Research and analysis

Fire safety: Smoke and toxicity (excluding appendices)

Published 22 December 2025

Applies to England

1. Introduction

1.1 Background

In June 2017, the Grenfell Tower fire resulted in the death of 72 residents, many others becoming homeless, and wider impacts on the local community. The incident also posed a significant challenge to the operational capabilities of the London Fire Brigade. In response to the Grenfell Tower fire, Dame Judith Hackitt conducted an independent review of Building Regulations and fire safety in England where she supported the recommendation to carry out “…further research with the construction industry to understand who uses Approved Documents, how they are used and where they are used to influence how they should be developed in the future…”.

This report contributes to research that forms part of Dame Judith’s recommended technical review by the Building Safety Regulator (BSR) at the Health and Safety Executive (HSE) (The research was originally commissioned by the Ministry of Housing, Communities and Local Government (MHCLG), which subsequently became the Department for Levelling Up, Housing and Communities (DLUHC), which then transferred its role to develop and maintain Approved Document B to the HSE of the Building Regulations and statutory guidance given by Approved Document B (AD B) for fire safety in buildings in England. The objective of this research has been to develop the technical evidence base to inform policy related to the smoke toxicity hazard from construction projects through the following questions:

1) Is there a need to regulate smoke toxicity arising from construction products and assemblies in fire safety frameworks?

2) What could the regulation of smoke toxicity from construction products look like? How does it integrate into existing frameworks and what test methods are appropriate?

At the start of the project the work programme associated was broken into two broad objectives. Objective A was to carry out a review of the then current research literature, standards and codes, test methods associated with the measurement of smoke and toxicity from fires. Objective B was to provide the technical evidence to develop recommendations underpinned by original research. This report should therefore be read in conjunction with the documents issued as part of this project as summarized in Section 1.3.

1.2 Reporting

This study has been carried out by a consortium of partners that has been led by OFR. The work has been jointly undertaken by members of the University of Edinburgh, University College London, OFR, and Efectis. The contributors and their organisations are identified within the reports contained within each appendix, relevant to the time of writing of the reports. External oversight of the project has been provided by an Expert Review Panel (ERP) consisting of stakeholders from various public and private organisations.

The appendices to this final report are those documents released throughout the research project with some minor editorial and formatting changes. The appendices retain the tenses as they were written at the time rather than being updated to reflect their current status. The earlier documents make reference to work that may or may not have subsequently been carried out as the result of ongoing findings, and feedback from the BSR and the ERP. The research also needed to adapt to various external factors such as newly published documents and events that have occurred over its duration.

Related to the release of the contents of the appendices, two articles have been published in the open literature. As a result of the review process, the articles contain updated and or additional content that has not been included in the original documents (the attached appendices). Further articles may be forthcoming which may again differ from some elements of the work presented here in response to the review process. The two articles are:

• J Reep, J L Torero, M Spearpoint, R M Hadden. Identifying the different processes occurring during the steady state charring of solids, ESFSS 2024: 4th European Symposium on Fire Safety Science, Barcelona, Spain, 2024.
• M Spearpoint, Y Kanellopoulos. Assessing the toxic contribution of fire smoke from the contents of residential spaces, Interflam 2025, Royal Holloway College, London, 2025.

In addition, the two related papers listed below have been published. These papers contain material that is not directly contained within the appendices.

• J Reep, J L Torero, R M Hadden. An approach for the improved measurement of pyrolysis products, Fire Safety Journal 142, 104037, 2024.
• J Reep, J L Torero, R M Hadden. Assessing the effect of oxidizer on flame geometry and effluent composition from burning solids, Fire Safety Journal, 152, 104347, 2025.

1.3 Work programme

1.3.1 Objective A1

Objective A1 (see Appendix A1) conducted a critical literature review of the technical aspects and potential case histories related to smoke toxicity, aiming to identify knowledge gaps and guide the direction, scope, and nature of subsequent objectives. The review was structured around several key topics: the definition and understanding of smoke toxicity, methods for assessing toxicity through experiments 3 and models, a summary of fire statistics and findings from real fire incidents, and approaches to addressing smoke toxicity in building fires across various international jurisdictions as well as in non-building contexts.

The problem of toxicity in buildings is currently addressed by separating people from the products of a fire. The separation is introduced by the expectations of building regulation through the concept of smoke management and depending on the means of achieving this it may not be explicitly directed towards preventing the interaction between people and fire. The approach of preventing the interaction between people and fire products needs to be retained but the means by which this is achieved needs to be studied in a manner that directly relates to the prevention of this interaction.

Generally, the review identified the key effluent species as carbon dioxide (CO2), carbon monoxide (CO), hydrogen cyanide (HCN) which has both narcotic or asphyxiant effects, hydrogen halides (for example, HCl and HBr), sulphur dioxide (SO2), nitric oxides (NOX) and aldehydes (with irritants effects). However, this list is not exhaustive.

The review found that small/bench-scale tests are adequate to characterize the products released from a burning material or product in a systematic way, while large-scale tests provide a ‘scenario-based’ description of the interactions between the combustion products, the fire and its immediate environment. These interactions are of critical importance, and while the existing data is relevant it is not sufficient in that it does not cover all the relevant variables in a methodical manner. The interactions between the fire and its immediate environment therefore need to be studied in a much more detailed manner, in an appropriate scale and with the combined used of experiments and calculations.

It was noted that fire statistics regarding smoke toxicity are misleading in that the data reported is not adequate to establish the nature of the failure and just focuses on the ultimate consequences. Thus, past case studies cannot be used in support of a better understanding of toxicity. If case studies are to be used, better data needs to be collected.

Objective A1 pointed towards the need to consider the identification of common building contents and construction products, including their composition, build-up, and associated material chemistry. This would involve assessing the nature and extent of the contribution of these construction components to smoke toxicity, as well as appraising the broader impact of this toxicity on safety and health. Additionally, the research would examine whether it is viable to develop a robust testing method to evaluate fire effluents from construction products, which would provide an understanding of their potential hazards.

1.3.2 Objective B1

Objective B1 (see Appendix B1) identified common building contents and common construction products, their build up and associated chemical composition. It then carried out an analysis of the potential contribution of construction products to smoke toxicity, with an appraisal of the relative impact of toxicity from construction products, as opposed to furnishings. Although the analysis resulted in some interesting findings, it was concluded that approach has no value to any form of regulatory control on the use of materials for construction and their potential toxic hazard. The method gives an inappropriate emphasis on materials present in large quantities and does not account for materials in lesser quantities that may have a proportionally greater impact. In terms of hazard to occupants, the calculations did not define whether the materials will be involved in a fire or not and did not include their sequence of participation should they be involved. The calculations did not consider the relative location of the occupants to the presence of smoke, at what stage of a fire any exposure occurs or for what duration, and how the fire protection provisions in a building contribute to its fire safety strategy.

Finally, Objective B1 proposed small-, intermediate and large-scale testing options to evaluate fire effluents from construction products. Not all the proposals were eventually employed in the study for various reasons that included the availability of laboratory facilities and the practicality of certain methods.

1.3.3 Objective B2

Objective B2 (see Appendix B2) describes the development of the crib configurations in preparation for the ISO 9705 room experiments undertaken in Objective B3. Different crib stick length and height combinations were assessed with measurements of heat release rate, CO, CO2 and O2 and other toxic gas species made. The work identified a specific crib arrangement and ignition protocol that would be suitable for the room experiments.

1.3.4 Objective B3

Objective B3 (see Appendix B3) explores the use of room-scale experiments to evaluate smoke toxicity from burning construction materials. The experiments were conducted using timber, polyisocyanurate (PIR) foam, and phenolic foam cribs burned in an ISO 9705 room [1] and in separate experiments under fully ventilated, free-burn conditions. Ventilation to the ISO 9705 room was varied by adjusting the opening size to modify combustion conditions. Additional reference tests with polystyrene beads validated the gas sampling methodology.

Findings indicate that while repeatable scenarios were developed for each material, complex interactions between crib combustion, the thermal environment, and fluid flows created varied combustion conditions that were difficult to isolate. Carbon monoxide (CO) and methane were identified in the tested materials, suggesting that assessing their yields under controlled burning conditions may provide a path to benchmark smoke toxicity.

Objective B3 recommended that further work to impose controlled thermal and oxygen environments on materials while monitoring effluent composition and mass loss rates. While full decoupling of burning rate from emissions may remain challenging, this approach could enable rapid material assessments and the development of a screening system based on CO and methane yields.

1.3.5 Objective B4

Objective B4 (see Appendix B4) involved the modification of the Fire Propagation Apparatus (FPA) flammability apparatus to evaluate effluent from burning construction foams. The work developed a testing methodology to decouple pyrolysis rates from the oxidative environment, enabling better assessment of flame geometry and burning processes for materials like polyoxymethylene (POM) and polymethylmethacrylate (PMMA). The method was validated by successfully controlling the mass loss rate (MLR) and investigating the pyrolysis and charring behaviours of insulation foams under varying oxidizer compositions. Observations revealed oxygen-dependent yields, with foams undergoing browning, cracking, and blackening, indicating simultaneous pyrolysis and charring. However, complexities in competing processes made determining the equivalence ratio challenging.

Despite systematic effects observed for simpler materials, challenges arose with lightweight and complex materials due to difficulties in maintaining a stable MLR and limitations of the load cell. Adjustments to optimize sample sizes, MLRs, and a proportional integral derivative (PID) controller were suggested for further studies. During steady-state MLR trials, an increase in energy demand and CO yields after ~350 s marked a transition from pyrolysis to char oxidation. Flaming combustion trials highlighted differences in the behaviour of PIR and phenolic foams, with PIR showing greater fluctuations due to less stable char formation.

The methodology demonstrated the importance of MLR in defining effluent composition, with CO and HCN yields from PIR and phenolic foams aligning with tube furnace data. However, it was noted that generating representative effluent streams requires caution, particularly for charring materials, as decompositional processes evolve during steady-state burning. Objective B4 did not resolve the question of which compounds should be prioritized in emissions analysis, as the species of interest depend on the material. Additionally, there is no current method to evaluate the additive impacts of different harmful species, making it necessary to assess effluent composition on a case-by-case basis. Scaling these measurements for broader application also remained unresolved.

1.3.6 Objective B5

Objective B5 (see Appendix B5) carried out an assessment using the results from the room-scale fire experiments in Objective B3 to estimate the yield of CO when cribs of mixed materials were burned in.

The study sought to estimate the CO yield from a mixed polyisocyanurate (PIR) and phenolic foam crib under fully ventilated conditions by combining results from single-material crib fires with bench-scale CO yield measurements taken from the FPA experiments in Objective B4. The findings suggest that using a theoretical approach to determine mass loss rates is not viable unless a travelling fire methodology is applied, necessitating the use of measured mass loss rates in subsequent analysis. Steady-state CO yield values for PIR and phenolic foam were selected subjectively, with adjustments made after initial values were deemed too high, informed by mixed material measurements.

Four weighted average methods were proposed to calculate CO yields for the mixed material crib. Two methods relied on single-material crib mass loss data, while two other methods required mixed crib mass loss measurements. These methods were specific to the tested conditions and cannot be extrapolated to other scenarios without adjustments. Experiments using different materials, arrangements, or combinations would require new assumptions.

A mixed wood and PVC crib was also tested, with subjective selection of CO yields from the literature. The results indicated that the slower-burning material (phenolic foam) dominated the burning rate in the PIR and phenolic foam crib, a novel finding. However, Objective B5 has significant limitations, including analysis restricted to well-ventilated conditions, specific material arrangements unrepresentative of real-world use, and an assumption of constant CO release relative to fuel mass loss, which conflicts with observed variations during combustion. CO release likely depends on the combustion location, with edge burning yielding different results from interior burning due to oxygen availability.

Ultimately, the findings did not provide a sufficiently robust methodology for regulatory assessment of smoke toxicity in building materials. The experimental approach of coupling bench-scale yields with burning rate measurements lacks the reliability needed for practical application in building construction regulations.

1.3.7 Objective B6

Objective B6 (see Appendix B6) evaluates whether flammability and smoke production classifications from standardised tests can serve as a proxy for regulating toxic gas species production from construction materials. It specifically reviews the EN 13501-1 [2] classification system, focusing on smoke production and its implications for visibility during fires.

The EN 13501-1 system categorises construction materials based on their reaction-to-fire, including smoke production. While restricting materials to class A1 can limit fire contributions and smoke generation (as non-burning materials produce no smoke), using this system to infer toxic hazard has significant limitations. Although some European codes apply the EN 13501-1 smoke classification to enhance safety in specific areas, such as escape routes, directly linking smoke production to toxic hazards remains problematic.

Challenges include the potential for over- or under-estimating toxic hazards, with over-estimation being conservative but under-estimation posing unacceptable risks. Furthermore, the EN 13501-1 test primarily evaluates flammability and provides limited insight into smoke generation or toxicity. Contextual variables, such as dilution by fresh air, further complicate the practical application of smoke classifications as a proxy for toxicity.

Objective B6 found that establishing a robust correlation between smoke production and toxic gas evolution could justify modifying the EN 13501-1 system or developing a new test method. However, even such improvements may struggle to account for scenario-specific factors. Thus, while the EN 13501-1 system indirectly addresses toxicity concerns, it is not a viable foundation for a comprehensive regulatory framework targeting toxic gas hazards from construction materials.

2. Context

2.1 Framework

The starting point for this project has been to recognise that all smoke, independent of its source of origin, is a hazard that has to be managed. Exposure to smoke will have negative outcomes to building occupants during a fire. Smoke may inhibit egress by reducing visibility and/or result in respiratory issues (either as an irritant or asphyxiant) reducing capacity of occupants to move to a place of relative safety. Ultimately smoke exposure may result in incapacitation or death of building occupants.

While this project is focused on construction products, it must be recognised that contents of buildings (which largely cannot be regulated) also have the potential to generate smoke. The composition and quantity of this smoke cannot be reliably predicted for the range of materials and scenarios that may be considered to reasonably occur within a building (fuel types, fire dynamics, compartment ventilation).

Finally, it is often claimed that the toxicity of smoke is not regulated. While it is true that the smoke toxicity hazard is not directly addressed in the current regulation or guidance in England, the risk arising from smoke is managed. This risk management is by controlling the spread of smoke and fire through compartmentation and guidance on how the linings used within a building must inhibit the spread of fire by having low rates of flame spread and reasonable rates of fire growth. As a result, the fire size should be small and hence the total mass of smoke produced should be small.

2.2 Definitions

The definition of smoke is not consistent across all technical areas and between different geographical areas. In the UK, BS EN ISO 13943 [3] is used which gives the following definition for smoke: “visible part of a fire effluent”. Fire effluent is defined as “all gases and aerosols, including suspended particles, created by combustion or pyrolysis and emitted to the environment”. This distinction between smoke and fire effluent is unhelpful in the context of this report.

An alternative definition is given in NFPA 92 where smoke is defined as “The airborne solid and liquid particulates and gases evolved when a material undergoes pyrolysis or combustion together with the quantity of air that is entrained or otherwise mixed into the mass”. This definition provides clarity as to the processes by which 9 the constituent components of smoke may be generated. In this report, the NFPA 92 definition should be assumed unless otherwise specified.

2.3 Hazard and consequence

The hazard of smoke production is inherent in any fire scenario. The consequences of smoke exposure include decreased visibility leading to impeded egress, discomfort due to irritant species leading to impeded egress, psychological challenges during movement through smoke leading to impeded egress, incapacitation due inhalation of toxic or asphyxiant species and or death. These topics have been covered in more detail in Objective A1 (see Appendix A1).

Standard approaches to risk management suggest that where possible a hazard should be eliminated or, if this is not practicable, minimised, and the severity and likelihood of the consequences should be reduced to a reasonable level. Since it is not practical to eliminate the production of smoke (as this would mean only non-combustible materials could be used in construction, and there still would be the presence of any smoke generated from the burning of combustible contents), the hazard should be minimised, and the consequences should be acceptable.

The hazard posed by smoke will be defined by both the volume of smoke produced and the composition of that smoke. Both depend on the material that is burning and the manner in which it is burning. The consequences arising from the smoke hazard depend much more on the interaction of smoke with people. This has a component related to the transport of the smoke through a building to locations where people are present. This transport is likely to be unique to each building design. Any physiological impacts will depend on the composition of the smoke, the time that each individual occupant comes into contact with the smoke and the vulnerability of individual occupants.

Both the hazard and the consequence of smoke must be assessed to evaluate the risk. In this context predicting the likelihood of interaction between smoke and people is extremely challenging as it depends on the probability of a significant fire occurring and understanding details of the reliability of construction typologies (doors, compartment boundaries) and the behaviour of building occupants during a fire event for which there are few or no data.

2.4 Evaluating risk

A flowchart representing the process by which occupants can become exposed to smoke is set out in Figure 1. The items in the shaded boxes are input variables which need to be measured, controlled or understood in making the assessments at each step of the process to make a specific assessment of the risk in a location. Note that the flowchart does not include toxicological effects.

In this study it is assumed that a fire has started, and human behaviour is not considered, therefore, the two central items in the flow chart are of most significance. It should be noted that the inputs in the shaded boxes carry significant uncertainty.

Figure 1 - Flow chart of the steps involved in assessing the toxic hazard of occupants exposed to smoke in a building

It is conceivable that if adequate information was available on each of the input variables, an assessment could be made at each location within a building regarding the evolution of smoke toxicity. This is explored in the following sections.

3. Assessing smoke toxicity risk

3.1 Generation of combustion-related fire products

The toxic hazard posed by smoke produced from a material has two primary components: 1) the generation of species during the burning of materials through pyrolysis and combustion processes, and 2) the transport of these products to areas where there are people. The concepts of pyrolysis and diffusion flames are presented in the following subsections, covering component (1); component (2) is presented separately in Section 3.2.

3.1.1 Pyrolysis

The thermal decomposition of solid materials yields low molecular weight species that exist in the gaseous state at standard temperature. These species mix with oxygen and when the conditions (concentrations, temperature) are appropriate they will ignite and burn forming products of combustion. However, where these conditions are not met, the pyrolysis products may be carried with the smoke plume. In general, the species produced by pyrolysis of a solid material are not well known and the mixture will depend on the elemental composition of the solid, the rate of heating of the solid, and the composition of the oxidative environment. Predicting the products of pyrolysis arising from a material and how this is affected by the burning conditions is a longstanding challenge in the field of fire science.

3.1.2 Diffusion flames

The quantity and composition of smoke produced from the burning of a condensed phase material will depend on the pyrolysis rate, and the factors which control the dynamics of diffusion flames (In this context a diffusion flame is one in which the gaseous fuel and oxygen are not mixed before burning. This is sometimes referred to as non-premixed flame). The pyrolysis rate is largely determined by the thermal environment around the condensed phase material while the dynamics of the flame will depend on the local oxygen concentration, the local temperature field, and the mixing/diffusion of pyrolyzate and oxygen. The effect of these variables on the structure of the flame are in general not well understood. This is a longstanding challenge in the field of fire science.

One of the earliest efforts to understand the diffusion flame was made by Burke and Schumann [4]. Their fundamental work identified that the shape of a diffusion flame was determined by the availability of oxidiser relative to the stoichiometric requirement. Flames in well-ventilated environments (in which there is an excess supply of oxygen) form a complete envelope and no pyrolysis products can escape through the flame sheet. Consequently, the emitted species are generally products of combustion which comprise of relatively high yields of carbon dioxide, water, soot (this is formed in almost all practical diffusion flames), and small quantities of carbon monoxide. As the oxidiser availability is reduced, the flame temperature decreases, and the flame sheet begins to ‘open’ (it becomes non-continuous). This is accompanied by the generation of products of incomplete combustion and allows pyrolysis products to escape through the flame. The composition of the smoke, therefore, includes products of combustion and products of pyrolysis.

The diffusion-controlled nature of fire flames means that it is not possible usually to study these in the context of the equivalence ratio (as this only applies to situations where fuel and oxidiser are mixed before burning). Instead, the problem needs to be explored in mixture fraction space. The combustion community has invested significant effort in understanding this problem (to reduce the emission of incomplete products of combustion from propulsion or energy conversion systems). However, these systems have turbulence intensities which are orders of magnitude higher than the ones occurring in fires, promoting mixing of the fuel and oxidiser streams. Consequently, this understanding cannot be applied to the context of fires.

3.2 Transport of smoke through the building

Once smoke has been generated the species must be transported to the location where occupants may become exposed. As species are transported, they will pass through locations of varying temperature and take different lengths of time to do this. This may result in the evolution of the species present and in the temperature of the smoke hence the mixing within a space. The volume occupied by the smoke will determine the concentration of species and hence the toxicological hazard. Tools exist to explore the transport of smoke through a space (for example, CONTAM, B-RISK, Fire Dynamics Simulator, FireFOAM); however, their use is complex and requires significant technical expertise. While these tools can predict smoke movement, the transport is dependent on properties which remain challenging to predict such as the heat release rate of the fire, and situational variables for which input data are not robust (for example, leakage through doors). These tools can also determine species concentrations using the calculated smoke volumes. However, they do not predict the generation of toxic species but instead typically require users to decide which species are of interest, proscribe yield values and the tools employ empirical correlations to determine rates of toxic gas release under different ventilation conditions.

3.3 Summary

The above sections have (briefly) laid out the fundamental issues related to the toxic hazard of smoke. Each one of these has significant complexity. The toxicological effects are beyond the scope of this project. The characteristics of diffusion flames and the chemistry arising during material burning are clearly a significant factor in determining the toxic hazard and we only have a superficial understanding of these. The transport of smoke is highly scenario dependent, and many factors will not be known a priori and there are few, if any, studies available to provide robust input data.

The dependence on the context required to resolve the transport element of the problem is such that it seems unwise to proceed in setting performance objectives at this level (Even if this were desirable, the toxic yields still need to be assessed). Instead, work should focus on evaluating the generation of smoke from a material as carried on in Objective B3 and Objective B4 (see Appendices B3 and B4). In this scenario it is necessary to be able to explore the dynamics of the diffusion flame (particularly the effects of oxygen concentration) separately from the dynamics of the burning rate of the material to define the hazard under the range of conditions that may be expected in a fire and to decouple the toxic hazard from the flammability hazard.

4. Review of the current guidance

4.1 Guidance associated with the Building Regulations

The five requirements of the Building Regulations in England all must be considered in the development of the fire strategy for a building. While all the requirements must be considered, Requirements B1, B2, and B3 of the Building Regulations address the hazard of smoke most directly.

Requirement B1 deals with the means of warning and escape from a building. Specifically, this requirement refers to “appropriate means of escape”. The intention of this requirement is further clarified to state that the requirement is met if, inter alia, “escape routes are sufficiently protected from the effects of fire and smoke”, and “…there are appropriate provisions to limit the ingress of smoke to the escape routes, or to restrict the spread of fire and remove smoke.” This can be read such that the intention of this requirement is that, other than for those parts of a building intimate with the fire, escape routes should be designed such that occupants evacuating a building should not come in contact with smoke and that if smoke can enter escape routes provisions should be made for its removal (It may be that there is conflict between Requirement B1 and Requirement B5 in that firefighting actions may require the opening of doors in escape routes for prolonged periods to enable firefighting intervention. This may violate the design intent that escape routes remain clear for all “material times”, especially in high rise buildings). This is the best approach to preventing the consequences arising from smoke exposure. The value in this approach is also that it is agnostic towards the composition of the smoke and does not require distinction to be drawn between for example, the negative visibility and the physiological impacts that smoke will have on the escape of occupants.

Requirement B2 manages the hazard of fire spread on internal linings of a building. The intention states that the “…building fabric should make a limited contribution to fire growth, including a low rate of heat release.” It is made clear that the intention of the guidance in Section B2 of AD B “…does not include guidance on the […] generation of smoke and fumes”. Nevertheless, by controlling the ‘rate of heat release’ and the ‘rate of fire growth’ on the products used in the linings of a building, this requirement also indirectly limits the quantity of smoke that can be generated (see Babrauskas and Peacock, [5]). A similar approach of managing the smoke generation by managing flammability is used in Requirement B4.

Requirement B3 states that a building should be compartmented by “fire resisting construction elements” (Note that in this context, and given that AD B does not define the term ‘fire’, the general definition given in BS EN ISO 13943:2023 is assumed: “…process of combustion (3.62) characterized by the emission of heat and fire effluent (3.147) and usually accompanied by smoke, flame or glowing or a combination thereof”). In most applications this means that the element should achieve an integrity (E) rating. The integrity criterion assesses the “ability of a separating element to prevent the passage of flames and hot gases”. Minimum periods of fire resistance are given in Appendix B of AD B.

The individual components of current guidance, while not requiring that the toxicity of construction products be assessed, does provide several methods towards the management of the risk posed by smoke. Although not explicitly stated, one reading of the intentions is that the approach is an onerous one requiring that smoke does not enter escape routes and hence it is ensured that people and smoke are kept separate during the escape from a building. If this is indeed the intention, then any change to how toxicity is regulated should result in a quantifiable improvement to this outcome. Were the toxicity of construction product smoke production become part of the regulations or guidance in AD B this could result in an argument to allow people and smoke to mix, so long as the conditions meet some prescribed limits. This could be seen as a retrograde step when compared to the current position.

4.2 Other approaches

Other approaches have been reviewed in detail in Objective A1 (see Appendix A1). A brief summary is given here.

4.2.1 Health and Safety at Work act – regulating toxic species in the workplace

In the workplace the Health and Safety Executive set exposure limits species which might be considered to be reasonably present. There are currently Workplace Exposure Limits (WEL) for approximately 500 compounds despite there being thousands of potential compounds in use. These are based on durations of exposure of either long term (8 hours) or short term (15 minutes). These limits are necessary as it is recognised that in some scenarios eliminating the hazard(ous material) is impractical and so the hazard must be reduced. These limits can be defined and enforced as it is reasonable to expect employers (the responsible person) to have knowledge of the species that may be present in their workplace. In addition, it is expected that there will be prolonged or repeated exposure to these species so regulating in this way is logical to prevent long term health impacts on employees. One challenge acknowledged in taking such an approach is the relative paucity in data that may be obtained by which to determine the exposure limits. Additionally, the exposure that may be expected during a fire is acute, and different toxicological effects may be present.

4.2.2 Transport for London – regulating toxic species production from burning materials

Transport for London requires screening of materials for the presence of halogens, nitrogen and sulphur using x-ray fluorescence. If these elements are found, then further screening is required using an approach based on the tube furnace or a smoke density chamber. The limits given by the Immediately Dangerous to Life and Health (IDLH) are used. This approach is different to workplace exposure due to two key differences: the ability to closely regulate the contents of the rolling stock and station construction; and a situation in which rapid movement away from a fire may not be possible. In the built environment, this situation primarily arises in the compartment of fire origin. In this case, the most likely items to be burning are contents which cannot be regulated from the perspective of building regulations (although the flammability of certain contents are regulated through the Furniture and Furnishings (Fire) (Safety) Regulations 1988, [6]). Therefore, transporting this approach to the regulation of construction materials is not appropriate.

4.3 Summary

Approved Document B already implicitly provides guidance on the management of the smoke toxicity hazard and risk. Scenarios where escape is difficult (for example, tunnels) or were there are other specific circumstances that increase risk, require more stringent regulations because the separation of people and smoke cannot be guaranteed as there is no compartmentation. Much of the current understanding sets thresholds based on exposure concentrations. In principle this is a rational approach. The use of IDLH or SLOT could be appropriate as there are existing metrics to do this. However, it is unclear how specifying maximum allowable concentrations as in WEL or implied by IDLH or SLOT would improve life safety as this requires detailed knowledge of the fire scenario to allow for their calculation. What is also unclear in the standards is why particular species have been selected to represent the toxicity hazard and what data has been used to set exposure limits.

5. Review of potential methods to assess toxicity hazard

As discussed in Section 3, the species produced from a burning sample are a function of the material and the combustion environment. There are numerous standardised approaches to how smoke toxicity may be assessed, and these have been reviewed in Objective A1. There are also methods which exist already to define the flammability hazard posed by materials and these are well established and provide classifications of materials based on a series of small and intermediate scale tests. There are other methods that can be used or adapted to evaluate the toxicity of combustion products, and these have also been reviewed.

5.1 Existing standardised approaches to measuring smoke toxicity

There are numerous standards which can be used to develop methods of assessing the toxic hazard of smoke and combustion products. These cover methods of sampling (ISO 19701 [7], ISO 19702 [8]), and calculation methods (ISO 19703 [9]). At the core of these methods is the need to generate data on yields of products. For this task the steady state tube furnace (ISO 19700) is used and is operated under conditions defined using equivalence ratio for different stages of the fire (specified in ISO 19706) (Equivalence ratio is not an appropriate way to define a diffusion flame representative of a fire and its utility to compartment fires (particularly through the global equivalence ratio) is not well established). It is recommended in the standard that an equivalence ratio of 0.75 is used for well-ventilated burning and an equivalence ratio of 2 is used to represent under ventilated burning. Data obtained from such trials can be used to perform calculations using standardised approaches (ISO 16732 [10]).

The steady-state tube furnace has been used for decades to calculate the yields of toxic species from materials. One disadvantage of this apparatus is that it does not have the ability to measure the real time mass loss from the specimen. This poses a significant challenge to the interpretation of the data. In addition to this, the range of conditions, while potentially related to stages of the fire development may not represent the ‘worst case’ scenario for every potential product that could be tested. In this respect, as part of the development of BS 7990 [11] and ISO 19700, a number of methods are assessed in BS ISO TR 9122-4:1993 [12] which concludes that “the choice of a fire model must be consistent with a good understanding of the characteristics of the real fire that is to be simulated”. This poses a significant challenge for any potential test method.

5.2 Desired properties of a smoke toxicity test

The strong coupling that exists between the combustion environment, the resulting gas phase oxidation reactions, the thermal feedback to the solid and the impact on the burning rate results in a highly complex process. In a real fire, the conditions under which a material may burn will vary in time and are not well-defined. Therefore, any test method will be limited by the practical constraint of being able to only evaluate a relatively small number of the possible conditions.

There are several categories of testing that could be developed to assess the toxicity of smoke. For the purposes of this report these are grouped by size. Because toxicity always requires understanding of which chemical species are produced and in which quantity, any test method must be able to sample the gaseous effluent, vary the combustion environment and record some characteristics pertaining to the burning rate of the material:

• The sampling process requires a closed environment for the collection of all the smoke.

• Varying the combustion environment requires the test specimen be in an environment that can be modified to change the ventilation, change configuration of the test specimen.

• The characterisation of burning requires some measurement of the solid sample mass during the test. It is also desirable for the burning of the material be independent of external factors.

It is also desirable to evaluate the performance of products rather than materials. In this case the term product is used to refer to an item in an “as sold” condition.

Indeed, the principle of almost all existing reaction-to-fire testing is aimed to achieve the same goal: to eliminate products which will have an unacceptably large contribution to fire growth from some applications. The issue of toxicity is much more nuanced. In designing a test, it is desirable to have a test that is both representative of real conditions but is repeatable and yields results consistently.

5.3 Large scale test methods

A number of large-scale test methods are used in the assessment of fire hazards. These are commonly deployed when the ability to make reliable predictions of fire behaviour are necessary, but the complexity is such that an assessment from first principles cannot be made. Objective B3 undertook experiments at compartment scale to evaluate the possibility of using this scale to evaluate the composition of smoke and the potential to relate this to the burning of the solid.

In any large-scale test scenario, an ignition source is chosen which is deemed to be representative or one relevant to the problem (although this is a somewhat arbitrary choice). Existing large-scale test methods are typically intended to assess some aspect of fire growth and hence the burning characteristics of the materials being tested. The geometric arrangements and dimensions are chosen to reflect typical scenarios (although again this is somewhat arbitrary). Nevertheless, this provides a consistent approach to evaluating the hazard posed by products in a close-to-end-use configuration. It is then up to a designer to interpret the appropriateness of the test method to their scenario, and the result.

Since these tests are concerned with the fire growth in a well-ventilated scenario then a credible worst case in terms of the combustion environment is presented by performing the tests in air. This leaves only the test geometry as a relevant consideration. The close-to-end-use geometries of the different tests, while arbitrary, allow some of the relevant factors to be explored for example, heat feedback between adjacent surfaces and effects of the smoke layer formation.

Applying a similar logic to the design of a test for the smoke toxicity of construction materials the following factors need to be considered:

• arrangement and form of combustible material
• size and duration of the ignition source
• the composition and flow of the gas-phase oxidiser
• the test geometry
• which stage of the fire is of most relevance
• what gas species to measure, and by what method

A representative arrangement of the combustible material/s cannot be easily justified as construction products as a category include products and systems used in vastly different scenarios. They will therefore be subject to different thermal exposures, will become involved in the fires at different stages and hence will be exposed to different conditions. This makes the size and duration of the ignition source difficult to define. Also, it is possibly and likely that many construction products will not burn in isolation (for example, timber, fire retarded polymers) but will burn in a system that produces a favourable thermal environment (reduced heat losses). In a large-scale test, it is not practical to control the concentration or flow of oxidiser therefore this variable must be measured. Highly localised measurements of the oxygen concentration at the location of the flame are not practical to make. Defining an appropriate geometry is also challenging due to construction products, as a category, including products and systems used in vastly different scenarios.

5.4 Intermediate scale test methods

Intermediate-scale test methods that are currently in use, such as the Single Burning Item (SBI) BS EN 13823 [13], all follow the same principles as large-scale test methods which so to say that these are assessments of the flammability of a solid; therefore, the burning is not decoupled from the emissions.

In a report published by ‘Fire Safe Europe’ [14] an assessment was made on whether the SBI was a suitable test method to obtain their proposed toxicity classification system discussed in Section 5.6. However, the report did not support the use of the SBI because it only addresses well-ventilated fire scenarios, the high exhaust flow makes the measurement of low concentrations of toxic gases difficult, and the lack of mass loss measurement of the non-steady state burning conditions means yields cannot be determined. The report notes that “…the Steady State Tube Furnace (ISO/TS 19700) is the most promising test method for such analyses and therefore relevant to be used as complementary bench-scale test method to the Single Burning Item (SBI) test.”

5.5 Bench scale test methods

Bench-scale methods include the Steady State Tube Furnace (SSTF), the cone calorimeter, the Controlled Atmosphere Pyrolysis Apparatus (CAPA), and the Fire Propagation Apparatus (FPA). These differ from large- and intermediate-scale tests in that the thermal environment can be imposed, and in the case of the CAPA and FPA the oxidative environment may also be imposed. However, methods such as the SSTF do not measure the material mass loss in real time and therefore cannot be used to determine yields. Furthermore, it is unclear whether the fire conditions specified in tube furnace/smoke chamber are appropriate to characterise the late stages of an enclosure fire.

Results from a study using a modified FPA were presented as part of Objective B4 (see Appendix B4). This method has significant potential for the evaluation of the species produced from the burning of materials. However, there are a number of challenges when scaling this to a product scale, as mechanical issues cannot readily be captured by this approach.

5.6 Classification

In addition to having a test method (or methods) to measure smoke toxicity, a functional regulatory environment needs to have a system of classification on which materials/products can be assessed. The classification system can be as simple as a pass/fail criterion or can use a system of categorisation (such as low/medium/high; A, B or C). The number of categories, how the categories are delineated, are not trivial questions. Within the Euroclass system there was considerable effort 21 undertaken to determine the flammability classes through the use of FIGRA, whereas, as discussed in Objective B6 (see Appendix B6), there is little information for the basis of the SMOGRA metric used to assign the smoke rating.

The ‘Fire Safe Europe’ [14] report suggested the creation of a three-level toxicity rating (referred to ‘tox’) similar to the smoke rating and dripping rating already defined in EN 13501-1 [2]. The report discusses the use of the Fractional Effective Dose and Fractional Effective Concentration methods to assess the hazard and presents some general suggestions on fire scenarios. However, when it comes to what trigger thresholds should be used to declare the toxicity rating of a material/product the report states “The decision on which values of n1, n2, n3 and n4 [the trigger points for different toxicity classes] should be adopted is the responsibility of the European Commission.” As far as it is understood, no such work has been performed.

5.7 Use of smoke density as a proxy for toxic hazard

It has been suggested in the literature that smoke density may be used as a proxy for the toxic hazard. Smoke density is fundamentally driven by the smoke point of a material. This is a poorly understood area of fire science. The smoke point (soot generation of a material) is again influenced by the diffusion flame structure and the competition between soot generation in the fuel-rich core of the flame and the oxidation of the soot in the flame sheet. While methods exist for the laminar flame smoke point of liquid and gaseous fuels, there is no single, simple method used for the measurement of the smoke points of a solid fuel and this quantity is subject to the same uncertainties around the oxidising environment as the direct measurement of non-soot combustion products. Further discussion on the use of soot production as a proxy for smoke toxicity is given in Objective B6 (see Appendix B6).

5.8 Summary

From the previous arguments, it is non-trivial to design a test that will allow a robust assessment of the yield of products under representative, reasonable conditions of a fire. The challenges lie in a lack of understanding of the dynamics of diffusion flames and in the interpretation of material scale testing and product testing. These have been supported by the experimental work undertaken in Objective B3 and Objective B4 as part of this project. It is likely that these issues may be overcome with advances in fire science and particularly in the understanding of the characteristics of diffusion flames originating from the burning of solid materials. However, even where there are advances in understanding, there remains the challenge of developing a classification system. Given the lack of understanding and a viable classification system, alternative methods should be considered in the management of the risk of smoke production in fires.

6. Alternatives to directly considering smoke toxicity

When considering a fire strategy to ensure “…escape routes are sufficiently protected from the effects of fire and smoke”, and “…there are appropriate provisions to limit the ingress of smoke to the escape routes, or to restrict the spread of fire and remove smoke” it can be posited that there is a hierarchy of already existing measures available such that:

1. Use materials that do not burn (class A1),
2. Use materials that are more difficult to ignite,
3. If items 1 and 2 are not appropriate, then use materials that are less flammable as determined through standard test frameworks,
4. Provide suppression systems to inhibit fire growth (although the activation of suppression system can affect the movement of smoke, its buoyancy, and change the burning regime of materials hence the production of toxic species),
5. Provide adequate separation between the source of smoke and the potential recipients (for example, have smoke seals on doors, use self-closers, install smoke curtains), and/or
6. Use active smoke management measures such as ventilation or pressurisation systems.

This is a comprehensive and robust set of measures. Where there might be a concern of the hazard from smoke being unacceptable then any or all these measures could be considered before resorting to assessing the toxic species that are generated from different products. In addition, many of the above measures will also address smoke from the contents of a building, not just the construction materials, and, so, provide significant benefits in managing the overall hazard to building occupants.

Were construction products to be regulated for their generation of toxic species, it may be taken as grounds to seek reductions in the performance expectations of one or more of the measures within the statutory guidance in AD B. For example, arguments may be made that materials with lesser reaction-to-fire performance could be permitted if it was deemed that the contribution to smoke toxicity was lower. This would result in an overall decrease in the robustness of the guidance.

7. Conclusions

7.1 Discussion

In the preceding sections the issues around the regulation of smoke toxicity have been summarised. The challenges of toxicity arise from the need to make a comparative assessment between different products with differing material compositions, burning behaviours, which may be burning in different environments, that may generate a vast array of possible toxic species. This means that assessing the hazard posed by the burning of a material is currently beyond the state of the art in fire science. Furthermore, the contextual variables required to make a robust assessment of the consequences of smoke production on egress of building occupants introduces further complexity and uncertainty.

In this project, the fundamental understanding of the issues of smoke toxicity have been evaluated and found that there is a fundamental lack of understanding of the way that toxic species are generated in a fire (Objective B1). These challenges arise due to a lack of understanding of the processes which occur in a diffusion flame which allow the escape of pyrolysis gases through the flame and the generation of products of incomplete combustion. It is recommended that further research be supported in these areas following on from Objective B4. This would allow the effect of oxidiser environment on the dynamics of the diffusion flame and, hence, the generation of species to be fundamentally understood (rather than correlated) thereby allowing a more rigorous testing framework to be developed.

Additional issues arise due to the complexity introduced by products (rather than materials). Many construction products are comprised of multiple materials with different thicknesses, compositions and properties. The burning behaviour of a product is therefore a function of the interactions between these materials (as illustrated in Objective B5). While this may be dealt with by testing in an ‘as used’ condition, such tests generally must be done at larger scale (as demonstrated by Objective B3) which has been shown to be incompatible with systematic changes in the burning environment. Testing the constituent parts of a product may be possible but requires the definition of which parts of a composite product must be tested. The use of the material composition as a screening for their use is deemed impractical as many products are designed to obtain other, non-fire-related, benefits.

As a consequence, while it may be appropriate to consider regulation of the toxicity of smoke from construction products, the fundamental understanding and the large range of potential variability that may be introduced as a result poses a significant 24 technical challenge to generate robust, repeatable testing framework that allows clear and straightforward interpretation.

7.2 Grenfell Tower Inquiry

In September 2024 the Grenfell Tower inquiry Phase 2 report was published. Specifically related to this research, paragraph 85.40 notes that:

“Professor Purser identified three groups of material that contributed to the smoke within the tower: the external cladding and insulation and the window infill panels, the window surrounds and the contents of the flats. During the early stages of the fire it is likely that the cladding, the window infill panels and the window surrounds were the main contributors to the smoke. The smoke from the fire on the outside of the building entered the tower and the contents of the flats became involved in the fire at different times as it progressed. The contribution made by the different materials to the smoke varied. We are satisfied that the cladding, insulation, window surrounds and window infill panels made a significant contribution but we are unable to say how much of a contribution each of them made at different times and in different places.”

This statement aligns with the findings of this research presented in Objective B1 (see Appendix B1) in that it is difficult to separate the contribution of smoke from the contents of a building from any construction elements. Objective B1 showed that contributions will depend on factors such as the relative mass of materials, the rate of smoke production, the relative time to ignition of combustibles. The Phase 2 report statement that “the cladding, made a significant contribution” does not appear to be definitively supported by the available information to the inquiry and should be viewed with caution. Furthermore, this significance is directed towards a contribution to the smoke and is not necessarily a contribution to the fatalities, albeit the two are clearly linked.

In the context of the toxic effluent species, paragraph 85.51 of the Phase 2 report states:

“Professor Purser was unable to say whether any of those who died had inhaled enough hydrogen cyanide to make a significant contribution to incapacitation and death. He was of the opinion, which we accept, that it had made some contribution but that the dominant toxic gas causing incapacitation and death had been carbon monoxide.”

Although this research has not only examined the yields of HCN and CO from the materials investigated, the statement from Professor Purser indicates that focussing the generation of CO in Objective B5 for example (see Appendix B5) has been a rational approach to assess the challenges of using small-scale test results to estimate the CO generated in large-scale experiments.

More broadly, the Phase 2 report makes numerous references to the reaction-to-fire performance of products in terms of flammability, rate of fire spread. Based on experiments undertaken after the fire, Paragraph 109.47 notes that:

“…the principal reason why the flames spread so rapidly up Grenfell Tower was the presence of ACM panels with polyethylene cores which had a high calorific value, melted and acted as a source of fuel for the growing fire. It is clear from the experiments that the principal factor which led to rapid growth of fire was the presence of unmodified polyethylene in the cores of the ACM panels…”

Furthermore, Paragraph 109.48 notes that:

“…although the contribution made by the insulation (with foil facers) to the total heat released during the period before the full involvement of the ACM is comparatively minor, representing less than 15% of the total energy released…”

These conclusions align with findings from this research that notes that considering the reaction-to-fire performance of materials is the primary way to reduce the fire hazard. In addition, the conclusions suggest that the contribution of the insulation to the early stages of the fire was minor, and it was the ACM external cladding panels with their polyethylene (PE) cores that contributed to the fire spread (and therefore the generation of smoke). According to Reisen [15] the yield of CO from PE is similar to wood in both well-ventilated and vitiated conditions.

Finally, paragraph 113.13 states that:

“We do not think it appropriate for us to recommend specific changes to Approved Document B, save in one respect. As we have pointed out in Chapter 48, the guidance proceeds on the assumption that effective compartmentation renders a stay put strategy an appropriate response to a fire in a flat in a high-rise residential building… One thing that has emerged clearly from our investigations is that in order to ensure the safety of occupants, including any with physical or mental impairments, those who design high-rise buildings need to be aware of the relationship between the rate at which fire is likely to spread through the external walls and the time required to evacuate the building or the relevant parts of it.”

The Phase 2 report clearly identifies that separating people from smoke is a key expectation of a fire strategy. This statement along with those regarding the reaction-to-fire performance of materials in terms of flammability (the rate of fire spread) aligns with the hierarchy of fire safety provisions available to a fire strategy discussed in Section 6. Importantly, none of the inquiry recommendations state there needs to be regulation on the smoke toxicity of construction products.

7.3 Implementation of regulations

As noted in Section 1.1, there are two components to this work which can be summarised as follows:

1. Should the generation of toxic species from the burning of construction materials be regulated?
2. If so, how would that be done?

These questions are inherently linked, and it is possible to determine a matrix of regions of differing outcomes as defined in Table 1.

Table 1 - Matrix of different outcomes regarding the ability to regulate effectively

		Regulate?
		Yes	No
Adequate assessment	Yes	A	B
	No	C	D

Of the four combinations, only Region A is a rational outcome in that there would be a need to regulate and there was an adequate assessment regime to do so. If there is no ‘need’ to regulate then it does not matter whether there is an adequate assessment method (Regions B and D) and should there be a ‘need’ to regulate but there is not and adequate assessment method then any regulations could not be meaningfully supported.

As discussed in Section 5, this work concludes there are many other fire safety measures currently available to mitigate the hazard of smoke toxicity. Therefore, enacting a regulation would likely provide little extra benefit to life safety compared to increasing one or more of those measures were it deemed necessary, therefore the answer to Question 1 above is ‘No’. Furthermore, given this work has shown that creating any form of adequate assessment in the context of construction materials is beyond the capability of current measuring techniques, design tools and sound engineering practice. The answer related to Question 2 above on whether there is an adequate assessment method is also ‘No’. Thus, promulgating any regulation would be a sub-optimal policy decision.

8. Recommendations

The primary recommendation from this work is that it is not necessary to explicitly regulate the toxicity of the smoke produced from burning construction materials. This recommendation is partly based on the current understanding of the production of toxic species in fires which prevents robust test methods to be generated but also because there are many other ways to manage the smoke toxicity risk with existing elements of fire engineering design guidance. This should be supported by a statement of intent from the Secretary of State in AD B clarifying the approach to smoke toxicity through the inclusion of reaction-to-fire requirements and the provision of fire protection measures such as compartmentation.

The following recommendations are also made:

• The toxicity arising from construction materials be kept under review as new products enter the market.
• The development of underpinning technical knowledge in this area (test methods, modelling tools, theoretical developments) be kept under review and supported by BSR.
• Consider how some form of cost-benefit analysis of regulating the smoke toxicity of construction materials could be carried out, noting that such an analysis would need to consider what toxicity thresholds might be unacceptable, what test methods might be used. This study has already indicated the challenges involved in the consideration of these factors.

9. References

[1] ISO, ‘ISO 9705-1:2016 Reaction to fire tests. Room corner test for wall and ceiling lining products. Test method for a small room configuration’, ISO, Geneva, 2016.

[2] BSI, ‘BS EN 13501-1:2018 Fire classification of construction products and building elements. Classification using data from reaction to fire tests’, BSI, London, 2018.

[3] BSI, ‘BS EN ISO 13943:2017 Fire safety. Vocabulary’, BSI, London, 2017.

[4] S. P. Burke and T. E. W. Schumann, ‘Diffusion flames’, Industrial & Engineering Chemistry, vol. 20, no. 10, pp. 998–1004, 1928.

[5] V. Babrauskas and R. D. Peacock, ‘Heat release rate: The single most important variable in fire hazard’, Fire Safety Journal, vol. 18, no. 3, pp. 255–272, Jan. 1992, doi: 10.1016/0379-7112(92)90019-9.

[6] UK Government, The Furniture and Furnishings (Fire) (Safety) Regulations 1988, 1988.

[7] BSI, ‘BS ISO 19701:2013 Methods for sampling and analysis of fire effluents’, BSI, London, 2013.

[8] ISO, ‘ISO 19702:2015 Guidance for sampling and analysis of toxic gases and vapours in fire effluents using Fourier Transform Infrared (FTIR) spectroscopy’, ISO, Geneva, 2015.

[9] BSI, ‘BS ISO 19703:2018 Generation and analysis of toxic gases in fire. Calculation of species yields, equivalence ratios and combustion efficiency in experimental fires’, BSI, London, 2018.

[10] BSI, ‘BS ISO 16732-1:2012 Fire safety engineering. Fire risk assessment. General’, BSI, London, 2012.

[11] BSI, ‘BS 7990:2003 Tube furnace method for the determination of toxic product yields in fire effluents’, BSI, London, 2003.

[12] BSI, ‘BS ISO TR 9122-4:1993 Toxicity testing of fire effluents. The fire model (furnaces and combustion apparatus used in small-scale testing)’, BSI, London, 1996.

[13] BSI, ‘BS EN 13823:2020+A1:2022 Reaction to fire tests for building products. Building products excluding floorings exposed to the thermal attack by a single burning item’, BSI, London, 2023.

[14] Fire Safe Europe, ‘Classification system for the smoke toxicity of fire exposed construction products’, c 2019.

[15] F. Reisen, ‘Inventory of major materials present in and around houses and their combustion emission products’, The Centre for Australian Weather and Climate Research, Aspendale, Victoria, Australia, Jan. 2011