Fire Resistance of Seats in Railway Vehicles

Featured Application

The presented test results can be used by designers and manufacturers of rolling stock seats to properly select upholstery fabrics

Abstract

This article discusses the current requirements for laboratory testing of the fire properties of seating in railway vehicles and the criteria for their assessment. The results of flammability and smoke tests performed on selected passenger seats and samples of the upholstery systems used in their construction are presented in order to find connections between them. It was demonstrated that the composition of the upholstery fabric has a significant impact on the burning behavior of the seats and the upholstery systems themselves, assuming that the same foam was used in their construction. Based on the conducted research, material composition analysis, and results, a lack of correlation was also found between the results of tests using a cone calorimeter and a furniture calorimeter. This confirms that the fire properties of upholstered products depend on many factors, including composition, shape, materials used, type of upholstery, and the design solutions of the finished seats. The tested upholstered products intended for railway applications are characterized by stochastic variability resulting from their specific applications and functional and operational properties.

1. Introduction

The increasing incidence of fires in passenger rolling stock in the 1970s prompted a search for tools to assess fire risk and identify directions for fire prevention. However, due to the varying development of rolling stock in individual countries, varying principles of its use and maintenance, and varying causes of fires, various testing methods for the flammability and smoke properties of materials and various classification systems were developed [1,2,3]. These remained in use until the EN 45545 series of standards was established in 2013 [4,5].

The main goal of fire safety measures for passenger rail vehicles is the protection of passengers and crew in the event of a fire on board by reducing the risk of fire, delaying its development, controlling the spread of combustion products, and ensuring appropriate evacuation conditions [4]. These tasks are achieved, among other things, through passive protection measures, which include the use of materials that meet specific fire performance criteria.

The selection of appropriate component materials for the construction of rolling stock equipment, including passenger seats, is one of the elements that directly influences ensuring the required level of fire safety.

A significant proportion of non-metallic materials susceptible to ignition are upholstered seating areas in railway carriages. These are directly exposed to ignition sources that may occur in vehicles, including heating system failures, electrical short circuits, fires, and arson. Despite advances in the development of various fire protection technologies, these products are very difficult to render flame-retardant. This is primarily due to the high mass fraction of foam and the difficulty of reconciling its functional properties, which are crucial for passenger comfort, with its flammability and smoke properties. Nevertheless, passenger safety is a priority in public transport, and therefore equipment components must meet stringent requirements, which are being revised as more advanced research is conducted and knowledge is gathered about the phenomena occurring during the spread of fire in a moving vehicle [6]. The flammability of complete passenger seats, and especially the upholstery fabric, is a crucial parameter for flame propagation in a railway carriage, which significantly determines the development of a fire within the vehicle. Another effective measure to reduce the fire properties of a seat is the use of flame-retardant nonwoven fabric glued to the seat foam [7].

2. Requirements for Flammability and Smoke Properties

Currently, the requirements for upholstery system components and complete seats are specified in the EN 45545-2:2020+A1:2023-12 standard [8]. Sets of requirements R for individual types of seats and their components are placed in group F according to [8] as shown below in

Table 1. Requirements of listed products [8].

Product No	Name	Details	Requirement
F	Furniture
F1	Complete upholstered seats in passenger areas	Complete passenger seat, including seat shell, upholstery, arm and head rests. Tip-up seats and staff seats (if accessible to passengers) are also included. Details of seat tests (including the conditions for vandalism testing) are given in EN 16989 [9].	R18
F1A	Upholstery and head rests for seats in passenger areas	Upholstery includes the trimming (e.g., suspension system), flexible foam core, intermediate layers (e.g., fire barrier, anti-vandal layer), seat covers (e.g., base, back, side cover), and head rest upholstery.	F1A
F1B	Armrests for seats in passenger areas	The surface on which the arm rests shall be tested. In addition, the downward-facing and vertical surfaces in the normal operating position shall comply with the requirements of 5.2.2.2 Fire integrity.	F1B
F1C	Seat shell–base in passenger areas	The external surface of the base shell (including all coatings or coverings) shall be tested.	F1C
F1E	Removable head rests for seats in passenger areas	Removable head rests shall be tested as if they were loose cushions.	R21
F1F	Seats in passenger areas without upholstery	All surfaces, outside the scope of F1C and F1D, which support the seat user.	R6
F2	Seats in staff areas	Staff seat upholstery and supporting structure (including the back/base shell) shall be tested according to the following conditions:	F2

. They include parameters such as smoke properties and toxicity of emitted gases and the amount and rate of heat released as a result of combustion.

Table 2. Material requirement set [8].

Requirement Set (Used for)	Test Method Reference	Parameter and Unit	Maximum or Minimum	HL1	HL2	HL3
R6 (F1C; F1D; F1E)	T03.01 ISO 5660-1: 50 kWm−2	MARHE kWm−2	Maximum	90	90	60
	T10.01 EN ISO 5659-2: 50 kWm−2	Ds (4) dimensionless	Maximum	600	300	150
	T10.02 EN ISO 5659-2: 50 kWm−2	VOF4 Min	Maximum	1200	600	300
	T11.01 EN 17084 Method 1 50 kWm−2	CITG dimensionless	Maximum	1.2	0.9	0.75
R18c (F1; 1E)	T06.01 EN 16989	MARHE kW	Maximum	80	55	25
	T06.02 EN 16989	TSP600, m2	Maximum	–	– b	45
	T06.03 EN 16989	Flame height above seat base, mm	Maximum	–	1180	1180
R19 (F2)	T03.02 ISO 5660-1: 25 kWm−2	MARHE, kWm−2	Maximum	75	50	50
R20 (F4)	T07 EN ISO 12952-2	After burning time s	Maximum	10	10	10
	T03.02 ISO 5660-1: 25 kWm−2	MARHE kWm−2	Maximum	50	50	50
	T10.03 EN ISO 5659-2: 25 kWm−2	Dsmax. dimensionless	Maximum	200	200	200
	T11.02 EN 17084 Method 1 25 kWm−2	CITG dimensionless	Maximum	0.75	0.75	0.75
R21 (F1A; F1B; F1E; F1F; F3)	T03.02 ISO 5660-1: 25 kWm−2	MARHE kWm−2	Maximum	75	50	50
	T10.03 EN ISO 5659-2: 25 kWm−2	Dsmax. dimensionless	Maximum	300	300	200
	T11.02 EN 17084 Method 1 25 kWm−2	CITG dimensionless	Maximum	1.2	0.9	0.75

b—TSP shall be measured to allow comparative data to be accumulated in anticipation of a possible future threshold value for HL2.

presents sets of requirements for individual R. As it follows, the values of the criterion parameters are differentiated depending on the hazard category (HL) for the rolling stock resulting from its design category and the assigned railway infrastructure in which the rolling stock operates, which is illustrated in

Table 3. Hazard level classification [8].

	Design Category
	N: Standard vehicles	A: Vehicles forming part of an automatic train having no emergency trained staff on board	D: Double decked vehicle	S: Sleeping and couchette vehicles
1	HL1	HL1	HL1	HL2
2	HL2	HL2	HL2	HL2
3	HL2	HL2	HL2	HL3
4	HL3	HL3	HL3	HL3

according to the standard [8].

3. Test Methods

Testing complete passenger seats is time-consuming and expensive (both in terms of testing and seat cost). However, the tests are intended to fully confirm compliance with safety requirements. The goal of the CEN Working Group (in which the Institute participates) was to create the ability to select seat component materials based on the results of pre-selected tests of upholstery samples according to R21 requirements.

As part of this project, it was decided to examine the correlation between the results of selected laboratory tests conducted for passenger seats according to the R18 requirements and for their upholstery systems according to R21 requirements.

The goal was to verify the feasibility of effectively classifying passenger seats based solely on tests on upholstery samples and to potentially submit proposals to the Working Group for changes to the standard. Additionally, it was decided to examine the impact of upholstery fabric on test results.

These tests were conducted in accordance with the requirements of EN 45545-2 [8] and the aforementioned standards [9,10,11,12]. The test methodology for R18 according to [8], using the furniture calorimeter, has changed compared with the previous version contained in Annex C of EN 45545-2 [5]. Currently, tests for complete passenger seats are conducted according to the EN 16989:2:2018 standard [9], which, compared with the previous version contained in Annex C of the EN 45545-2:2013 standard [5], introduced the following changes:

• installation of three walls under the exhaust hood;
• installation the wall on the side of the seat;
• modification of the burner to increase its power from 7 kW to 15 kW;
• measurement of the flame height above the seat surface;
• recording of smoke emissions (Total Smoke Production);
• method of preparing a non-vandal-resistant seat.

The testing methodology for upholstery systems for the scope according to R21 has not changed.

Figure 1, Figure 2 and Figure 3 below show the test stands where the above-mentioned fire tests were performed. Figure 1 shows the furniture calorimeter used in the tests, Figure 2 shows the cone calorimeter according to [10], and Figure 3 shows the smoke chamber with the Fourier Transform Infrared Spectrometer (FTIR) according to [11,12].

In order to achieve the assumed goals of this project from the tested passenger seats proposed for use in long-distance transport, those that were similar in shape and weight and had the same polyurethane foam composition were selected for analysis.

Table 4. List of tested passenger seats.

Seat Number	Seat Shell	Foam	Fire Blocker	Upholstery Fabric
1	PC/ABS	polyurethane	none	100% polyester
2	PC/ABS	polyurethane	mAR and pAR	70% polyester + 30% cashmere wool
3	PC/ABS	polyurethane	mAR and pAR	85% wool + 15% polyamide
4	PC/ABS	polyurethane	mAR and pAR	100% polyester
5	PC/ABS	polyurethane	mAR and pAR	85% wool + 15% polyamide
6	PC/ABS	polyurethane	mAR and pAR	85% wool + 15% polyamide
7	PC/ABS	polyurethane	none	100% polyester
8	PC/ABS	polyurethane	none	100% polyester
9	PC/ABS	polyurethane	mAR and pAR	80% wool + 20% polyamide
10	PC/ABS	polyurethane	none	100% polyester

lists their components.

As can be seen from the table above, the tests were carried out on complete passenger seats in which the shell was made of PC/ABS plastic, while the upholstery systems consisted of polyurethane foam and upholstery fabric. Some of them also contained a fire-resistant layer made of meta- and para-aramid (mAR and pAR) fibers. The test seats used polyester/polyurethane materials and a mixture of synthetic fibers (polyester or polyamide) with wool in various proportions.

All tested objects were conditioned in a climatic chamber to constant mass, in accordance with the requirements of the relevant standards: temperature 23 °C ± 2 °C, humidity 50% ± 5%. Environmental conditions were also observed in the laboratory rooms.

Figure 4 and Figure 5 present sample heat-release rate graphs recorded during laboratory tests using a cone calorimeter and a furniture calorimeter. A summary of all the results obtained for each seat/upholstery system is provided in

Table 5. Results of tests carried out on a furniture and cone calorimeters and in a smoke chamber.

Parameter, UNIT	Test Method (Object)
	Seat 1 30/24	Seat 2 50/24	Seat 3 107/24	Seat 4 75/23	Seat 5 122/24	Seat 6 123/24	Seat 7 156/24	Seat 8 149/24	Seat 9 52/25	Seat 10 41/25
	EN 16989:2018 [9] (complete passenger seat)
RHR Peak, kW	46.6	60.5	24.4	29.5	22.6	22.1	36.5	48.6	20.8	22.8
MARHE, kW	24.6	26.5	19.4	18.6	17.4	17.3	25.7	29.5	15.5	17.0
TSP, m2	91.3	72.3	71.7	92.3	19.2	38.3	97.2	129.8	35.2	13.7
HL classification for R18	HL2	HL2	HL2	HL2	HL3	HL3	HL2	HL2	HL3	HL3
	System 1 35/24	System 2 17/24	System 3 103/24	System 4 76/23	System 5 115/24	System 6 105/24	System 7 102/24	System 8 152/24	System 9 7/25	System 10 44/25
	EN ISO 5660-1:25 kWm−2 [10] (upholstery system)
MARHE, kW/m2	41.4	44.3	23.8	43.2	7.3	27.6	47.4	44.9	37.7	24.4
HRRmax, kW/m2	118.2	107.0	71.6	104.9	30.6	110.2	162.5	148.9	220.5	129.6
Sa	763	687	451	854	542.3	42.8	580	619	416.8	221.4
EN ISO 5659-2:25 kW/m2 [11] (upholstery system)
Ds. max	275.3	208.9	115.5	242.8	168.7	76.8	154.9	157.2	233.6	101.2
HL classification for R21	HL2	HL2	HL3	HL2	HL3	HL3	HL3	HL3	HL2	HL3

RHR Peak—maximum heat emission rate, kW; MARHE—maximum average rate of heat emission, kW/m², [10]; TSP—total smoke production, m²; MARHE—maximum average rate of heat emission according to EN 16989 [9], kW; HRRmax—maximum heat release rate, kW/m²; S_a—total smoke production, m²/m²; D_S max—maximum smoke optical density.

.

As part of the smoke chamber tests according to ISO 5659-1 [11], combustion product toxicity measurements were also performed using a Fourier Transform Infrared Spectrometer (FTIR) EN 17084 [12], but they were not the subject of the analysis of this project.

Based on tests conducted using a cone calorimeter and a smoke chamber (for upholstery samples) and a furniture calorimeter (for complete seats without cut upholstery), the results of the selected parameters was analyzed.

In the first stage, an attempt was made to determine the correlation between the intensity of smoke emission during the testing of the upholstery systems using the cone calorimeter method (Sa) and the single-chamber test method (D_S max), presented on Figure 6.

In the next step, a correlation was sought between the test results for the upholstery systems, i.e., those conducted on small samples, and the results of full-scale tests, i.e., those conducted on complete seats. The resulting correlations are presented in the graphs (Figure 7, Figure 8 and Figure 9).

As can be seen from Table 5 and the graphs (Figure 6, Figure 7, Figure 8 and Figure 9), despite adhering to strict test procedures, a significant scatter of results occurred, allowing only a rough approximation to determine the trends in the existing relationships. In contrast to the results of the research carried out in 2017 [13], no polynomial character (character of a 4-degree polynomial curve) was obtained in the relationship between the MARHE values of the upholstery systems and the MARHE of the complete seats, for which the values obtained ranged from 7.67 kW to 39.0 kW. However, it should be taken into account that the earlier tests were carried out according to the methodology compliant with Annex C of the EN 45545-2:2013 standard [5] which differs, among other factors, in the use of a burner with more than 50% lower power, which has a significant impact on the combustion process and the heat emission values.

At the same time, the MARHE values from the cone calorimeter according to [10] were in the range of (44.2–62.55) kW/m², indicating that the upholstery of the previously tested seats was made of materials with higher fire susceptibility than those used in the current tests.

The current tests also confirmed the lack of correlation between the HL classification results obtained for the seats and the HL classification for their upholstery systems. Similarly, a lack of correlation between the results of tests using the cone calorimeter and the furniture calorimeter was demonstrated in studies conducted at the TGM Laboratory in Austria [7] and at the National Institute of Standards and Technology in the USA [14].

Due to the lack of a clear correlation between the results of the compared parameters, the results of tests performed as part of the interlaboratory comparison studies organized by CERTIFER France in 2023 [15], in which the Institute also participated, were subjected to a detailed analysis. As can be seen from the summary in

Table 6. Test results of passenger seats in Inter-Laboratory Comparison Tests [15].

Laboratory No	MARHE, kW	TSP, m2	HL Classification
N	21.12	21.80	HL3
P	56.67	369.33	HL1
R	47.10	167.83	HL2
T	64.50	230.27	HL1
U	43.77	224.30	HL2
V	72.73	71.57	HL1
X	44.52	346.73	HL2
Z	78.73	73.43	HL1
ZD	37.13	44.67	HL2
ZG	57.67	37.50	HL1
ZK	26.80	153.10	HL2
ZS	11.63	7.73	HL3

and the graphs in Figure 10 and Figure 11, although the tests discussed were performed on seats from a single production batch and in accredited laboratories with consistent measurements and competent personnel, the obtained values show significant discrepancies. The MARHE values are in the range of (11.63–78.73) kW, and the TSP values are in the range of (7.73–346.73) m².

The results were evaluated by CERTIFER using a statistical method in accordance with ISO 5725-2:2019 [16] using the mean and standard deviations for different laboratories, repeatability, and reproducibility, the 2σ criterion (two standard deviations) and the Mandel test. As a result, only the results from the ZG laboratory were not accepted due to excessive intra-laboratory variance. However, as shown in Table 6, the tested seats are classified according to EN 45545-2 [8] into three risk levels, from HL1 to HL3. Based on the analyzed results and the obtained variances, it can be concluded that conducting tests in accordance with this test method requires high skill and accuracy, and the results used to assess fire safety in rail vehicles according to the TSI LOC&PAS [17] requirements should be one of many criteria when estimating the risk associated with fire hazard.

The authors of this article also analyzed studies on the ignition susceptibility of furniture armchairs that were conducted in a laboratory at the University of Edinburgh [18]. In this study, 25 upholstered armchairs from a single batch were subjected to calorimetric tests, which determined the heat release rate (HRR), mass loss rate (MLR), and toxic gas emissions (CO, CO₂, N₂O, NO, CH₄, HCN). Significant discrepancies were observed in the obtained results. However, additional analysis by the authors of the discussed publication showed that by taking into account the visual observation of combustion of individual objects, divided into their different processes (i.e., pyrolysis, flame, and smoldering), greater consistency of the obtained values was achieved.

This confirms the influence of many factors on the combustion process of multi-component objects, such as upholstered products, which exhibit stochastic variability as their complexity and scales increase. Therefore, the use of calorimetric test results of such objects for FDS simulations or other engineering applications requires conducting laboratory tests for a sufficiently large population, while taking into account all the observed phenomena occurring in them.

The test results for the tested railway seats, presented in Table 5 and Figure 6, Figure 7, Figure 8 and Figure 9, allow us to conclude that the composition of the upholstery fabric had the greatest impact on the combustion process of the tested objects and the obtained values (despite the high mass fraction of foam), which is also confirmed by other studies [19]. It was observed that the more synthetic fibers (polyester, polyamide) in the material composition, the higher the product’s flammability. It should also be noted that the tested wool-upholstered seats contained a fire-retardant layer that limited the combustion of the foam filling.

This is due to the chemical decomposition mechanism that occurs during combustion. Wool typically contains a relatively high nitrogen content (approx. 14%) and water, resulting in a very high ignition temperature in the range of (570–600) °C and is characterized by relatively low heat emission. The natural fire-resistant properties of wool fibers are related to their chemical composition and the formation of a charred layer on the material’s surface, which acts as a barrier to heat and oxygen, thus delaying combustion. Furthermore, wool products emit less smoke and toxic gases during combustion than synthetic polymers [20]. This was observed during the tests, as illustrated in the graphs presented (Figure 6 and Figure 7). On the other hand, fabrics made of synthetic fibers, or those with a dominant share of synthetic fibers, are characterized by lower ignition temperatures, approximately (432–488) °C. Furthermore, synthetic fabrics such as polyester and polyamide ignite quickly with intense flames and often emit dense, black smoke [21].

Therefore, to achieve the best fire resistance, 100% wool fabric should be used in the production of passenger seats. Even a 15% addition of polyamide lowers the ignition temperature by 100 °C and increases smoke emission.

Furthermore, it’s important to consider that the density and type of fabric (plush, bouclé, jacquard, chenille, alcantara, microfiber, etc.), the thickness of the fabric, and other seat components, particularly the type of fire blocker, foam, and adhesive used, also play a significant role in the burning process.

4. Conclusions

The tests conducted allowed us to formulate the following conclusions.

The fire properties of passenger seats are closely related to the flammability of the materials used in their construction. Proper material selection and testing for fire properties are crucial to ensuring passenger safety in the event of a fire. However, tests according to the R21 requirements should be considered a screening method allowing for the preliminary selection of component materials for finished products.

This is due to the fact that the highly precise calorimetric method developed by Babrauskas [22], applied to homogeneous, repeatable materials, produces a scatter of results for other materials, especially for multilayer samples (permissible by the ISO 5660-1 standard [10] up to 10%). However, seat classification cannot be based solely on R21 results. The analyses conducted confirmed the lack of correlation between the HL classification results obtained for the seats and the HL classification for their upholstery systems. Similarly, the lack of correlation between the results of the tests using the cone calorimeter and the furniture calorimeter was demonstrated in studies conducted in other laboratories [7,15].

The combustion processes of multi-component objects, such as upholstered products, are highly variable on a natural scale. Within a single production run, these objects may differ in workmanship, which affects flame propagation. Furthermore, conducting tests according to this test method requires considerable skill, accuracy, and adherence to environmental conditions, particularly airflow.

However, despite the lack of reproducibility of the results, the applied method of classifying these products based on the current requirements of the EN 45545-2 standard [8] according to R21 and R18 is the most adequate method for assessing the fire risk of upholstery products in rail transport. Nevertheless, the results used to assess fire safety in rail vehicles according to the TSI LOC&PAS requirements [18] should be one of many criteria when estimating the risk associated with fire hazard.

The composition of the upholstery fabric has a significant impact on the combustion behavior of seats and upholstery systems when using the same foam. Based on research, it has been determined that the more synthetic fibers a fabric contains, the higher the heat and smoke emissions. This is particularly noticeable when using a fire blocker, which prevents ignition sources from reaching the foam. Therefore, to achieve the best fire resistance, 100% wool fabric should be used in the production of passenger seats.