an ignitability index (${\mathrm{I}}_{\mathrm{i}\mathrm{g}}$), two heat release indices (${\mathrm{I}}_{\mathrm{Q}1}$ and ${\mathrm{I}}_{\mathrm{Q}2}$), and three integral limits (${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$, ${\mathrm{I}}_{\mathrm{Q}2\mathrm{m}\mathrm{i}\mathrm{n}}$, and ${\mathrm{I}}_{\mathrm{Q}12\mathrm{m}\mathrm{i}\mathrm{n}}$) [53]. The materials are then classified into four groups based on comparisons between heat release rate indices and integral limits based on the following four conditions:

• Group number 1—no ignition exceeds 50 kW/m² or ${\mathrm{I}}_{\mathrm{Q}1}\le {\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ and ${\mathrm{I}}_{\mathrm{Q}2}\le {\mathrm{I}}_{\mathrm{Q}12\mathrm{m}\mathrm{i}\mathrm{n}}$;
• Group number 2$–{\mathrm{I}}_{\mathrm{Q}1}\le {\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ and ${\mathrm{I}}_{\mathrm{Q}2}>{\mathrm{I}}_{\mathrm{Q}12\mathrm{m}\mathrm{i}\mathrm{n}}$;
• Group number 3$–{\mathrm{I}}_{\mathrm{Q}1}>{\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ and ${\mathrm{I}}_{\mathrm{Q}2}\le {\mathrm{I}}_{\mathrm{Q}2\mathrm{m}\mathrm{i}\mathrm{n}}$;
• Group number 4$–{\mathrm{I}}_{\mathrm{Q}1}>{\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ and ${\mathrm{I}}_{\mathrm{Q}2}>{\mathrm{I}}_{\mathrm{Q}2\mathrm{m}\mathrm{i}\mathrm{n}}$.

Two additional group numbers, 1-S and 2-S, are derived from the first and second groups, respectively, based on the average specific extinction area being less than 250 m²/kg, as calculated per ISO 5660.

3. Materials and Methods

This study compares the regulations described previously based on four main criteria: (1) scale, (2) cost and size, (3) measurement, and (4) performance.

The scale criteria relate to the scale representativeness of the specimen, material, or system being tested. The material scale characterizes only the fire behaviour of the fuel, while the system scale includes the effects of ventilation and the system in which the material will be used. When considering cost criteria, the cost of a test will be more closely related to the size of the test specimen. When evaluating measurement criteria, it is essential to consider both the measurement’s repeatability and reproducibility and list the quantities (properties) being measured. Finally, performance criteria refer to the interpretation of the physical measurement carried out and its implementation in a building code as it relates to the safety of people during a fire.

Table 5. Characteristics of the materials.

N°	Material	Thickness (mm)	Density (kg/m³)	Combustible (Y/N)	FSR
M01	Rock fiber insulation	36.6	128	N	0 (1)
M02	Fibreglass insulation	50.0	12.8	N	0 (1)
M03	Polyvinyl chloride board (PVC)	19.2	557	Y	15 (1)
M04	Gypsum board (Type X)	15.9	673	N	15 (1)
M05	Gypsum board (regular)	12.7	614	N	25 (1)
M06	Fire-retardant treated plywood (Douglas fir)	11.6	535	Y	25 (1)
M07	Douglas fir (Pseudotsuga menziesii)	19.1	561	Y	40 (2)
M08	White spruce (Picea glauca)	15.2	349	Y	50 (2)
M09	Parallel strand lumber (PSL)	31.9	680	Y	50 (1)
M10	Laminated veneer lumber (LVL)	24.4	599	Y	50 (1)
M11	Red oak (Quercus rubra)	22.4	764	Y	100 (2)
M12	Sugar maple (Acer saccharum)	20.3	780	Y	104 (2)
M13	Medium-density fibreboard (MDF)	16.2	766	Y	120 (3)
M14	Wood–plastic composite (WPC)	19.0	1022	Y	125 (1)
M15	Particleboard	16.5	607	Y	150 (3)
M16	Spruce plywood	12.9	501	Y	150 (1)
M17	Oriented strand board (OSB)	15.0	620	Y	175 (3)
M18	Extruded polystyrene insulation (XPS)	24.2	25	Y	190 (1)
M19	Wood fiber insulation (WFI)	12.8	232	Y	218 (1)
M20	Expanded polystyrene insulation (EPS)	25.0	24	Y	240 (1)

⁽¹⁾ Manufacturer’s product sheet. ⁽²⁾ Surface flammability and flame-spread ratings, Canadian Wood Council, 2020 [57]. ⁽³⁾ Fire design specification for wood, American Wood Council, 2022 [58].

lists twenty materials carefully chosen based on specific criteria. They should exhibit a range of different fire behaviours, be commonly used in building construction, be available on the market, have been fire tested and classified according to Canadian standards, and have an equivalent marketed in Europe. The surface flame spread (FSR) of the Canadian classification of the materials following the CAN/ULC-S102/S102.2 is also summarized in Table 5. According to the National Building Code of Canada, M01 and M02 are considered noncombustible materials, while M04 and M05 are acceptable to be used in noncombustible construction.

The materials were conditioned, prepared, and tested at the Université Laval fire research laboratory using a dual cone calorimeter supplied by Fire Testing Technology. The equipment is presented in the Figure 4. The cone calorimeter was calibrated according to ISO 5660-1 [27], including the polymethylmethacrylate (PMMA) calibration, and the testing followed the procedure provided in the standard. The data collected for each material were evaluated according to the first method’s criteria for the Japanese classification and the stochastic approach [56] of Kokkala for the New Zealand classification. All the experiments were carried out in triplicate, as required in ISO 5660-1.

The materials are subjected to a constant radiant heat flux of 50 kW/m² in a horizontal position and 25 mm below the conical heater. PVC, however, was positioned 60 mm away from the heat source due to its intumescent behaviour. According to the standard, the materials with a dimension of 100 × 100 mm were tested using the recommended stainless steel retainer frame and a low-density fibre blanket as backing material.

All specimens were wrapped aluminium foil, with the reflective side facing the specimen. The foil was pre-cut to cover the bottom and sides of the sample, with the foil not extending over the top of the specimen.

The objective is to evaluate the two methods in terms of their abilities to judge the degree of combustibility of materials based on cone calorimeter test results and how they correlate with current Canadian classification based on surface flame spread tests.

4. Results

4.1. Cone Calorimeter Test Results

Given the multitude of materials that have been tested and to facilitate the understanding and interpretation of the data, the heat release rate data have been appropriately divided into four distinct categories. These categories are designated as follows: noncombustible materials, miscellaneous materials, wood species, and bio-based materials.

Two main behaviours were observed for noncombustible materials. Fibreglass (M02) and rock fiber (M01) did not ignite, and their heat release rate (HRR) curves remained near zero, except for the first 20 s, when a slight increase of less than 10 kW/m² was noted. Gypsum boards (M04 and M05) exhibited rapid growth just after the ignition of the surface paper, followed by a short steady-state HRR caused by the gradual disappearance of the paper. The gypsum boards’ HRR_max values remain less than 120 kW/m². Figure 5 shows the heat release rates of the four noncombustible materials.

Four behaviours were reported for materials with miscellaneous behaviours. XPS (M18) and EPS (M20) insulations showed a fast increase in heat release with HRR_max exceeding 500 kW/m², followed by an equally fast decrease just after the HRR_max. The materials were entirely consumed after a short period of less than 160 s. PVC (M03) exhibited an oscillatory behaviour of the HRR, which could be attributed to its intumescent behaviour. The PVC curve oscillates between the ignition of the material; the first HRR_max occurs below 200 kW/m², and then a gradual extinction occurs until a fresh layer of material appears and combustion starts again.

The wood–plastic composite (M14) showed strong growth and a high HRR_max of 697 kW/m², followed by a decrease until a char formed. The char helped to limit the heat release rate. On the other hand, the fire-retardant treated plywood (M06) had a relatively limited heat release rate compared to other materials in the same category. After ignition, a char was formed, and the HRR_max and the heat release rate were decreased. The heat release rates of the miscellaneous materials are shown in Figure 6.

For wood specimens, a similar behaviour was observed among different wood species during combustion. After ignition, there was a fast increase in the heat release rate, which then decreased and stabilized until a char layer formed. However, a second peak appeared at the end of the combustion of the wood species.

The levels of the steady state and the decay rate varied depending on the type of wood. White spruce (M08) ignited quickly compared to other species but had a lower HRR_max and steady state. Despite a delayed ignition and a weaker spike than red oak (M11) and Douglas fir (M07), sugar maple (M12) had a higher steady state. Figure 7 shows the heat release rates of these wood species.

In the category of bio-based materials, the burning behaviour was quite similar to that of wood. It showed fast growth after ignition, followed by a decrease to a steady state before the appearance of a second peak at the end of combustion. In the case of wood insulation, a short combustion with a low growth rate was observed when compared to other bio-based materials. This behaviour appeared to be reasonable, considering the material thickness and density. LVL had the highest HRR_max among the bio-based materials and wood species. The heat release rates of the bio-based materials are shown in Figure 8.

The data obtained from the cone calorimeter experiments will be analysed hereafter and extrapolated specifically to the regulatory frameworks of Japan and New Zealand in compliance with the guidelines and stipulations defined by each respective authority.

4.2. Classification according to the Japanese and New Zealand Regulations

The Japanese classification based on the cone calorimeter test is used to demonstrate compliance with the prescriptive requirements of the Japanese BSL shown in Table 2. For the NZBC, the classification based on the cone calorimeter involves calculating the Kokkala correlation indices. The material is automatically classified as Group 1 (noncombustible) if there is no ignition.

Regarding the NZBC cone calorimeter classification, the ignitability index, ${\mathrm{I}}_{\mathrm{i}\mathrm{g}}$, is calculated as a function of the ignition time (${\mathrm{t}}_{\mathrm{i}\mathrm{g}}$). The latter is defined as the time when the measured heat release rate exceeds 50 kW/m².

The heat release rate indices, ${\mathrm{I}}_{\mathrm{Q}1}$ and ${\mathrm{I}}_{\mathrm{Q}2}$, are calculated using a numerical integration. ${\mathrm{I}}_{\mathrm{Q}1}$ indicates whether flashover is reached before 10 min or 20 min of testing, while ${\mathrm{I}}_{\mathrm{Q}2}$ specifies if the flashover is reached after 2 min of exposure to 100 kW or 300 kW.

The integral limits relate to the conditions when flashover occurs after 2 min (${\mathrm{I}}_{\mathrm{Q}2\mathrm{m}\mathrm{i}\mathrm{n}}$) when the burner intensity is at 100 kW (${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$), or after 2 min (${\mathrm{I}}_{\mathrm{Q}12\mathrm{m}\mathrm{i}\mathrm{n}}$) when the burner is increased to 300 kW. These integral limits represent the linear regressions of the flashover times obtained in the ISO 9705 test for 2, 10, and 12 min as a function of the ignitability index ${\mathrm{I}}_{\mathrm{i}\mathrm{g}}$.

The results of the 20 tested materials according to the Japanese classification criteria and the Kokkala et al. indices are presented in

Table 6. Cone calorimeter data projections according to the New Zealand and Japanese criteria.

	New Zealand Criteria						Japanese Criteria
N°	${\mathbf{I}}_{\mathbf{i}\mathbf{g}}$ min−1	${\mathbf{I}}_{\mathbf{Q}1}$ s0.66kW/m2	${\mathbf{I}}_{\mathbf{Q}2}$ s0.07kW/m2	${\mathbf{I}}_{\mathbf{Q}10\mathbf{m}\mathbf{i}\mathbf{n}}$ s0.66kW/m2	${\mathbf{I}}_{\mathbf{Q}2\mathbf{m}\mathbf{i}\mathbf{n}}$ s0.07kW/m2	${\mathbf{I}}_{\mathbf{Q}12\mathbf{m}\mathbf{i}\mathbf{n}}$ s0.07kW/m2	${\mathbf{HRR}}_{\mathbf{max}}$ kW/m2	${\mathbf{THR}}_{5\mathbf{min}}$ MJ/m2	${\mathbf{THR}}_{10\mathbf{min}}$ MJ/m2	${\mathbf{THR}}_{20\mathbf{min}}$ MJ/m2
M01	NI	NI	NI	NI	NI	NI	13 ± 3	0.6 ± 0.1	-	-
M02	NI	NI	NI	NI	NI	NI	11 ± 1	0.4 ± 0.1	-	-
M03	3.2 ± 0.5	8607 ± 482	730 ± 24	5076 ± 257	1948 ± 78	1123 ± 78	175 ± 8	15.7 ± 2.0	-	-
M04	1.3 ± 0.1	897 ± 15	316 ± 10	6124 ± 38	2268 ± 12	1443 ± 12	99 ± 4	2.2 ± 0.1	-	-
M05	1.6 ± 0.0	795 ± 323	258 ± 110	5924 ± 0	2207 ± 0	1382 ± 0	106 ± 29	2.4 ± 0.2	-	-
M06	4.6 ± 0.8	6607 ± 171	724 ± 5	4327 ± 416	1719 ± 127	894 ± 127	149 ± 16	13.1 ± 1.0	32.7 ± 0.6	-
M07	2.0 ± 0.4	12,982 ± 289	1263 ± 71	5731 ± 204	2148 ± 62	1323 ± 62	244 ± 23	27.3 ± 0.8	51.0 ± 1.5	115 ± 3
M08	3.9 ± 1.4	10,274 ± 70	963 ± 44	4692 ± 757	1831 ± 231	1006 ± 231	184 ± 8	20.7 ± 0.8	43.8 ± 1.8	-
M09	2.5 ± 0.9	14,797 ± 335	1483 ± 158	5427 ± 471	2056 ± 144	1231 ± 144	280 ± 54	35.0 ± 0.7	61.3 ± 0.6	114 ± 2
M10	1.5 ± 0.9	15,417 ± 336	1239 ± 571	5979 ± 511	2224 ± 156	1399 ± 156	308 ± 55	31.8 ± 1.8	61.2 ± 1.5	142 ± 3
M11	1.7 ± 0.1	13,512 ± 199	1273 ± 76	5898 ± 50	2199 ± 15	1374 ± 15	245 ± 19	29.3 ± 0.9	52.6 ± 0.9	116 ± 1
M12	1.3 ± 0.2	16,003 ± 458	1331 ± 37	6074 ± 81	2253 ± 25	1428 ± 25	207 ± 10	32.7 ± 1.0	63.6 ± 2.3	162 ± 4
M13	1.9 ± 0.1	19,352 ± 342	1572 ± 8	5795 ± 67	2168 ± 20	1343 ± 20	291 ± 1	39.9 ± 0.7	67.6 ± 1.1	-
M14	1.8 ± 0.1	28,068 ± 538	2380 ± 48	5856 ± 32	2186 ± 10	1361 ± 10	710 ± 13	71.5 ± 1.0	112.0 ± 2.4	-
M15	1.7 ± 0.2	15,193 ± 229	1327 ± 14	5860 ± 92	2188 ± 28	1363 ± 28	249 ± 10	34.5 ± 0.1	57.4 ± 0.1	-
M16	3.0 ± 0.5	12,163 ± 599	1167 ± 76	5175 ± 278	1978 ± 85	1153 ± 85	220 ± 2	25.1 ± 2.6	62.4 ± 3.4	-
M17	2.2 ± 0.0	15,395 ± 159	1306 ± 42	5614 ± 25	2113 ± 8	1288 ± 8	211 ± 4	34.8 ± 0.7	65.8 ± 2.6	-
M18	2.7 ± 0.4	7697 ± 75	1781 ± 8	5366 ± 241	2037 ± 74	1212 ± 74	697 ± 8	20.4 ± 0.3	-	-
M19	9.3 ± 1.0	8120 ± 133	1046 ± 34	1786 ± 545	943 ± 167	118 ± 167	209 ± 4	32.2 ± 0.1	-	-
M20	2.3 ± 0.1	7285 ± 161	1676 ± 94	5584 ± 51	2103 ± 16	1278 ± 16	541 ± 59	19.8 ± 0.4	-	-

NI: no Ignition. -: flameout occurred or the absence of heat release.

. The results will be used to discuss the classification based on the various approaches detailed herein. For the Japanese criteria in Table 2, the first HRR_max for wood species and bio-based materials is considered.

5. Discussion

5.1. Comparison of Fire Classifications of Materials

The fire classifications of the 20 tested materials based on the Canadian, Japanese, European and New Zealand methods are presented in

Table 7. Comparison of fire classifications.

N°	CAN/ULC-S114	CAN/ULC-S102	Japanese Class	Euroclass	New Zealand Group
M01	Noncombustible	0	NC	A1	1
M02	Noncombustible	0	NC	A1	1
M03	Combustible	15	UC	Bfl 2	3
M04	Noncombustible	15	NC	A2	1
M05	Noncombustible	25	NC	A2	1
M06	Combustible	25	UC	B 1	3
M07	Combustible	40	UC	D	3
M08	Combustible	50	UC	D	3
M09	Combustible	50	UC	D 1	3
M10	Combustible	50	UC	D	3
M11	Combustible	100	UC	D	3
M12	Combustible	104	UC	D	3
M13	Combustible	120	UC	D	3
M14	Combustible	125	UC	Efl 2	4
M15	Combustible	150	UC	D	3
M16	Combustible	150	UC	D	3
M17	Combustible	175	UC	D	3
M18	Combustible	190 3	UC	E	3
M19	Combustible	218	UC	E	4
M20	Combustible	240 4	UC	E	3

NC: noncombustible. UC: unclassified. ¹ Projection on similar materials. ² Marketed and classified for floors. ³ According to CAN/ULC-S102.2 and equal to 10 according to ASTM E84. ⁴ According to CAN/ULC-S102.2 and equal to 25 according to ASTM E84.

. It can be observed that the choice of a fire classification should not depend essentially on the choice of the test to be used because there is no good or bad test. All the previous tests classify the materials based on specific physical properties and/or behaviours. Instead, the classification should depend more on the engineering variables and the objective to be considered [12].

The Japanese cone calorimeter classification system determines the degree of noncombustibility of materials. As expected, noncombustible materials M01, M02, M04, and M05 fall under the noncombustible class, while all other materials are grouped under a single combustible class. It should be noted that, among the materials tested, there is a lack of material in the quasi-noncombustible and fire-retardant classes. It is also noteworthy to mention that the current Japanese regulations have criteria essentially designed for materials exhibiting low levels of combustibility. It should also be noted that the concept of fire retardancy materials is not identical under Japanese and Canadian regulations.

For example, M06 is a fire-retardant plywood with an FSR of 25 under Canadian regulations, but it is not considered a fire-retardant under Japanese regulations. This difference is because the concept of fire retardancy in Japanese legislation is based on a material’s contribution to fire dynamics. In contrast, Canadian legislation is based on a material’s ability to reduce surface flame spread. The correlation between the surface flame spread rating and fire contribution is not reciprocal. Gypsum board and fire-retardant plywood both have FSRs of 25. However, M05 is noncombustible, while M06 is neither noncombustible nor fire-retardant according to Japanese regulations.

According to the comparison between the FSR and the peak HRR in Figure 9, the materials with an HRR lower than 175.1 kW/m² fall within the FSR index range of 0–25. This correspondence is also seen between the FSR and the THR in Figure 10, which shows the THR at 5 min. The correspondence mentioned above is no longer valid for materials with an FSR less than 25 if the THR_5min exceeds 16 MJ/m². However, a discrepancy is observed between the HRR_max or THR and the FSR indices for materials with FSR indices greater than 25.

The New Zealand classification system is similar to the Japanese classification system, as it groups noncombustible materials in a single group. Except for M14 and M19, all other materials come under Group 3. It should be noted that no materials used in this study fell under the Group 2 classification according to New Zealand’s classification. Materials that are typically classified as Group 2 include gypsum board covered with layers such as vinyl wallpaper or polystyrene layers [59].

It can be seen from the comparison of the Japanese, New Zealand, and Canadian fire classifications in

Table 8. Comparison of the Japanese, New Zealand, and Canadian fire classifications.

		New Zealand Groups
		1	2	3	4
Japan	NC (1)	M01, M02, M04, M05	-	-	-	NC (1)	Canada
	QC (2)	-	-	-	-
	FR (3)	-	-	-	-
	C (4)	-	-	M03, M06, M07, M08, M09, M10, M11, M12, M13, M15, M16, M17, M18, M20	M14, M19	C (4)

⁽¹⁾ NC: noncombustible material. ⁽²⁾ QC: quasi-noncombustible material. ⁽³⁾ FR: fire-retardant material. ⁽⁴⁾ C: combustible materials that do not meet noncombustibility requirements.

that all noncombustible materials, according to the Canadian classification, are classified in the first class, also called noncombustible materials. The similarity between the two noncombustible categories can be attributed to the resemblance in the standards used for the second method of the Japanese classification and the CAN/ULC-S114 classification. Both classifications require a temperature increase of 30 °C for the Japanese case and 32 °C for the Canadian case.

Although the cone calorimeter classification has a more stringent heat of combustion criterion of 3 MJ/m² for the Canadian classification compared to 8 MJ/m² for the Japanese classification, the tested materials are classified in the same category. These classification methods divide materials into two categories based on their fire classifications without providing further details on their levels of combustibility.

When comparing the Canadian and New Zealand classifications, it is observed that the two classifications have a reasonably similar separation between noncombustible and combustible materials. In contrast to Japan and Canada, the New Zealand model permits the subdivision of combustible materials into Groups 3 and 4. The model provides greater flexibility for classifying materials based on their combustibility.

The level of combustibility provided by the New Zealand method for Group 3 is relatively low compared to the cone calorimeter and vertical tube furnace classifications. Except for materials M14 and M19, as mentioned previously, all materials exhibiting different fire behaviours are classified into the same group.

Notably, Group 4 encompasses the M14 and M19 materials, which exhibit the shortest ignition time (the highest ${\mathrm{I}}_{\mathrm{i}\mathrm{g}}$) for M19 and the highest HRR_max and THR_10min for M14. This finding agrees with the results of the study by Kokkala et al. [56].

Regarding the flame spread rating criteria, materials classified as noncombustible in Japan and Group 1 in New Zealand would typically have an FSR between 0 and 25, as observed in

Table 9. Comparison between the New Zealand groups and the flame spread rating.

		New Zealand Groups
		1	2	3	4
Flame spread rating	[0–25]	M01, M02, M04, M05	-	M03, M06	-
	[26–75]	-	-	M07, M08, M09, M10	-
	[76–250]	-	-	M11, M12, M13, M15, M16, M17, M18, M20	M14, M19

. However, it should be noted that PVC and fire-retardant plywood are outliers in this classification. Although fire-retardant plywood has an FSR no greater than 25, it would not meet the noncombustibility requirements of New Zealand classifications. The PVC tested has an FSR of 15 and is classified as Group 3 in the New Zealand classification. This difference can be explained by PVC’s intumescent fire behaviour, which inhibits the surface flame spread when it is subjected to the CAN/ULC-S102 tunnel test.

The Kokkala correlation appears to classify bio-based materials, wood species, and thermoplastic materials in the same fire risk group. In the section of New Zealand fire classification, it is mentioned that the ${\mathrm{I}}_{\mathrm{Q}1}$ and ${\mathrm{I}}_{\mathrm{Q}2}$ heat release indices with limiting integral combinations are intended to correlate with the flashover time in a room corner test.

The noncombustible materials M01 and M02 will not be considered in the following discussion between the FSR and the Kokkala model indices given that they both did not ignite. To determine whether flashover occurs before or after the burner intensity is increased (after a 10 min of the test), the ${\mathrm{I}}_{\mathrm{Q}1}$ index and ${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ limit are compared. In Figure 11, it is shown that materials with an ${\mathrm{I}}_{\mathrm{Q}1}$ level below ${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ experience flashover either after a change in burner intensity or without one. The FSR value falling within the [0–25] range does not necessarily indicate the relationship between the FSR and flashover time after the change in burner intensity. This observation is validated by the examples of M03 and M06, where their FSR falls within the [0–25] range, but their ${\mathrm{I}}_{\mathrm{Q}1}$ exceeds the ${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ threshold. However, the heat released from PVC does not meet the criteria set by New Zealand regulations for the first group because its index ${\mathrm{I}}_{\mathrm{Q}1}$ is greater than the integral limit ${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$, as observed in Figure 11.

From Figure 12 describing the ${\mathrm{I}}_{\mathrm{Q}2}$ index with the ignitability index, the Kokkala model classifies combustible materials with the 2 min limit of flashover time, to some extent, into Group 3 and Group 4. Additionally, materials exceeding the 2 min limit in Group 4 exhibit the highest THR (in the case of M14) or the shortest ignition time (in the case of M19). According to Kokkala’s model, materials such as XPS fall between the limits of 2 and 12 min in flashover time, whereas real-life room corner tests have shown flashover times of less than 2 min [60]. It is important to note that the materials that fall under Group 4 are more innovative than the ones tested by Kokkala during the development of these stochastic indices and integral limits, ${\mathrm{I}}_{\mathrm{Q}10\min }$, ${\mathrm{I}}_{\mathrm{Q}2\min }$ and ${\mathrm{I}}_{\mathrm{Q}12\min }$.

The range of validity for these stochastic indices and limits depends on the materials tested during that period [56]. The materials located below the 12 min limit, except for M04 and M05, are not in a range covered by the Kokkala model. The New Zealand regulations do not consider the case “${\mathrm{I}}_{\mathrm{Q}1}$ > ${\mathrm{I}}_{\mathrm{Q}10\mathrm{m}\mathrm{i}\mathrm{n}}$ and ${\mathrm{I}}_{\mathrm{Q}2}$ < ${\mathrm{I}}_{\mathrm{Q}12\mathrm{m}\mathrm{i}\mathrm{n}}$” and the materials in this interval are assimilated to Group 3. This criterion is absent due to the linear interpolation of the 12 min limit, which was adapted from the results of the Kokkala studies [56]. In the Kokkala tests, all materials classified in Group 3 were considered to be between the 2 and 12 min limits. A larger quantity of room corner test data for a broader range of materials is warranted to further strengthen the limits of the Kokkala model.

Regarding the European classification, materials classified as noncombustible according to CAN/ULC-S114 or CAN/ULC-S135 are equivalent to the European noncombustible classes A1 and A2. Additionally, the THR shown in Table 6 and calculated using the oxygen consumption method complies quite well with the ISO 1716 criteria presented in Table 3. In examples, materials M01 and M02 have, respectively, a THR of 0.62 MJ/m² and 0.42 MJ/m², which are below the 1,4 MJ/m² criterion for class A1. Similarly, materials M04 and M05 have a THR of 2.21 MJ/m² and 2.37 MJ/m², which are below the 4.0 MJ/m² criterion for class A2. When comparing the gross heat of combustion of the ISO 1716 criteria, determined by the rise in temperature and including the latent heat of vaporization of water, to the THR using the gas analysis method, the difference is quite reasonable. The NBCC [1] criterion of 3 MJ/m² appears reasonable compared to the criteria for classes A1 and A2.

Interestingly, the combustibility level of European classes increases with each increase in the FSR of materials, as shown in Figure 13, except for materials M03 and M14 intended for floor applications and therefore tested per CAN/ULC-S102.2 rather than CAN/ULC S102 used for the others. PVC is classified as Bfl under ISO 9239-1, while the wood–plastic composite is classified as Efl, equivalent to class E. Materials with an FSR of zero are equivalent to class A1 noncombustibility in the European classification. However, according to ASTM standard E84 [36], a material that does not spread flame should not be considered noncombustible when tested in the Steiner tunnel.

According to European regulations, materials grouped in classes A2 and B would exhibit FSRs between 15 and 25. Class D, covering woods and bio-based materials, would exhibit FSRs ranging between 40 and 175. Class E covering thermoplastic materials and thin wood fiber insulation covers an FSR range from 125 to 240. In the case of the United States of America (USA), the XPS and EPS have flame spread indices (FSIs) of 10, which will be correlated with the A2 classes of the European regulations. As shown in Figure 6, these materials are far from being noncombustible and not even near being of low levels of combustibility.

However, in the USA, the surface flame spread is evaluated per ASTM E84, which consists of a ceiling-mounted specimen, regardless of its fire behaviour. This is a significant difference between Canada and the USA for these materials that can melt and/or drip, and a major drawback of ASTM E84. In Canada, XPS and EPS have FSRs of 190 and 240, respectively, which is at the opposite spectrum of the FSI of 10 in the USA. An interesting observation can be made about M14, which has a high FSR of 125 and is classified as a high flammability risk material Efl (equivalent to class E) according to European standards. This classification is due to its high concentration of highly flammable thermoplastic material, which places it in class E. However, M14 performs almost as well as red oak (FSR of 100) in the CAN/ULC-S102 tunnel test, which is somewhat paradoxical considering its higher heat release rate shown in Figure 6 compared to that of red oak shown in Figure 7.

When conducting a performance-based fire safety design, it is essential to note that the classifications outlined in Table 8 may not be as valuable for modelling and calculations. Studies need quantitative input data for modelling and calculations. For instance, while the Canadian classification system utilizing the Steiner tunnel may provide some data, it does not reflect quantitative and/or physical data.

5.2. Comparison of Fire Reaction Tests

The fire reaction of a material is the discipline that characterizes the response to ignite, to spread, or to release the heat of a material subjected to extreme thermal conditions (fire) under controlled and reproducible conditions. Currently, no known test is suitable to fully characterize the fire behaviour of a material representative of realistic fire exposures. However, each test is designed to characterize a specific behaviour in a specific exposure and with high confidence.

5.2.1. Scale

To distinguish between the system classification and material classification, an analogy between structural design and fire safety design can be made. Structural design studies the mechanical behaviour of materials, while fire safety design studies the reaction of materials to fire. A structural design begins by determining the measurable criteria to be achieved (maximum stress, deformation, etc.). Then comes the stage of characterizing the mechanical behaviour on a material scale using mechanical tests on small and medium scales (tests of bending, tension, compression, etc.). It allows a given material to be selected over another before being projected to larger scales using a mechanical performance approach (modeling, etc.). Testing structural behaviour at the system scale is typically uncommon, with perhaps the exception of seismic resisting systems. Fire safety studies, somewhat, are the opposite. They start by testing materials on the system scale (ISO 9705, etc.) and then follow down to the material scale (by stochastic correlation, reduced testing, etc.) before judging the fire risk of a material.

A large-scale fire is a complex phenomenon influenced by several factors, including the distribution of materials in a system, ventilation, and the choice of combustibles. It is a non-linear process that requires a careful consideration of all these factors to reasonably predict the outcomes. However, the material scale represents a controlled environment designed to study a particular behaviour. The cases of the cone calorimeter and the vertical tube furnace are examples where the effects of ventilation and the boundary effects of the test are controlled to characterize the fire behaviour of a material. This is widely acknowledged in numerous standard fire tests, where it is stipulated that these standards do not incorporate all factors required for fire hazard or risk assessments under actual fire conditions. They are only used to measure and describe the behaviour or response of a material to heat and the flame of a material under controlled conditions.

The variability of results obtained in different large-scale tests for material classification highlights the complexity of determining a material’s real contribution to the system scale. This difficulty, exacerbated by differences in testing protocols and conditions for the same large-scale test [61], poses challenges from the scientific perspective and regulatory processes. Indeed, while these topics spark significant research interest, their practical applications in standards and regulations remain limited, potentially leading to a conservative approach towards combustible materials. For example, thermoplastic materials are highly combustible on a material scale, whereas, on a ceiling scale, their ability to melt gives them an advantage in some tests (e.g., Steiner tunnel tests).

In this context, a classification approach based on the intrinsic properties of materials is more relevant to a fire risk assessment. Adopting such an approach enables the risks associated with their large-scale use to be understood and characterized, encouraging the adoption of performance-based fire safety regulations.

As a result, the cone calorimeter test appears to be the most acceptable solution at the material scale, notwithstanding the importance of suitable classification criteria. During the development of the Euroclass in the 1990s, the option of a cone calorimeter was mentioned but was quickly dismissed [15]. There were concerns that cone calorimeter tests would be suitable for complex materials, as well as representative of products in their end-use state. The cone calorimeter is suitable for simple materials but not suitable for the fixings and joints of products. Without a consensus among the European members, the SBI was developed and highly criticized by the scientific community. The current test does not effectively distinguish the ignition time of the material or differentiate the impact of the system from that of the material. Babrauskas [62] reported that the SBI test is unique because it was mainly designed by government regulators rather than by scientists.

The selection of a preferred test at the system scale is complex, with tests such as the room corner test, façade tests, and tests in specific configurations such as Steiner tunnels, each having advantages and limitations, depending on the application’s requirements. In this context, the judgment of the fire safety engineer is the optimal choice for assessing risks on a large scale. The fire safety engineers’ expertise allows them to comprehend the subtleties of different tests and apply them suitably to the project’s requirements. They consider material properties, actual fire conditions, architectural design, and regulations to conduct a comprehensive risk assessment and design integrated solutions that ensure occupants’ safety and asset protection.

5.2.2. Cost and Size

The cost of testing materials depends largely on the size of the specimen. Material-scale testing is generally less expensive than large-scale testing when equipment and maintenance costs are not considered. This affordability of material-scale tests makes it easier to test more specimens and ensures repeatable and reproducible measurements. In contrast, large-scale specimen testing, such as the room corner test, is more expensive and complex. The system allows for greater possibilities, including the material distribution. Research [61] comparing room corner test protocols has shown that the results are influenced by factors such as the burner power and material distribution, potentially affecting the regulator’s assessment of materials.

In the context of the fire classification of materials, small-scale and low-cost size specimens are the most appropriate choice for industrialists and regulators. For large-scale specimens, expert judgment on fire safety is typically required to select the scenario according to its application. During a project design, materials classified in a particular scenario will not necessarily be used for the function for which they were classified and per the same conditions as in the test.

Determining the impact of the size of a test specimen on the measurement can prove to be complex, as the specimen’s dimension is typically adjusted based on the specific characterization objective of the test. It is essential to recognize that each test has specific parameters, and the specimen size is often deliberately chosen to meet the requirements of the experiment and/or the test apparatus.

In the case of the Steiner tunnel, the dimensions of the specimen and the tunnel are not established to characterize the behaviour of various materials, but rather to provide a comparative measure specific to the behaviour of red oak. The dimensions and test protocol are maintained to ensure the consistency and comparability of results obtained in the test with red oak. Comparing flame spread between wood species may be relevant, but when it comes to fire safety, comparing flame spread between red oak and a melting thermoplastic material is highly questionable.

Thus, although the question of the influence of the specimen size on the measurement remains challenging, exceptions such as the Steiner tunnel test underscore the need to carefully consider the specific parameters of each test to interpret the results obtained from such tests correctly.

5.2.3. Measurement

In the tests discussed in this article, it is essential to note that there are three main categories of measurements. These are the rise in temperature, the production and consumption of gas, and the visual monitoring of the flame spread phenomenon.

The vertical tube furnace and bomb calorimetry tests both involve measuring temperature rises. The vertical tube furnace measures the temperature increase caused by the combustion of a material, with specific measurements stipulated in the test standard. The relative mass loss is also measured before and after the test. The bomb calorimeter calculates the gross calorific value of a material by measuring water temperature rises resulting from its combustion.

The tests conducted with the cone calorimeter, room corner, and SBI are designed to measure gas consumption and production. The gas analyses involve monitoring the amount of oxygen consumed during a material’s combustion to determine the heat release rate. A gas analysis can also be performed to calculate the heat release rate based on carbon dioxide, monoxide, and water vapour production. The cone calorimeter is used to measure the heat release rate of a material, while the room corner and SBI tests evaluate how a material combusts in specific scenarios. Measuring the heat release rate in a scenario can be challenging due to the impact of the system on the measurement. The scenario’s effect varies depending on the material being tested.

The Steiner tunnel tests, ISO 11925-2 and ISO 9239-1, involve the operator measuring the flame front along a system visually. In the Steiner tunnel test, the operator monitors the flame spread until it reaches the end of the tunnel. However, limited visibility of the flame can make it difficult to monitor its progress. In such cases, the standard suggests using the time taken for the temperature of the thermocouple at the end of the tunnel to exceed 527 °C. However, this method has some limitations because the tunnel’s dimensions are restricted. As a result, the measured temperature does not allow for differentiation between the temperature of the flame front and that of the combustion gases.

5.2.4. Performance

According to Angus Law’s taxonomy of UK fire tests [14], tests can be classified based on the ease with which users can apply the results without negative consequences for fire safety. Law’s study presented three categories: non-representative tests, model tests, and technological proof tests.

Non-representative tests do not replicate actual building fire scenarios. Their thresholds are conservative enough that users do not need to worry about whether the test applies to their intended application. Model tests rely on “models” of expected fire scenarios, and so users must be confident that the model is similar enough to their application. Technological proof tests provide a more realistic test of a real building system, but users must analyse the similarities between their test and the real building before applying the results.

Despite including temperature rise and mass loss analyses in the CAN/ULC-S114 noncombustibility test, the test does not allow access to properties relevant to fire dynamics, such as the heat release rate from a material. This makes the CAN/ULC-S114 a binary test preferred for determining the noncombustibility of a homogeneous material despite the exposure of all its faces, which is an unusual exposure scenario in a real fire.

In the case of the ISO 5660 cone calorimeter test, the standard combines the two analyses of mass loss and gas analysis to provide access to a wide range of highly significant properties relevant to fire dynamics, such as the heat release rate, smoke production, and toxicity of fire effluents, among others.

It should be noted that many countries are beginning to reform and amend their classifications to introduce the cone calorimeter into their legislation (NZBC, etc.) with different acceptance criteria [63]. It is important to remember that CAN/ULC-S135 specified in the NBCC is not to be confused with the generic version of ISO 5660. CAN/ULC-S135 is only intended to evaluate materials exhibiting a low level of combustibility at a given heat flux level. It is not intended to evaluate degrees of combustibility.

Although the cone calorimeter has advantages, it also has several disadvantages. These include sensitivity to low-emissivity materials that reflect part of the radiant heat, the absence of assessment of a material’s fire stability in terms of integrity and resistance to melting, the influence of specimen holder edge effects on test conditions [64,65], and the lack of consensus on the classification criteria to be adopted and the limitations of correlation models.

Unlike the other tests, the Steiner tunnel test (CAN/ULC-S102 [35]) is not based on one of the previous methods. Although the test represents a scenario more like a fire in a concealed space rather than in a corridor, the test measures flame spread along the cavity and obscuration with a thermocouple at the end of the tunnel. Despite some advantages of the test, the test concedes many critical disadvantages for a fire classification [19]; a review by the authors of the evolution of fire classification in Canada highlighted the limitations of the Steiner tunnel method.

The single burning item (SBI) test is a relatively recent fire test whose primary intent is to harmonize exchanges within the European Union. The SBI test does not provide any information about the material’s behaviour, such as the ignition time, heat release rate, etc. It should be noted that the SBI test procedure involves placing the burner in a corner between two walls without a ceiling. This creates an artificial system that deviates significantly from real-life scenarios. While the walls undoubtedly play a crucial role in the lateral spread of flames, current regulations do not consider their effects. Furthermore, it is essential to note that the effect of the ceiling on flashover [66] is not considered when assessing scenarios using the SBI test.

The decision to remove the ceiling from the SBI test is part of an approach aimed at resolving the problems inherent in the low repeatability and reproducibility of the measurements [67], combined with the need to rationalize the dimensions of the equipment used. This strategic choice stems from a critical assessment of the constraints encountered during the implementation of the test, aimed at improving the reliability and accuracy of the results. Another limitation of the European test-based classification is that the classification was correlated to the set of materials tested during the development process of the ISO 9705 test method. The relevance of this test for innovative products remains questionable when the SBI test itself only managed to predict 90% of the ISO 9705 test results (26 out of 30 specimens tested in both tests [68]). Therefore, this test alone is insufficient to classify a material but must be associated with a second test (ISO 1716 or ISO 11925-2), as shown in Table 3.

ISO 9705, in many situations, is presented as a full-scale reference test. Its cost and the choice of ventilation conditions for the system significantly influence its adoption. The ISO 9705 test provides results for a particular configuration of a material, which means that any configuration change may modify the classification of the material and could limit the possible design configurations without performing additional testing. Several models [69,70], such as the Kokkala model, have been developed to establish a correlation between room corner results and the cone calorimeter results. Although a correlation is used to establish a relationship between different tests, it is essential to remember that a correlation does not imply causality. A correlation measures the extent to which two variables are linearly related without making statements about cause and effect. Conversely, causality is a statistical measure of the relationship between two variables where one variable affects the other. Causality happens when the value of one variable changes in response to a modification in another variable.

Recent studies [59] have shown that the correlation between the Kokkala model’s cone calorimeter and the room corner classification is not always satisfactory. Therefore, the link between the group number detected from ISO 9705 and the group number expected by the Kokkala method using ISO 5660 data is nowadays questionable. Interestingly, the results of the fire classification of materials using the Kokkala model according to ISO 5660 are more severe than those obtained in a full-scale test according to ISO 9705 [59]. The Kokkala correlations are primarily designed for uniform materials that do not melt or shrink when exposed to flames. Therefore, they may not be suitable for specific products. The main advantages and disadvantages of the various test methods described herein are summarized in

Table 10. Comparison in terms of the combustibility of the test selected.

Test	Advantage	Disadvantage
Vertical tube furnace “Unrepresentative tests”	Controlled test environment Insights into material properties (temperature rise, mass loss, and ignition time) Low cost Material representativeness Repeatability and reproducibility	Binary result (pass/fail) Absence of fire dynamics Sensitivity to heterogeneous and/or composite materials Exposure of all faces System representativeness Product representativeness Absence of fire effluent measurements (smoke and toxicity)
ISO 1716 “Unrepresentative tests”	Controlled test environment Material representativeness Insights into material properties (gross heat of combustion) Repeatability and reproducibility Low cost	Absence of fire dynamics System representativeness Product representativeness Absence of fire effluent measurements (smoke and toxicity) Product representativeness
ISO 5660 “Unrepresentative tests”	Controlled test environment Low cost Insights into material properties (HRR, fire effluents, ignition, mass loss, etc.) Repeatability and reproducibility Material representativeness	No classification criteria Thickness limitation Constant radiant heat flux exposure Sensitivity to material emissivity Product representativeness
CAN/ULC-S102/S102.2 “Model tests”	Accounts for thermal inertia Material representativeness Comparative test Model tests	Influenced by airflow turbulence Controlled test environment Unrealistic fire scenario Absence of fuel input Relevance of FSR and SDC ratings Sensitivity to material type Repeatability and reproducibility Test not recommended on its own Cost Product representativeness
EN 13823 “Model tests”	Sample representativeness Controlled test environment Material representativeness	Cost Absence of ceiling influence System effect Unrealistic scenario Correlation with ISO 9705 (around 90%) Product representativeness
ISO 9705 “Technological proof tests”	Controlled test environment Reference scenario test Product representativeness	Cost Bidirectional flow on opening System effect Representative of one fire scenario Material representativeness

.

6. Conclusions

Comparing the fire classifications of materials highlights the diversity of methods and results obtained in different testing contexts. Each classification method brings its advantages and limitations, which underlines the importance of choosing the most appropriate method according to the specific objectives of the fire risk assessment of materials. It is essential to consider the intrinsic characteristics of materials and the requirements and standards applicable in the industries and regulations concerned to ensure accurate and reliable assessments of their reaction-to-fire performance.

Reaction to fire is tightly regulated but often not well quantified. Apart from comparing wood species, the Steiner tunnel test only allows some materials to be classified. With its sensitivity to the different behaviours of materials and its non-quantifiable measurement, the Steiner tunnel flame spread rating cannot realistically determine the degree of combustibility of a material on its own, which makes it difficult to correlate these results with other tests.

The Japanese classification represents a classification of the degree of noncombustibility of materials. As a result, the cone calorimeter (according to the Japanese Building Standard Law) is similar to the CAN/ULC-S135 test in the Canadian classification used for classifying materials exhibiting low levels of combustibility. The European classification system combines multiple tests of different scales. Material-scale tests are preferred for noncombustible or highly combustible materials, whereas a system scale is considered for moderately combustible materials. Although the single burning item (SBI) test provides valuable insights into the fire performance of materials under controlled conditions, its relevance and representativeness in real-life scenarios are uncertain. In the case of the New Zealand classification, the classification based on the Kokkala model divides the fire reaction of materials globally into two main groups: noncombustible (Group 1) and combustible (Group 3). In addition, some materials classified according to ISO 9705 and ISO 5660 using the Kokkala model do not provide the same group correspondence. Thus, the stochastic limitations of the Kokkala model require revision based on a more extensive range of room corner test results.

A comparison of tests in scale, cost, measurement, and fire performance indicates that the most preferable method of classifying the combustibility of materials is to use the scale of the material with access to quantifiable fire dynamics measurements at a low cost. Using large-scale quantifiable tests makes it possible to obtain a system’s response in a single scenario at a high cost. However, it does not allow an accurate analysis of a material’s contribution, and its relevance to the classification of general-purpose materials is particularly complicated for non-performance-based codes. Currently, the most advanced test that can best characterize the material scale is the cone calorimeter, but the criteria for assessing the fire risk are still undefined. A revision of the building codes on the material scale with access to the degree of combustibility of the materials is necessary to establish the transition from a prescriptive to a performance-based regulation. More research is needed on the criteria to be considered in a cone calorimeter test to classify materials’ combustibility degrees.