An overview and experimental analysis of furniture fire safety regulations in Europe

The regulation requirements for upholstered furniture are nationally defined and lead to very different safety levels across Europe. This paper studies the fire performance of upholstered furniture across Europe and demonstrates the differences in the fire safety of sofas in different countries. Real sofas were purchased over Europe and tested first alone then for several of them in a room scenario. The fire performance of such upholstered furniture is related to their constitution and possibly to the improvement of their performance by physical and/or chemical means. Results are presented from a performance point of view and not focusing on the different technical solutions that can be envisaged, such as fire barriers or flame retardants. The results clearly show the effect of stricter regulations, not necessarily on the maximum heat release but mainly on the time available to escape, which is critical in many fire scenarios. Such regulations may have a positive effect, but they have to be performance‐based, evaluated properly, and implemented considering market surveillance.


Summary
The regulation requirements for upholstered furniture are nationally defined and lead to very different safety levels across Europe. This paper studies the fire performance of upholstered furniture across Europe and demonstrates the differences in the fire safety of sofas in different countries. Real sofas were purchased over Europe and tested first alone then for several of them in a room scenario. The fire performance of such upholstered furniture is related to their constitution and possibly to the improvement of their performance by physical and/or chemical means. Results are presented from a performance point of view and not focusing on the different technical solutions that can be envisaged, such as fire barriers or flame retardants. The results clearly show the effect of stricter regulations, not necessarily on the maximum heat release but mainly on the time available to escape, which is critical in many fire scenarios. Such regulations may have a positive effect, but they have to be performance-based, evaluated properly, and implemented considering market surveillance. have also been published. [6][7][8] The concept of "flame retardant" is often misleading as it refers to various effects (in gas or solid phase), among a large number of chemical families. This leads to confusion when flame retardant is used as a generic name to deal with toxicity. Some flame retardants, mainly a small number of brominated chemical species, have been banned in the past 1 and are known to be persistent pollutants. 9 There is also a concern that newer flame retardant chemicals are being applied before their consequences are properly evaluated and that they are no better than those originally banned. 7 The United Kingdom, which has the most stringent furniture fire safety regulations in the European Union (EU), is in fact actively trying to change its regulations to "encourage the reduction in usage of flame retardants" (FEU, p. 22). 2 Herein, an overview is presented of (recent) literature related to the furniture fire safety and halogenated flame retardants debate and the results of sofa reaction to fire tests conducted at Efectis in 2010.
[The copyright line for this article was changed on 4 April 2020 after original online publication.]

| Statistical evidences
There are several reports available concerning fire statistics related to upholstered furniture in different degrees of detail, from the United States, [10][11][12] the United Kingdom, 13 the EU, 15 and Sweden. 14  sofas, FR-treated sofas were shown to release more toxic combustion products than nontreated sofas. Based on fire statistics from 1999, 50 times more fires started in nontreated sofas ignited than in FR treated sofas. These statistics were used to offset the more hazardous emissions of FR treated sofas, which lead SP to conclude that nontreated sofas in total had a larger negative environmental impact than FR-treated sofas.
The emission results, however, seem to only account for sofas as item first ignited and not to account for sofas that are burned during living room fires. According to the statistics stated in the report, over 3300 living room fires spread beyond the first item, compared to a total of 510 sofas as first ignited item. Using the statistics and assumptions from the report, it can be shown that in fact more FR-treated sofas than nontreated sofas have been burned in residential fires. It is unclear whether this analysis was included in the overall life cycle analyses of the different sofa types.
The results of this report should therefore be handled with caution.
The EU report presents statistical data of the number of domestic fire fatalities per million inhabitants from 19 EU countries. 15  fatalities per million inhabitants is still higher in these countries than in for example Germany or the Netherlands. Overall, no correlations could be established between the presented results and fire safety regulations for furniture and textiles in each of the included EU countries. The results primarily suggest that also national culture and other legislation, such as the taxation on alcohol, influence domestic fire safety.

| Testing
All of the implemented testing standards (ie, TB 117, 16  implemented. There were also two projects of European standards in the nineties for stronger ignition sources than a cigarette and a match, but these projects have been cancelled (projects prEN1021-3 and -4).
Cigarette and match tests are the principal test methods (see Table 3).

| Regulations across Europe
Sainrat presented in 2006 38 the status on fire regulations related to upholstered furniture. Table 4 is a summary of his work with the updates from Chivas et al 1 and further ones, such as. 15 Since these studies, the introduction of low ignition propensity cigarettes in Europe makes the "cigarette" test scenario less relevant but the change is too recent to observe any noticeable effect on fire statistics.
Recently in France (2014), ANSES (the French national safety and sanitary agency) performed a review at the request of ministries to balance the consequences of a possible reinforcement of regulation on upholstered furniture flammability that could lead to an increase in use of flame retardants, and the related impact. 39 As conclusion of the study, ANSES recommended the status-quo on the French regulation.
The European Commission started to investigate the possibility to develop a regulation on smoke toxicity of construction products. 40 The conclusion of the report was that a prescriptive approach could not be adapted, and that further analyses are needed. Consecutively, Fire Safe Europe 2 launched recently an initiative to develop a methodology to assess and classify the smoke toxicity of fire-exposed construction products. These studies are not impacting upholstered furniture but discuss about measuring toxicity and relations between contributions of the contents (ie, furniture) vs the construction products to fire growth and smoke generation and effects. Relations between the mass loss and the toxic effects (incapacitation and lethality) due to the combustion of upholstered furniture in a  43 This study was adapted to large volumes such as the ones available in theaters and other spectacle rooms, and the toxic risk was considered as covered during the ignition test (gas burner equivalent to 20 g paper cushion) if there was no lateral spread of flame from one seat to another and if the mass loss of an individual seat was less than 300 g. In these conditions, it was considered that burning 300 g in such a volume would not lead to untenable conditions for occupants, whatever the burning seat was.

| Fire dynamics
There are several papers that discuss the fire dynamics of upholstered furniture fires. 44

| Discussion
The peak heat release rate (HRR), maximum HRR, time to peak smoke, and total smoke production. The authors also highlighted that this level of performance was not achieved at the expense of the smoke toxicity.

| Analyses from literature review
The

| EXPERIMENTAL ANALYSIS OF EUROPEAN FURNITURES 7
The goal of this study was to determine the fire performance of upholstered sofas from each country of the EU that were available on the consumer market at the time. Two experimental series were conducted as a part of this study, the first series tested a single sofa from each country of the EU and the second series tested a selection of sofas in a furnished room setup.
In the first series, the sofas were ignited using a wood crib #5 (BS 5852), and the HRR and smoke production rate were measured during the experiments. Test configuration was free burning of sofas under a 3 m × 3 m calorimetric hood, mechanically ventilated. When possible, the sofas were allowed to burn until natural extinction. For ten sofas, the temperature in the ventilation ducts became critical, typically for fires that reached an HRR over 1200 kW, and the fire was extinguished prematurely.
In the second series, six sofas were ignited using the wood crib #5 in identically furnished rooms. The test configuration was similar to ISO 9705-1 52 /EN 14390 53 standards, with the described test room. The heat release rate, smoke production rate, temperature in the room and time to flashover were measured using the standardized calorimetric hood placed at the door.
The heat release rate, expressed in kW, is calculated using the oxygen consumption method, as proposed by Thornton 54 and modified by Huggett. 55 In practice, resolution on the heat release rate is driven by dilution and the size of the equipment. So, to allow accurate measurement, the used hood size and flow rate in the duct have to be adapted to the expected heat release rate to obtain a depletion factor not exceeding 17% of oxygen in the effluents, but being significantly different from the initial oxygen concentration. The measurement uncertainty of the heat release rate is evaluated as ±10%.The smoke production rate is calculated according to the light transmission method, using a white light opacimeter. The beam attenuation is expressed as the extinction coefficient in m −1 . The smoke production rate, expressed in m 2 /s, is given by multiplying the extinction coefficient with the flow rate and its measurement uncertainty is evaluated as ±20%. Temperatures during the second series of tests have been measured using K-type 1.5 mm Inconel shielded thermocouples, which have an evaluated measurement uncertainty of ±2.5 K.
The crib #5 ignition source used in this study best describes the ignition scenario of sofas being ignited by open flames from another item on fire. Statistics from the United States indicate that this scenario is the most frequent ignition scenario (25%) for upholstered furniture fires. 6 These statistics show that the tested scenario is relevant in terms of upholstered furniture fires.
The data set was analyzed and presented relative to the best performing sofa and the results of the room experiments were evaluated with respect to evacuation. Pictures of the different Sofas and rooms tested are available as supplementary material.

| Phase 1: Sofa experiments
There was no selection criteria for the tested sofas, apart from their commercial availability. The tested sofas therefore showed a wide variety of physical properties and used materials. In a certain degree, they represent customer preferences in each country. As such, the fire performance of the sofas could not be related to any specific sofa property or physical parameter, including construction materials. Each sofa should, however, adhere to the fire safety regulations of its respective country where it was sold. The data set thus provides the opportunity to analyze whether the fire performance of sofas improves with increasingly stringent fire safety regulations. Only for Ireland, the country with the second most stringent fire safety regulations, two sofas were tested because the first sofa performed very poorly. Based on this poor fire performance, the first sofa was initially estimated to not comply with the Irish fire safety regulations. A second sofa that should have complied with the Irish fire safety regulations was therefore purchased and tested as well. It should be noted that while the UK sofa had the best fire performance, it did in fact not pass the crib 5 test as described by BS 5852 (no flaming at 10 minutes after ignition of the crib). This may be a typical example of a mock-up that had passed the test while the results are not representative of the actual sofa, which has been observed during other experimental series as well. [34][35][36] The main results are available in Tables 5-7 Burns were extinguished prematurely "to prevent damage to the measuring equipment in the system and to the ventilator fins" for which a maximum equipment temperature of 350 C was set. N/A means not applicable.  Figures 2 and 3. Open symbols represent the experiments that were extinguished manually to prevent the measurement systems from overheating (see also Table 5). Since the data set is very limited, any observed trends from the results should be considered as preliminary results until more data becomes available.
The different types of fire safety regulation levels related to upholstered furniture that were active in the EU in 2010 can be sum-

| Heat release rate analysis
By plotting the time to the peak HRR as a function of the regulation level ( Figure 4), the data clearly show that only the second sofa from Ireland No correlation was found between the peak heat release rate and the total heat release, as shown in Figure 6. All tests showed peak heat release rates over 500 kW, independent of the total heat released. Figure 7 also does not show any correlation between the initial mass of the sofa and the total heat released, but does indicate that for this type of dataset, with furniture items lighter than 100 kg, the heat released is limited to 600 MJ. Overall, all sofas ignited and were either extinguished manually due to too rapid fire growth or burned out fully, extinguishing because of a lack of fuel.

| Smoke release analysis
The relevant plots related to the smoke production are shown in . Due to a measurement error there is no smoke data available for the sofa from the United Kingdom. There does not seem to be any sort of correlation between the different regulatory levels and the smoke production (rate). There was also no strong correlation found between the HRR and smoke production (rate), indicating that different types of materials were used in the fillings of the sofas.
Because the material composition of the tested sofas was unknown, it is not possible to provide a more detailed analysis of the data. As for the total heat released, Figure 10 shows no correlation between the initial mass of the sofa and the smoke released.

| Phase 2: Room experiments
In this phase, six sofas are tested in a small furnished room to demonstrate the consequences of their performance in a semi-realistic consumer application. The room chosen is based on the room corner test The results are summarized in Table 8

| CONCLUSIONS
The literature study and the experiments on the European sofas have shown that the current furniture fire safety regulatory frameworks have reduced domestic furniture fires only to a limited degree. The ignition scenarios that are commonly covered by the regulations remain the primary ignition causes of furniture fires. This is also observed during the tests, where instead of preventing ignition, compliance with the most stringent regulations only delayed ignition for high ignition sources such crib #5.
The prescribed tests and regulations should therefore be updated to better reflect the actual modes of accidental ignition for upholstered furniture. Furniture fire safety regulations should ideally be based on the fire performance of the full piece of furniture, and not just mockups, to reach the expected level of fire safety. Since tests on mock-ups seem to provide only limited data that is relevant for end-use scenarios, there is a need for further research on scaling between mock-ups and real-scale tests. Furthermore, assessments are needed for variations inside a family of upholstered furniture using the mock-up scale test samples. Laurens van Gelderen https://orcid.org/0000-0002-1690-8334 ENDNOTES 1 "Penta-BDE and Octa-BDE were voluntarily phased out of production in 2004. Also in 2004, the sale of penta and octa in concentrations higher than 0.1% by mass were banned in the European Union. In 2006, EPA required that any new manufacture or use of penta or octa would constitute a new use and require prior evaluation before approval." Cited from NFPA. 10