Multi-Scale Characterisation of the Fire Hazards of Timber Cladding

Abstract

Timber-clad facades, traditionally prevalent in North America and Scandinavia, are gaining popularity in central Europe and the UK for applications beyond low-rise buildings. Timber differs from typical cladding materials, such as masonry, due to its non-uniformity, combustibility, and moisture sensitivity, requiring unique design considerations to manage these characteristics. This paper investigates the fire hazards associated with timber cladding, particularly focusing on thermally modified timber, motivated by the 2019 Samuel Garside House fire in the UK. The study aims to address five key research questions: (1) the impact of thermal modification on external fire spread hazards, (2) the fire risk associated with slatted timber configurations, (3) the effectiveness of fire-retardant treatments, (4) the correlation between small-scale standard tests and large-scale behaviours, and (5) the adequacy of current fire safety guidance in addressing these hazards. The experimental campaign involved four timber sample variants: (i) virgin timber, (ii) new thermally modified timber, (iii) aged thermally modified timber, and (iv) fire-retardant-treated thermally modified timber. These samples were tested across four different methods, including the single-flame source test, mass loss cone test, single burning item (SBI) test, and an intermediate-scale test. Results indicated that thermal modification slightly increased the peak heat release rate (HRR) compared to virgin timber. The configuration of timber slats significantly impacted HRR, with vertically oriented slats demonstrating higher HRR than horizontally oriented flat cedar cladding. Fire-retardant treatments substantially reduced HRR, achieving Euroclass B in vertical slatted configurations. However, the long-term efficacy of these treatments under ageing and weathering conditions remains unexplored. This research underscores the need for clarifications in the guidance in timber cladding design, considering the observed fire hazards in different slat configurations and the efficacy of fire-retardant treatments.

1. Introduction

Timber-clad facades are commonly used in North American and Scandinavian countries, while there is an increasing interest in Central Europe and the UK in their application beyond low-rise buildings [1]. Compared to typical cladding materials like masonry, timber is distinctly different because it is non-uniform, combustible, and moisture sensitive [2]. These differences necessitate a distinctive approach in the façade design—addressing changes in moisture content while managing the non-uniformity and combustibility of the timber material.

Timber cladding comes in various types, differentiated primarily by the type of wooden material, profile styles, and board arrangement. The wood can either be softwood, hardwood, or modified wood. The typical profiles can be tongue and groove, rectangular, feather edge, shiplap, or parallelogram. Meanwhile, the board can be arranged horizontally, vertically, or diagonally [3].

Despite the design challenges, timber offers benefits that make it a material of choice for facades. Firstly, it is practical for being strong but lightweight and easy to work with, among other things. Secondly, its aesthetics provide a sense of connectivity with nature. Thirdly, it is environmentally friendly with its renewability, recyclability and ease of disposal, as well as its raw tree material’s CO₂ absorption capability [3]. Finally, modern treatment techniques are said to have enhanced its durability and resistance to weather, pests, and decay. With proper maintenance, timber cladding can withstand the elements for decades.

1.1. Thermal Modification of Timber

One of the possible treatments to enhance the durability and stability of timber is thermal modification. This process, also known as heat treatment, fundamentally changes the physical and chemical properties of wood, making it more suitable for various applications, especially in environments where wood is exposed to harsh conditions [4].

The process of thermal modification includes a heating phase, where the timber is heated in a controlled chamber in a temperature range from 160 °C to 230 °C. This heating is performed in a low-oxygen environment to prevent the wood from burning, and it can be achieved through partial vacuum, injecting an inert gas like nitrogen, or injecting steam. The duration of the heat treatment varies depending on the desired properties and the type of wood. It can last from a few hours to a couple of days.

The changes arising from the thermal modification process are set out below:

• Chemical Changes: The high temperatures break down hemicellulose (a type of carbohydrate in wood), reduce the amount of hygroscopic (moisture-absorbing) sugars, and cause chemical modifications in cellulose and lignin. These changes reduce the wood’s ability to absorb moisture.
• Physical Changes: The wood becomes less prone to swelling and shrinkage because of its lower equilibrium moisture content. This leads to increased dimensional stability, which is crucial for applications like flooring or cladding where changes in moisture can cause warping or deformation.
• Colour Alteration: Thermal modification often darkens the wood, giving it a richer, more uniform colour. This colour change is consistent throughout the wood, not just on the surface.
• Decreased Weight: The process may result in a slight reduction in the wood’s weight due to the loss of volatile compounds and moisture.

The process of heat-modifying timber is intended to induce improved durability performance, specifically the resistance to decay. This is achieved by reducing the moisture content, making the wood less attractive to fungi and insects, leading to enhanced decay resistance and longevity. However, the heat-modifying process has some trade-offs, such as cost and changes in some mechanical properties depending on the wood species, as presented in [5,6].

1.2. Research Objectives

Despite the many benefits that timber cladding brings to the built environment, it also introduces a fire hazard that must be addressed. In England, the fire safety hazards associated with timber cladding can be considered to fall within the remit of Regulation B3—Internal fire spread (structure) and B4—external fire spread [7]. In the case of the former, timber cladding can provide a means by which compartmentation is circumvented, leading to a larger fire within the building of origin. This might be through fire spread over or within the cladding system. Regardless, the outcome is a larger fire within the building. From an external fire spread perspective, the timber cladding can facilitate fire spread to and over the walls and, owing to the increased area of burning, influencing the prospect of fire spread from one building to another.

The significant fire that happened in June 2019 at Samuel Garside House, Barking, in the UK, is an example of the fire hazard introduced by this type of cladding. As illustrated in Figure 1, this complex of buildings had several balconies and façade components composed of vertically slatted timber. Post-fire investigations suggest that the timber was thermally modified Scots pine, and the analysis of video footage established that in some instances the fire spread horizontally at a rate of 50 mm per second. Eventually, the fire completely consumed all the combustible content in eight flats, as reported in [8], meaning that compartmentation was circumvented. Additionally, there was noticeable heat damage to other properties across the street [9].

This fire, together with concerns raised through the Collaborative Reporting for Safer Structures (CROSS) in the UK [11], was the main driver for the UK government to promote a research programme that would help quantify the risks and challenges posed by timber cladding. This paper represents part of this research programme, and the main research questions investigated are as follows:

• Does the thermal modification of timber result in a greater external fire spread hazard than virgin timber, and how does this change with ageing/weathering?
• Does the vertically slatted configuration of timber cladding, the like of which was observed at Samuel Garside House, result in a greater external fire spread hazard than more traditional flat/lapped without gaps timber cladding?
• To what extent can the external fire spread hazard associated with timber cladding be mitigated through the application of fire-retardant treatments?
• To what extent does SBI scale standard testing correlate with the behaviour at a larger scale with more realistic heat fluxes?
• Is the English guidance (ADB) [12] able to capture and differentiate any such hazards that might arise from material changes (e.g., thermal modification) or configurational changes (e.g., slats).

2. Methodology

To answer the research questions presented in Section 1.2, the experimental campaign was set up with four different types of samples and with four testing methods across different scales [13].

2.1. Sample Types

To align with the timber species seen at Samuel Garside House, Pinus sylvestris (PS) was adopted throughout (Scots pine). Four different sample variants have been identified to address the research questions. These are described in

Table 1. Sample description.

Nomenclature	Description	Density (kg/m3)	Sample Sourcing and Characteristics
PS-Virgin	Virgin timber	485	Sourced from a Scottish sawmill. The trees were cut down in October and collected in November 2021. The slats were stored in a conditioning room until they had reached constant mass, as defined by BS EN 13238:2010 [14].
PS-TM-New	New thermally modified timber	425	Sourced from a timber cladding specialist supplier, based in Birmingham, from Scandinavian forests certified under the Pan European Forestry Certification (PEFC) scheme. The thermal modification process achieved a durability rating ‘D’ by following the methodology presented in [13]. The material was introduced to a conditioning room until it reached constant mass, as defined by BS EN 13238:2010. This PS-TM-New was not exposed to weathering.
PS-TM-Aged	Aged thermally modified timber	444	A ‘batch’ (80 lengths at approximately 1.5 m length) of the PS-TM-aged timber slats was collected from Barking Riverside redevelopment, the same development where the Samuel Garside House fire occurred, in May 2020. It was estimated from Google Earth imagery that the slats had been weathered for approximately seven years. The slats were stored in a conditioning room until they had reached constant mass, as defined by BS EN 13238:2010.
PS-TM-FR	Thermally modified timber treated with fire retardant	503	These were the same PS-TM-New samples. Once the material was imported to England, the supplier arranged for it to be chemically treated using chemical(s) approved by the Wood Protection Association (WPA). The fire-retardant product was Sentrin FRX. The resulting grade was described as external, Leach Resistant (LR) grade. This was achieved through vacuum impregnation, with subsequent kiln drying and heat curing. It should be noted that whilst the Treatment Certificate for the product states that the treatment was of such ‘loading’ necessary for the thermally modified material to achieve a Euroclass B classification under BS EN 13501-1:2018 [15], it is not known what end-use arrangement was envisaged (by the supplier) for the BS EN 13823:2020 [16] (Single Burning Item) test. The slats were stored in a conditioning room until they had reached constant mass, as defined by BS EN 13238:2010.

.

2.2. Testing Methods

The chosen testing methods provide a means to characterise differences in fire behaviour at the material and system scale. Additionally, the tests referenced in the English guidance ADB (BS EN 11925-2:2020 [17] and BS EN 13823:2020) are also adopted to assess their utility in classifying timber cladding systems. The testing methods are presented as follows:

2.2.1. BS EN 11925-2:2020 Indicative Single-Flame Source Test [17]

This is a material scale test, and it is a prerequisite for progressing to Euroclass classification in the grades B to D, in accordance with EN 13505-1:2018 [15].

The method specifies a test for determining the ignitability of products by direct small-flame impingement using vertically oriented test specimens. The flame is applied either to the surface or the edge of the specimen. Then, the vertical flame spread is measured. The determination of the production of flaming droplets depends on whether the filter paper placed beneath the specimen ignites.

One run of tests was conducted, comprising one surface exposure and one edge exposure to the flame (i.e., two tests). Both tests were conducted at an exposure period of 30 s.

2.2.2. BS ISO 5660-1:2015+A1:2019—Mass Loss Cone [18]

This is also a material scale test. Two types of experiments were conducted with this testing rig for all the samples, except the FR-treated samples (PS-TM-FR). This is because the FR treatment expanded as heated, which introduced changes to the boundary condition on the surface of the sample, complicating consistent and repeatable testing.

Samples of approximately 92 mm × 99 mm were placed in aluminium foil ‘boats’, as described in BS ISO 5660-1:2015+A1:2019 [18].

The first type of experiment consisted of exposing the samples to a constant incident heat flux of 35 kW/m². Each sample was tested three times. These experiments allow a direct comparison of the burning rate and heat release rate across the samples. Thus, any differences in burning characteristics arising from thermal treatment, ageing, etc., can be isolated from the impact of the configuration.

The second type of experiment aimed to establish the critical heat flux for ignition. To achieve this, eleven samples were prepared. A high heat flux was selected (for the first run), which was known (by experience) to ignite the surface of the sample and give rise to sustained flaming. The process was repeated with heat flux reduced (in 5 kW/m² decrements) to the point where no ignition and no sustained flaming was observed. Thereafter, the heat fluxes were reduced to finer 1 kW/m² differentials to determine the critical heat flux for ignition and sustained flaming within 15 min of exposure.

2.2.3. BS EN 13823:2020 (Single Burning Item, SBI) Test [16]

This is a system test, and it is after completing and passing the single-flame source test for the purpose of obtaining a Euroclass. This test provides the fire growth index (FIGRA), smoke production rate, and droplet production characteristics of building products (excluding floorings) when exposed to a fire from a sand-box burner supplied with propane. This test is necessary to achieve an external surface classification Euroclass in grades B to D.

The specimen is mounted on a trolley that is positioned in a frame beneath an exhaust system. The reaction of the specimen to the burner is monitored instrumentally and visually. Heat and smoke release rates are measured, and physical characteristics are assessed by observation.

In this experimental programme, each sample was fixed into the test rig to closely resemble the end-use arrangement at Samuel Garside House. The ‘slats’ of the façade were ventilated (to the front, sides and rear). There were 68 mm gaps between the faces of adjacent slats, i.e., slats were installed at 110 mm centres (note each slat was nominally 42 mm wide). Angle irons were 60 mm × 60 mm × 8 mm thick, with a localised cavity behind the angle iron 20 mm wide. The slats stood on the trolley tight against the head and base welded U-profiles. Both wings were secured from behind the U-profiles with L-profile clamps in the C-shaped profiles, i.e., the SBI test equipment. See Figure 2 and Figure 3.

2.2.4. BRE Intermediate Scale Test [19]

This is a system-scale test, and it was designed by BRE Global’s Fire Engineering Team for a separate fire research project. The detailed protocol of the test is explained in [19].

This test is intended to provide information about flame spread over the external surface under representative heat fluxes of an external plume emerging from a window (i.e., higher heat fluxes than in the SBI test). This test permits the measurement of additional parameters, such as heat release rate via oxygen depletion calorimetry.

The experimental rig was constructed from a hot-rolled steel frame made up of rectangular hollow steel sections. Lateral mild steel angles were fixed to the frame, onto which the timber cladding was installed (Figure 4).

The cladding was subject to fire exposure via a crib, consisting of 40 sticks of 50 × 50 × 500 mm softwood, placed in contact with the centre of the experimental sample. Through calibration experiments undertaken by BRE (explained in the detailed protocol), the crib was shown to induce an incident heat flux of between 45 and 75 kW/m² on a non-combustible reference sample, at a point 1 m above the crib (on the centreline) and a peak heat release rate of ca. 300 (±20) kW.

In terms of measurement, the rig was instrumented with thermocouples at various locations across the sample. Additionally, the heat release rate (HRR) from the combination of the cladding sample and the crib was monitored through oxygen consumption calorimetry, with the rig located below an extractor hood. The HRR is the focus parameter for this research.

3. Results

3.1. Results from the Single-Flame Source Test

A summary of the results of the single-flame source tests is presented in

Table 2. Summary of results from BS EN 11925-2:2020 indicative single-flame source test.

	PS-Virgin	PS-TM-New	PS-TM-Aged	PS-TM-FR
Flame spread ≤ 150 mm within 60 s	Yes	Yes	Yes	Yes
Ignition of filter paper	No	No	No	No
Threshold achieved	Class B-E	Class B-E	Class B-E	Class B-E

.

Additionally, Figure 5 presents how the different sample types looked after conducting the tests.

Whilst this test produces no quantitative outcomes, a qualitative review of Figure 5 indicates that there were varying extents of flame spread across the different samples, with PS-TM-Aged showing the most and PS-Virgin the least.

3.2. Results from the Mass Loss Cone Test

Figure 6 plots the results for the experiments conducted under a constant heat flux of 35 kW/m². It presents the relationship between time and three metrics from the mass loss cone test: heat release rate, heat of combustion, and mass loss rate. These are plotted in 60 s increments relative to ignition and reflect the transient phase of the burning of wood as the char develops and the mass loss rate tends towards a steady value. This was observed to take ca. 6 min from ignition.

Table 3. Time to ignition for samples under different heat flux exposure in the cone.

Heat Flux [kW/m2]	Time to Ignition [s], ‘NI’ Indicates No Ignition. ‘-‘ Indicates Not Tested at This Exposure Heat Flux
	PS-Virgin	PS-TM-New	PS-TM-Aged
20	398	188	121
19	-	-	-
18	NI/648/862	631	-
17	432/NI/NI	NI	-
16	NI	-	-
15	NI	NI	860
14	-	-	NI
10	-	-	NI

summarises the results for the second type of experiments where the samples were exposed to different heat fluxes as per Section 2.2.2. The table presents the time to ignition in function of the incident heat flux. Where multiple tests are undertaken, these are separated by “/” in the table. One of the timber samples, PS-Virgin exhibited a cross-over, whereby the lowest heat flux at which ignition occurred was lower than the highest heat flux at which no ignition occurred. It was therefore necessary to conduct triplicate runs at each heat flux to determine an average time to ignition.

A summary of the experimental results obtained with the mass loss cone is presented in

Table 4. Summary of parameters for the three sample typologies tested in accordance with BS ISO 5660-1:2015+A1:2019.

Parameter	Unit	PS-Virgin	PS-TM-New	PS-TM-Aged	Heat Flux Exposure
Specimen thickness	mm	43	43	42	Constant 35 kW/m2
Specimen initial mass	g	191.4	159.7	175.0
Time to ignition	s	43	35.3	35.7
Total heat release	MJ/m2	97.4	106	111.2
Peak heat release rate	kW/m2	141.2	151.3	148.5
Critical heat flux for ignition	kW/m2	17.7	17.5	14.5	Varying

. This includes the results for both types of experiments; the experiments that were conducted under a constant heat flux and a varying heat flux.

3.3. Results from the SBI Test

Figure 7 compares the four sample types in terms of heat release rate with time. For comparison against the classification criteria, the FIGRA limits for Class B to D materials are indicated as black hatched lines. Where the HRR plot remains below a given line, it would achieve the applicable classification, i.e., all samples except that treated with a fire retardant exceed the FIGRA for Class D.

The PS-TM-FR sample had such a low HRR that it is barely distinguishable from the x-axis in Figure 7.

Table 5. Summary of SBI results.

Parameter	Unit	PS-Virgin	PS-TM-New	PS-TM-Aged	PS-TM-FR
Fire growth rate—FIGRA(0.2 MJ)	W/s	-	-	-	37.1
Fire growth rate—FIGRA(0.4 MJ)	W/s	855	1283	1187	-
Total heat release—THR600 s	MJ	75	88	80	2.3
Smoke growth rate—SMOGRA	m2/s2	4.5	19.1	5.1	6.9
Total smoke production—TSP600 s	m2	42.5	47.4	30.8	65.8
Flaming particles within 600 s	-	Yes	No	No	No
Flaming > 10 s within 600 s	-	Yes	-	-	-
Lateral flame spread reaches sample edge	-	No	No *	No *	No
FIGRA for Class B	W/s	120
THR600 s for Class B	MJ	7.5
FIGRA for Class C	W/s	250
THR600 s for Class C	MJ	15
FIGRA for Class D	W/s	750
THR600 s for Class D	MJ	No limit
Classification achieved	-	E-s1, d2	E-s1, d0	E-s1, d0	B-s2, d0

* Test terminated prematurely due to high temperatures in exhaust duct.

summarises the SBI testing outcomes for the four samples, noting their final classification. It should be stated that no further differentiation is provided beyond Class E, i.e., anything performing worse than Class D is denoted Class E, meaning substantial differences in fire hazard can exist under the worst classification.

3.4. Results from the BRE Intermediate Scale Test [19]

Figure 8 shows the extent of fire spread and consumption of slats for each of the four experiments, in 5 min increments from ignition of the timber crib.

The HRR with time for the four sample types is shown in Figure 9. All plots are offset such that 0 min corresponds with the time of ignition of the timber crib. The peak heat release rate of the crib in isolation is ca. 300 (± 20) kW. Any heat release above this threshold can be attributed to the sample.

Qualitatively, it is observed that no significant difference exists between PS-Virgin, PS-TM-New and PS-TM-Aged. PS-TM-FR results in substantially reduced HRR, i.e., by circa a factor of two considering the timeframe of 5 to 15 min from ignition. Nearly all the heat release rate for the PS-TM-FR sample can be attributed to the timber crib, which is estimated to have a peak heat release rate of ca. 300 (± 20) kW.

A summary of the BRE intermediate-scale test results is given in

Table 6. Summary of intermediate-scale results.

Parameter	PS-Virgin	PS-TM-New	PS-TM-Aged	PS-TM-FR
Peak HRR [kW]	792	975	893	382
Time to peak [s]	999	993	1143	336
Average HRR during first 8 min [kW] (including crib)	362	433	437	249

.

4. Analysis and Discussion

The discussion of the results is provided at two scales. Namely, that of the material, informed by testing in the small-flame test and the mass loss cone, and that of the system, informed by testing in the SBI or BRE intermediate-scale rig.

Ahead of the analysis and discussion, it is important to note that there is potential variability in wood properties even within the same tree species and batch. To ensure that the selected samples had comparable prerequisites, such as similar densities and latewood ratios in their initial states, the material-scale tests were conducted with multiple repeats to capture any inherent variability in the wood samples.

Additionally, the system-scale tests utilised larger and multiple samples simultaneously, which further helped in minimising the impact of material variability on the overall test results. As a result, the data collected from the tests already account (to some extent) for potential material variability.

4.1. Material Scale Discussion

4.1.1. Ignition Characteristics

According to the results presented in Section 3.1, all four timber samples would have been permitted to progress to classification in the Euroclass bandings B to E. Of the non-fire-retardant treated samples, qualitatively, it can be said that virgin timber exhibited the lowest extent of flame spread, followed by the new thermally modified sample and then the aged thermally modified sample. However, differences in this extent of flame spread were nominal.

Section 3.2 presents the results for the critical heat flux for piloted ignition for the non-fire-retardant-treated samples. The results for PS-Virgin, PS-TM-New, and PS-TM-Aged were 17.7, 17.5, and 14.5 kW/m², respectively. This is generally greater than that quoted in Drysdale [20] for Red Cedar (13.3 kW/m²), Radiata Pine (12.9 kW/m²), and Douglas Fir (13 kW/m²).

For the new thermally modified cladding material, the critical heat flux for piloted ignition appears to be unaffected by the process, with the virgin material having nominally the same value.

Ageing of the sample appears to have reduced the critical heat flux for piloted ignition by ca. 3 kW/m². This could be for several reasons related to weathering and heat cycles, such as physical degradation, leading to fissures, or further changes in the timber composition, as documented in studies exploring the fire behaviour of historic timber structures [21,22]. However, as aged virgin timber samples have not been studied in this experimental programme, it is not possible to determine if thermally modified timber is unduly affected by ageing relative to virgin timber.

4.1.2. Heat Release Rate Characteristics

Mass loss rate and heat release rate when subject to an incident heat flux of 35 kW/m² have been reported in Section 3.2 for non-fire-retardant treated samples. This indicates comparable trends for both parameters. Like the critical heat flux for piloted ignition, there was no appreciable difference between the new virgin and thermally modified samples. The aged, thermally modified timber produced a slightly higher heat release rate per unit area and, correspondingly, had a higher heat of combustion. It also produced the highest total heat release rate over the course of the 30-min test. These results question the adequacy of testing virgin samples, whether thermally modified or not, to characterise the burning behaviour of cladding in a real building scenario, where the timber may be exposed to the elements for several years.

4.2. System Scale Discussion

4.2.1. SBI Results Discussion

All non-fire-retardant-treated samples exceeded the FIGRA for Class D and, therefore, were classified as Class E. In both tests involving thermally modified timber (aged and new), the test was terminated prematurely due to high temperatures in the calorimetry ductwork. This is attributed to an increase in heat release rate after ca. 800 s, indicating a greater extent of spread. The same increase in heat release rate was observed for the virgin timber case, but developed much later, i.e., after ca. 1300 s.

When the new thermally modified timber was treated with a fire retardant, the cladding FIGRA was below the threshold for Class B. The resulting total heat release rate was an order of magnitude below that of the non-fire-retardant treated samples.

These results indicate that the fire-retardant treatment, when newly applied, can be an effective method for controlling the fire growth rate on vertical surfaces, as represented by the SBI test, even in challenging configurations comprising slats with gaps.

4.2.2. BRE Intermediate Scale Results Discussion

At an intermediate scale, results for all four timber typologies are shown in Section 3.4. From these results, it is observed that no significant difference exists between samples PS-Virgin, PS-TM-New, and PS-TM-Aged. In both thermally modified cases, there are indications of further spread, apparent through an increase in HRR after ca. 15 min. This increase was not observed for the virgin timber case. The samples with fire retardant (PS-TM-FR) result in substantially reduced HRR, i.e., by circa a factor of two, considering the timeframe of 5 to 15 min from ignition.

To place these results in context, Figure 10 compares the HRR from the samples used in this research with two other façade cladding types presented in [19], where the same testing methodology was used. The description of these four samples is as follows:

• PE-ACM experiment 1 (PE_ACM_1): 4.0 mm thick aluminium composite material (ACM), consisting of 0.5 mm thick aluminium faces and a 3.0 mm thick PE core. Based on bomb calorimeter tests, the gross heat of combustion was determined as c. 46 MJ/kg,
• PE-ACM experiment 2 (PE_ACM_2): as per PE ACM experiment 1,
• Cedar cladding experiment 1 (Cedar_1): Untreated cedar wood timber slats of the tongue and groove type, 19 mm in thickness and 140 mm of width, horizontally oriented and without gaps. The density of the cladding was 355 kg/m³, with a moisture content of between 13 and 15%,
• Cedar cladding experiment 2 (Cedar_2): As per Cedar cladding experiment 1.

Comparing PS-Virgin, PS-TM-New, PS-TM-Aged and PS-TM-FR with PE-ACM and Cedar cases and taking into account that the HRR of Scots Pine is on average slightly lower that of Cedar [20], the results indicate that: (i) the initial growth rates of all experiments are comparable, with this similarity attributed to the consistent burning of the initial crib heat source, (ii) that non-fire-retardant treated timber cladding, in a vertical orientation with gaps in-between, results in a HRR of circa twice that of horizontally oriented Cedar cladding without gaps in-between, (iii) non-fire-retardant treated timber cladding, in a vertical orientation with gaps in-between, results in a HRR of circa half that of PE-ACM, and (iv) new fire-retardant treated timber cladding, in a vertical orientation with gaps in-between, results in the lowest HRR of all samples, i.e., below that of the Cedar reference experiments. This suggests only the crib materially contributed to the HRR.

Considering the benchmarking exercise presented in Figure 10, it demonstrates that the configuration variable (vertical slatted vs. horizontal flat) greatly impacts the fire dynamics of the façade cladding. The vertical slatted, whether thermally modified or not, aged or new, results in substantially higher HRR compared to horizontal flat timber cedar cladding. Additionally, the HRR of the PS-TM-FR sample was comparable to that of the flat cedar cladding, indicating that, in a new state, fire retardant treatments can result in slatted configurations that are of a comparable hazard to that of flat timber (cedar) surfaces.

Finally, the PE-ACM samples did not contribute significantly to the baseline HRR from the crib for the first 5 min of the test, since the aluminium sheets were protecting the highly combustible PE core. However, as soon as the PE core ignited, the HRR of the PE-ACM panel accelerated rapidly to an HRR of ca. twice that of the timber cladding types studied herein (PS-Virgin, PS-TM-New, and PS-TM-Aged). This puts into perspective the quite unique hazard that PE-ACM cladding can introduce to a building when compared to the other cladding systems presented in this paper.

4.3. Impact of Scale

There is some consistency in the results of samples regardless of scale, i.e., high release rates in the cone generally corresponded with high heat release rates in both the SBI and intermediate-scale rig if the configuration of the timber slats is consistent.

Nevertheless, as indicated in the discussion for Figure 10, the configuration of the slats has a great impact on the HRR. This can only be captured with a system-scale test, where the effect of different timber slat configurations is reflected.

5. Conclusions

This research has aimed to better understand the hazards associated with timber cladding through experimentation and testing across several scales. The experimental campaign included samples that were virgin timber, thermally treated, thermally treated and aged and thermally treated coated with fire-retardant. This study was motivated by the fire at Samuel Garside House and is intended to inform future technical policy, i.e., Approved Document B guidance, on the matter of timber cladding. The conclusions respond to the research questions posed in the Introduction.

Firstly, the impact of thermal modification has been studied in small and intermediate-scale tests comparing virgin and thermally modified samples (PS-Virgin and PS-TM-New). From the small-scale experiments, there are no appreciable differences in the burning and ignition behaviour between both samples. The intermediate scale tests indicate a slightly increased peak in heat release rate for the thermally modified samples, which suggests marginally faster fire spread. These results were still substantially lower than those of PE-ACM. Additionally, the thermally treated and aged sample (PS-TM-Aged) had a lower critical heat flux for piloted ignition, indicating that it would typically be easier to ignite. Given aged virgin timber was not subject to investigation, it has not been possible to establish if thermally modified timber is unduly affected by ageing in terms of its burning rate or ignition characteristics.

Secondly, the results and comparison with other research programmes indicate that the orientation of the timber slats has a considerable impact on the heat release rate and fire spread. All the vertically oriented samples (except the samples with fire retardant) had approximately 1.8 times greater HRR compared to the horizontal flat cedar samples presented in [19] in the BRE intermediate scale test. This indicates that reaction-to-fire classes should not be generalised and reflect the behaviour of products used in the as-tested configuration.

Finally, both system-scale tests indicate a considerably reduced HRR for the samples with fire retardant. From the SBI tests, the samples achieved a Euroclass B in the vertical slatted configuration, and in the BRE intermediate-scale tests, the samples presented a lower HRR than the horizontally oriented flat cedar samples from [19]. However, none of the experiments captured how ageing/weathering would affect the performance of the fire-retardant treatment.

6. Potential Areas of Further Investigation

Two areas have been identified for potential further investigation:

• It is highlighted that aged, thermally modified timber supported fire spread more rapidly than new, thermally modified timber. It is not known whether this is an artefact of the thermal modification process or the ageing process. Therefore, consideration should be given to exploring the impact of ageing on the reaction-to-fire characteristics of unmodified timber cladding.
• It has been established that fire retardants are effective in mitigating the burning rate of timber cladding systems, even under challenging configurations such as vertically slatted. However, this conclusion is drawn based upon fire-retardant-treated timber that is both new and has not been subject to an external environment or ageing. If reliance were to be placed on such coatings to improve the reaction-to-fire performance of timber cladding, it would be prudent to undertake research with the aim of establishing its long-term efficacy. Additionally, other types of treatments available on the market, such as organic resin-modified wood, should be studied for a full overview of the different products.