You are on page 1of 11

ARTICLE IN PRESS

Fire Safety Journal 45 (2010) 238–248

ARTICLE IN PRESS Fire Safety Journal 45 (2010) 238–248 Contents lists available at ScienceDirect Fire Safety

Contents lists available at ScienceDirect

Fire Safety Journal

journal homepage: www.elsevier.com/locate/firesaf Low-cost wool-based fire blocking inter-liners for

Low-cost wool-based fire blocking inter-liners for upholstered furniture

P. Martini a , M.J. Spearpoint b, n , P.E. Ingham c

a Aurecon, Christchurch, New Zealand

b Department of Civil and Natural Resources Engineering, University of Canterbury, New Zealand

c Textile Science and Technology Section, AgResearch, Lincoln, New Zealand

article info

Article history:

Received 19 May 2009 Received in revised form 3 March 2010 Accepted 24 March 2010 Available online 14 April 2010

Keywords:

Wool

Upholstered furniture

Inter-liner

Calorimetry

abstract

New Zealand is reluctant to implement mandatory fire safety regulations for domestic furniture because of the cost/benefit even though up to 30% of household deaths and injuries over the past 20 or more years can be attributed to soft furnishings. Work has been carried out to develop a low-cost inter- liner using fire retardant treated wool with altered proportions of different fire resistant synthetic fibres that may be attractive for voluntary inclusion by manufacturers. Bench- and full-scale tests have been conducted to characterise the fire performance of composites of polypropylene upholstery fabric, different inter-liner materials and a polyurethane foam core. Inter-liner blends that reduced the peak rate of heat release, extended the time to peak rate of heat release and reduced other combustion product output were identified. A blend of 75% Zirpro treated fire retardant wool and 25% Panox fibre was found to be the best candidate material when fire performance, cost and practical application were considered.

& 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Studies [1,2] of New Zealand residential building fires from the 1970s until 2000 have determined that a relatively constant percentage of between 20% and 30% of deaths and injuries can be related to fires involving bedding and upholstered furniture. Common, untreated or unmodified upholstered furniture exhibits fire growth characteristics leading to considerable risk to life and property. Work by Ingham et al . [3–5] in the late 1970s and early 1980s provided evidence of this risk, while more recent work [2,6–8] has continued to highlight the fact that New Zealand furniture has not made any progress in terms of fire safety. Concerns have been raised many times but moves to introduce mandatory measures have been met with reluctance from those with commercial interests and those organisations which would inherit the responsibility of implementing and enforcing the regulations [9] . Jurisdictions such as the UK and the state of California have introduced mandatory regulations and minimum standards, such as BS 5852 [10] and California Technical Bulletin 117 [11] . Organisations such as the Upholstered Furniture Action Council (UFAC), USA and the Consumer Protection Safety Commission (CPSC), USA have implemented voluntary standards such as Technical Bulletin, TB 116 [12] , although not all are in agreement as to which standards should be adopted [13] . However, the difficulties of regulating the fire safety of New Zealand

n Corresponding author. Tel.: +64 3 364 2237; fax: +64 3 364 2758. E-mail address: michael.spearpoint@canterbury.ac.nz (M.J. Spearpoint).

0379-7112/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.

upholstered furniture can be attributed to the issue of ‘cost’. For governing bodies, this is the cost/benefit outcome dominated by the increased costs due to regulation, administration and enforcement against the expected numbers of lives and property saved while for manufacturers it is the cost of production, compliance and market competitiveness. The best solution to reducing combustibility of soft furnishings has been a point of debate since the early 1980s [9] because of the expected costs of regulating every component material and the inevitable increased costs of production which will be passed onto the consumer. The alternative of regulating only the most combustible or greatest contributor to the fire risk also raises debate because it singles out and disadvantages only selected products and manufacturers. The choice of upholstery fabric and underlying component materials has pronounced effects on the burning characteristics of soft furniture composites [7,14,15] . Protection of the foam component does nothing to reduce the ignition susceptibility and burning of the surface material. In some cases, it may increase the heat release rate (HRR) contribution from the upholstery once ignition has been established. Despite this, the greatest fire risk is generally considered to be from the mass bulk of foam filling contained within the furniture. A furniture fire may develop from a smouldering ignition source, such as a cigarette, and either remain smouldering or transition into a rapid flaming fire. Damant [16] suggests that life threatening conditions are not often encountered during the smouldering phase of full-scale cigarette ignition tests, and therefore this work focuses on the flaming fire scenario. One option available to reduce the risk is the use of a fire resistant inter-liner between the upholstery fabric and the foam

ARTICLE IN PRESS

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

239

core. The use of an inter-liner is by no means a new idea as many materials and composites have been used for this purpose; however, the availability and usage of these in domestic furnish- ings appears to be under-utilised. Published data and studies on inter-liner development are sparse and contribute a very small percentage of the total literature on domestic fire safety research. Three likely reasons for this could be: the lack of support for the development of such materials; a shift in focus towards the improvement of the polyurethane foam compounds; and the patenting of fire-blocking fabrics for commercial benefit. The introduction of intermediate layers of material between the surface upholstery and the resilient polyurethane foam core has been evident since at least the 1970s when the aircraft industry began to introduce higher standards of testing and performance for aircraft seating and fixtures. It was recognised that the greatest fire load contribution within the passenger cabin was exclusively from the polyurethane flexible foams used in the seating assembly. Regulatory requirements were enhanced in 1986 that subsequently required the upholstery/cushion combi- nations to pass the FAR 25.853(c) kerosene burner test [17] . The easiest solution for compliance was to encapsulate the foam within a durable fire blocking fabric, subsequently leading to enhanced popularity and inclusion of inter-liners in aircraft seat design. Some early fire barriers to emerge contained a high percentage of wool blended with other fibres which, when treated, demonstrated fire resistant capabilities above the wool’s natural fire retardant (FR) qualities. Research and development into synthetic fibres produced FR fibres with improved fire resistance, and eventually these began to replace the wool or natural fibre portion of FR materials but at a much higher cost of production. The objective of the research described in this paper was to develop a low-cost inter-liner for use in New Zealand domestic furniture by blending FR treated wool and a high temperature resistant fibre in a non-woven structure. By combining low-cost FR treated wool with altered proportions of different fire resistant synthetic fibres, the aim was to identify the best performing fibre blend and proportion producing equivalent performance to some existing, more expensive commercial fire-blocking fabrics. If successful, the low-cost barrier may make it more attractive for voluntary inclusion into furniture products by New Zealand manufacturers in the absence of mandatory requirements. It was not the aim of this work to create a blocking fabric that would meet a particular test standard, and so the results given here are likely to be applicable to jurisdictions outside of New Zealand.

have to endure a stringent and frequent cleaning process. Consequently the wool fibre content for the test samples consisted of mid micron ( 25 m m) scoured wool treated with an 6% w/w treatment of K 2 TiF 6 (potassium hexafluorotitanate) per kilogramme of wool fibre. Zirpro is not the only successful FR treatment for wool available, but it has withstood the test of time because of its simplicity of application, its reliability in fire tests, durability and compatibility with other post-treatment processing such as shrink resist, insect-resist and fabric finishes. The reason that the Zirpro treatment was chosen for this work is the result of many years of experience and familiarity of the process at the Wool Research Organisation of New Zealand (now the Textile Science and Technology Section of AgResearch), the relatively simple process of application and the superior intumescent properties of this treatment. This research does not exclusively promote the use of Zirpro above other FR alternatives. Any other treatment which can demonstrate equivalent performance would be acceptable.

2.2. Fire resistant fibres

For a flame retardant to be effective, it must interrupt the combustion mechanism. The modes of interruption can be broadly described by those given in Table 1 . Although the majority of textile related fibres associated with heat resistance and flame retardancy have fire retardant chemicals applied after the manufacture of the textile, about 20% by mass of textile fibres produced can be grouped into the category of synthetic flame or heat resistant specialty fibres [19] . The inherent high temperature characteristics of specialty fibres can mostly be attributed, but not exclusively limited to, several mechanisms described by modes 2 and 3. These qualities, pertaining to the more successful FR fibres, are significant in a handful of aromatic polymers: aromatic poly- amides (also known as aramids), polyamide–imides, polybenzox- azoles, polybenzimidazoles, carbonised polyacrylonitrile and melamine based fibres. One indication of a material’s ability to resist combustion is the measure of its limiting oxygen index (LOI). It is the concentration of oxygen in the test environment needed to sustain combustion of a sample of the material for a definite time or until a specified amount has been consumed. Fibres which do not burn at an LOI of 25 or above are considered to be flame retardant [20] . Many synthetic fibres are now

2.

Material selection and development

2.1.

Fire retardant wool

Wool has a high degree of inherent non-flammability through

a combination of flame retardant attributes. Wool has a high

ignition temperature due to high moisture regain but suffers from brittleness once charred and this brittleness compromises the protection offered by the char layer to anything behind it. The standard flame retardant treatment for wool has been the ‘Zirpro’ process [18] developed by the International Wool Secretariat (IWS) in the 1970s. The treatment uses a zirconium

or titanium complex which consistently produces an intumescent,

robust char boundary layer against the penetration of flame and hot gases. The application of the FR chemical results in absorption into the fibre so that the treatment obtains resistance against mechanical wear without the loss of FR qualities. It is effective for up to 50 commercial washes, although as a treatment for the intermediate barrier in domestic furniture it is unlikely it will

Table 1 Different mechanisms for fire retardant action observed in fibres (adapted from Horrocks [19] ).

Modes of interruption

Example

1 Removal of heat

High heat of fusion and/or degradation and/or dehydration

2 Increased decomposition

Not usually exploited by flame retardants and more usual in inherently flame and heat resistant fibres e.g. aramids.

temperature

3 Increase in char formation and decrease in flammable volatiles

Most phosphorus- and nitrogen- containing flame retardants for cellulose and wool; heavy metal complexes in wool e.g. Zirpro treatment.

4 Reduced access to oxygen or flame diffusion

Hydrated and some char-prompting retardants release water, halogen containing retardants release water.

5 Interference with flame chemistry and/or increase fuel ignition temperature

Halogen containing flame retardants, often in combination with antimony oxides.

ARTICLE IN PRESS

240

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

available which obtain LOI values between 30 and 40 with some exceptional fibres having an LOI around 60 [19] .

A range of synthetically produced ‘technical’ fibres designed

for specific engineering applications was sourced with a demon- strated ability to withstand flame and high temperatures. Most of the fibres acquired have been or are currently used in protective clothing applications or applications involving high temperatures and/or flaming. An approximate relative cost of each raw fibre has been included in Table 2 . These estimates are solely for the cost of the fibre and do not include processing costs. The pricing of the Zirpro wool component is an approximation based upon the wool price and the treatment cost. Table 2 also gives indicative LOI data which where available from the published literature or manufacturers’ information.

2.3. Foam

The selection of polyurethane foams and suppliers is broad such that no single foam can be considered as being a representative material used in furniture construction. Despite new formulations and variations being continuously introduced, the combustion properties of these foams, excluding the highly specialised formulations like those used in aircraft, fall within a narrow band of values, especially in terms of HRR. It was not the purpose of the research to achieve a predetermined improvement in combustion qualities or to reach predefined targets, but to demonstrate that given the same circumstances a simple fire barrier can offer an increase in fire safety comparable to other accepted products. For these reasons, the foam material needed only have reasonably ‘normal’ combustion characteristics with respect to other furniture foams.

A sufficient quantity of CTU43-180, 43 kg/m 3 , polyurethane

foam sourced from a local supplier was available. The foam was claimed to be of fire retardant quality and to be self extinguishing in terms of BS 4735 [25] , a horizontal burning test for small cellular plastic/rubber specimens subject to a 60 s flame applica- tion time. An HRR comparison between CTU43-180 against other commonly available polyurethane foams is shown in Fig. 1 . The tests were conducted in the cone calorimeter at an irradiance level of 50 kW/m 2 and curves are an average of three samples of each type. The majority of fire retardant foams are designed to delay the onset of ignition. The CTU foam did demonstrate this feature to a certain degree, but in comparison to the other common foams it was not overly convincing. The CTU foam had an average ignition time of 40 s, while the domestic untreated 25 kg/m 3 foam had an average ignition time of 38 s. The important characteristics to consider after ignition are the HRR and total energy available and it would appear that the CTU foam did not exhibit any significant advantage. Although peak HRR occurs about 50 s later

in the CTU foam than in the other foam types, the maximum HRR ( 300 kW/m 2 ) was not significantly less than the non-treated foams which peaked in the range of 300–390 kW/m 2 . CTU43-180 also maintained a high heat release for a longer period than the three domestic foams and it is questionable whether any fire retardant quality existed. The combustion response of ‘true’ fire retardant foam can be seen in the output curve of the aircraft foam, where a substantial reduction in HRR is obtained over the entire test period as a result of the low intensity flame generated. The demonstrated combustibility of the CTU43-180 foam was sufficient to provide a basis for the performance of the inter-liner selections. Because of the available supply of this foam and its negligible fire retardant aspect, the foam was used in all further tests. The average effective heat of combustion of three samples of the foam was 24.1 MJ/kg.

2.4. Upholstery fabric

The worst performing common upholstery in fire involving domestic furniture is expected to be that using 100% polypropy- lene fabric [19] . This has an LOI of 17.4 [26] and high heat of combustion compared to most other popular covering materials. Hirschler and Shakir [14] found that the covering fabric has the greatest influence on the fire performance of the system, so by including the worst expected performing covering material the samples represented the higher end of the flammability risk spectrum for soft furnishings. A commercially manufactured 100% polypropylene plain weave fabric, with a fabric weight of 260 g/m 2 was used throughout.

with a fabric weight of 260 g/m 2 was used throughout. Fig. 1. Bench-scale heat release

Fig. 1. Bench-scale heat release rate comparison of different polyurethane foam types.

Table 2 Identification, LOI data and approximate relative cost per kilogram of raw material of fibres examined in the research.

Type

LOI

Relative cost

Wool

25 [19] 27–28 (Zirconium); 31–32 (titanium) [21] 28 [20] 30–32 [22] 45–60 [22] , 55 [20] 32 [23] 30–32 [20,22,23] 38 [22] 4 40 [24]

1.0

Zirpro wool

1.5

FR Viscose

Viscose Melamine formaldehyde Oxidised poly(acrylonitrile) Aromatic polyimide–amide Aromatic polyimide–amide Polyimide Epoxy, chopped carbon

3.7

Basofil

3.7

Panox

5.5

Kermel Tech

7.0

Kermel

9.7

P84

10.5

Grafil

13.3

ARTICLE IN PRESS

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

241

Martini et al. / Fire Safety Journal 45 (2010) 238–248 241 Fig. 2. Cross-section of the

Fig. 2. Cross-section of the thermal protection index (TPI) apparatus (adapted from BS EN 367 [27] ).

2.5. Production of non-woven material

Samples of blended non-woven material both for bench- and full-scale experiments were produced through the forced entanglement of fibres, where the mechanical process of needle punching was used to form a non-woven fabric. The process is particularly useful for applications where the use of spun yarn, such as in woven or knitted fabric, would introduce an unnecessary expense. Zirpro treated wool and the selected technical fibre were blended and mixed together by hand in the required proportions according to mass fraction. The loose blend was then introduced to a sample card, where by the action of several needled rollers spinning at differential velocities, the fibres were further blended and combined to form continuous web. Needle punching of the fibres was achieved using a punch machine fitted with regular barb (RB) needles to produce even felting on both sides of the processed material. Apart from the obvious aesthetic qualities, double-sided needle-punching also gives a denser, tighter and more entangled fibre arrangement that improves both the residual strength and barrier performance qualities. Three wool/technical fibre ratios were selected as 95/5, 85/15 and 75/25. A nominal fabric weight of 300 g/m 2 was chosen as the target fabric weight of the finished blended material as most currently available fire-blocking fabrics fall within the fabric weight range of 250 and 350 g/m 2 .

3.

Comparative heat transfer performance of pure fibre

3.1. Procedure

In order to isolate and distinguish the best performing technical fibre types, non-woven, 3 mm thick samples were made

consisting of 100% pure fibre. Assessment was conducted using a modified setup of the BS EN 367 test [27] consisting of a total net heat flux of 50 kW/m 2 produced by a Meker burner flame incident on one side of the sample and a copper disc calorimeter on the opposite side ( Fig. 2 ) with a thermocouple used to measure the heat transfer to the disc. Heat transfer performance of a material is a good indicator for sample superiority, but is not a sufficient criterion by itself. Heat transfer through the sample incorporated a combination of heat passage through the network of fibres and, where/if the sample structure had been ruptured, split or decomposed, directly from flame to the calorimeter. The test did not distinguish between the modes of heat transfer and the physical changes of the sample during testing. Consequently the heat transfer tests gave a relative comparison of performance. Reducing the likelihood of mechanical failure of the inter-liner is an important factor when choosing a suitable fibre to combine with wool. By choosing a fibre which also has a strong dimensional stability during flaming exposure, the stress on the char structure is minimised, leading to a lesser likelihood of rupture of the barrier substrate. The network of fire resistant fibres would also act as a scaffolding and support matrix, upon which a continuous intumescent thermal boundary can form. Extra 300 g/m 2 samples were tested with the modified convective heat equipment and the projected, intact, surface area was calculated before and after testing to gauge the dimensional stability during high temperature thermal exposure.

3.2. Results

A parameter referred to as the thermal protection index (TPI) is defined here as the best linear fit line to the test data of the time taken for the copper disc calorimeter to measure a temperature of

ARTICLE IN PRESS

242

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

300 1 C minus the TPI of the calorimeter without a sample for a range of sample densities. The choice of 300 1 C follows from the approximate ignition temperature of polyurethane foam which has been shown to fall in the vicinity of 280 1 C [8] . The TPI results ( Fig. 3 ) show clear linear trends between each sample type and especially between the best and worst performing samples. The

between the best and worst performing samples. The Fig. 3. Comparison of the thermal protection index

Fig. 3. Comparison of the thermal protection index (TPI) results for each pure fibre type at different fabric weights.

Table 3 Ranking of pure fibre needle felts according to dimensional stability during exposure to flame and comparison with thermal protection index (TPI) results.

Ranking

Fibre

Area change %

Fibre

Calculated TPI

1 Grafil

0

Grafil

210

2 Panox

8

Panox

150

3 Basofil

16

Basofil

81

4 FR Viscose

39

FR Viscose

90

5 Zirpro wool

49

Zirpro wool

56

6 Kermel

61

Kermel

79

7 P84

84

Kermel Tech

38

8 Kermel Tech

88

P84

28

9 Wool

n/a n

Wool

18 a

n Could not be measured due to extensive decomposition. a TPI values for lambs wool was experimentally determined at only one density and only for comparison against the FR fibres listed in the table.

trends become more defined and separate as the range of sample density was increased. The relative ranking of pure fibre samples based upon TPI and dimensional stability are shown in Table 3 , where the calculated values for dimensional stability are an average of 10 tests. Table 3 demonstrates a strong level of agreement between the heat transfer and dimensional stability performance. The highest TPI values are given by those samples showing the least dimensional change during exposure to flame and generally those with a higher LOI. The trend is consistent and supports the intuitive conclusion that the best protection (i.e. higher TPI) is achieved by those samples which have fewer tendencies to shrink, crack or deform when exposed to fire.

4. Cone calorimeter assessment of bench-scale composites

4.1. Procedure

Having identified the top four performing technical fibres from the heat transfer tests, non-woven fabric samples were made using the process described previously. Grafil fibre has the ability to transport static charge which can lead to an accumulation of fibres on rollers and cards and on some occasions causing electrical malfunctions [28] . Only two out of the three composi- tion ratios were made containing Grafil because of these difficulties. Resulting non-woven fabric weights had an average of 296 g/m 2 and a coefficient of variation of 5.4%. Table 4 shows the variants of non-woven fabrics with the inclusion of the two commercial fabrics for comparison. Each fire blocking fabric was incorporated into bench-scale furniture composites consisting of the polypropylene upholstery fabric, the inter-liner and the CTU foam. Specimens were a nominal (100 100 50) mm 3 block as specified ISO 5660 [29] . The foam core was cut to within a 7 5% mass tolerance level in keeping with ISO 5657 [30] and BS 476: Part 13 [31] , but with no allowance for the thickness of the covering layers. Square pieces measuring (260 260) mm 3 for both inter-liner and upholstery fabric were cut, ensuring they were free from blemishes such as density variation or weave irregularities. The uniformity of the commercially produced upholstery was very consistent and all samples weighed were within 1% variation from the mean. The inter-liners contained more irregularities, however they were within 7 5% tolerance. The shape for the fabric shells were transferred by tracing around a template modified from ISO 5657. Samples were tested using the edge frame and grid to eliminate

Table 4 Sample identification, average density, CO, CO 2 , specific extinction area and heat of combustion for cone calorimeter tests.

Material or blend

Designation

Fabric weight (g/m 2 )

Specific extinction area

CO yield

CO 2 yield

Heat of

   

(m

2 /kg)

(g/g)

(g/g)

combustion

 

(MJ/kg)

Zirpro control Zirpro/Panox 95/5 Zirpro/Panox 85/15 Zirpro/Panox 75/25 Zirpro/Basofil 95/5 Zirpro/Basofil 85/15 Zirpro/Basofil 75/25 Zirpro/FR Viscose 95/5 Zirpro/FR Viscose 85/15 Zirpro/FR Viscose 75/25 Zirpro/Grafil 95/5 Zirpro/Grafil 85/15 Kevlar Panotex

Zirpro

307

502

0.065

1.859

27.9

297

513

0.055

2.094

31.4

Z/P

295

436

0.040

1.970

30.1

317

378

0.035

2.129

29.3

312

485

0.063

2.009

28.9

Z/B

297

409

0.054

1.898

30.0

275

412

0.043

1.457

26.6

327

503

0.069

2.133

32.0

Z/FRV

276

480

0.044

1.221

27.9

291

526

0.073

2.173

31.2

289

497

0.037

1.851

26.4

Z/G

266

416

0.041

1.813

26.0

Kevlar

280

355

0.021

2.162

31.5

Panotex

349

503

0.049

2.091

31.2

ARTICLE IN PRESS

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

243

movement of the specimens off the sample holder or expansion up to the conical heater coil. The use of the edge frame and grid resulted in slight compression of the sample within the holder; however, when samples were exposed to the radiant heat, the foam softened and receded from the heat source prior to combustion, therefore relieving the compression force. Previous work by Babrauskas et al. [32] showed that the edge frame had no effect on ignition, but generally resulted in lower peak and 180 s period heat release rates. Since the main objective of this work was to provide a comparative ranking rather than absolute values, the reduction in HRR due to the frame was not critical. Triplicate samples were tested in the cone calorimeter for each non-woven variant at an irradiance level of 50 kW/m 2 .

4.2. Results

4.2.1. Ignition and heat release rate

Ignition of the surface upholstery is difficult to control with the use of an underlying protective barrier, although it may have a slight influence if the melting upholstery is allowed to penetrate into a porous FR barrier, thereby temporarily reducing the amount of melted residue exposed to an external heat flux. Ignition times were between 15 and 20 s for all tests and melting of the fabric into the barrier was observed although it was difficult to quantify its effect. The isolated peak HRR from the covering layer is often of a relatively short duration compared to the remaining heat output and less than the peak caused by the combustion of the foam core. Instead, an effective fire barrier aims to reduce the peak HRR from combustion of material beneath it and/or increase the time at which this maximum HRR occurs so that the combination of increased time to peak HRR and decrease in maximum HRR is the optimal result. Fig. 4 shows the average HRR output from the unprotected foam/fabric sample and 85/15 blend composites. The results for the 85/15 blend are typical of the other two blends. A lowering of the peak HRR from the covering upholstery was observed for certain fibre types e.g. Basofil. Despite this occasional occurrence, the upholstery produced a HRR spike early in the combustion sequence, and, as might be expected, this peak occurred at nominally the same time due to the identical sample preparation and presentation to the heat source. Fig. 4 clearly distinguishes the peak HRR results (excluding the upholstery peak) that were used to identify combinations with superior fire blocking ability. The Z/G and Z/P blends were better performers, having a lower magnitude and occurring at increased time than the Z/FRV and Z/B blends.

occurring at increased time than the Z/FRV and Z/B blends. Fig. 4. Comparison of heat release

Fig. 4. Comparison of heat release rates of bench-scale composites tested in the cone calorimeter with a wool/FR fibre ratio of 85/15 and an unprotected composite.

wool/FR fibre ratio of 85/15 and an unprotected composite. Fig. 5. Cone calorimeter heat release rate

Fig. 5. Cone calorimeter heat release rate results for all wool/FR fibre sample combinations tested: (a) average peak value and (b) average time to peak.

Fig. 5 (a) shows the average peak HRR for each composite blend and confirms the major trend and order of the best performing composites. However, the expected trend of decreasing HRR with increasing technical fibre proportion is not conclusive. The density of inter-liner samples ( Table 4 ) shows there is a noticeable variation over the three blend ratios. In the case of the Z/B and Z/FRV blends the 95/5 composition obtained a higher density than the 85/15 or 75/25 blends, and this may have contributed to the reversal in the expected trend. Times to reach maximum HRR are displayed in Fig. 5 (b). The best performing combinations are those maximising the time to peak HRR and the same order of merit is obtained with the Z/G and Z/P blends outperforming the Z/FRV and Z/B blends and also the plain Zirpro treated wool. The dominant trend is clear in that the Z/G and Z/P combinations were superior. Although Z/G had a slight perfor- mance advantage over Z/P, Grafil fibre is a very rigid, high carbon content ( 4 99%) fibre and is not unlike glass fibre when in contact with human skin, causing discomfort and irritation [28] . Despite the success of the Z/G blend when exposed to fire, the difficulties in processing and the effect on humans made the Z/G combination a less desirable choice for use in furniture applications.

4.2.2. Combustion products

The use of multiple measured parameters goes some way to reinforce possible significant trends that may only seem sugges- tive from the data of a single parameter. Table 4 also gives a summary of average CO, CO 2 , specific extinction area (SEA) and heat of combustion obtained from the cone calorimeter. The Z/P blend at the 75/25 blend ratio was found to be the superior product with an SEA of 378 m 2 /kg, which is comparable to Kevlar

ARTICLE IN PRESS

244

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

having a SEA of 355 m 2 /kg. All other composites ranged 409– 526 m 2 /kg. For CO yield, the response to increasing the technical fibre proportion was not the same for all sample types. The Z/P and Z/B composites exhibited a decrease in average CO yield with an increase in technical fibre, whilst the Z/FRV and Z/G composites did not show much difference regardless of the blend. Only Z/B displayed a decreasing trend in CO 2 yield with increasing Basofil proportion. All other samples displayed a relatively constant CO 2 yield regardless of the inter-liner composition. This may suggest that the dominant proportion of the CO 2 produced originates from complete combustion of the fabric and foam and that the CO yield

is more likely to be influenced by the incomplete combustion of the fire resistant layer. This is also supported, in part, by the effective heat of combustion values which were likewise, relatively invariant.

5. Furniture calorimeter tests

5.1. Procedure

The results provided by the cone calorimeter identified the Z/P (75/25) blend as the best practical candidate material for the next

blend as the best practical candidate material for the next Fig. 6. Photographic timeline at indicated

Fig. 6. Photographic timeline at indicated times after ignition of the wood crib for the Control B sample.

after ignition of the wood crib for the Control B sample. Fig. 7. Photographic timeline at

Fig. 7. Photographic timeline at indicated times after ignition of the wood crib for the Zirpro/Panox (Z/P) C sample.

ARTICLE IN PRESS

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

245

stage of assessment. Furthermore, the Panox technical fibre has a lower relative cost compared to Grafil and is not significantly greater than Basofil and FR Viscose. The Z/P (75/25) blend inter- liner was tested in an ASTM E 1537-99 [34] comparable furniture calorimeter as a more realistic representation of a polyurethane filled chair, alongside unprotected (upholstery and foam without inter-liner) control chairs, Zirpro protected and chairs incorporat- ing Kevlar and Panotex commercial inter-liner products. Unfortu- nately there was only sufficient material to carry out one Kevlar and two Panotex tests; however, the results up to the 20 min mark strongly indicated that any expected variation would occur after this time and this was enough of an indication for comparison with the Z/P (75/25) blend. Cleary and Quintiere [33] recommend the use of small ignition sources in the assessment of the fire hazard of upholstered furniture in conjunction with a measure of peak HRR and time to peak in addition to ignition from larger sources such a gas burner. Greater reproducibility and consistency can be achieved using a burner, but this approach can mask the initial fire resistance of a sample to a realistically increasing heat source, and in many cases

the immediate severity of the ignition flame is sufficiently high to guarantee combustion, regardless of the existence of moderate, passive fire resistant measures. Initial behaviour of a sample assembly, whether on the surface or involving the substrates, is an important consideration which is more easily observed using ignition sources such as wood cribs. These give a small but steadily intensifying heat source that more strongly resemble the early stages of a real fire. Wood cribs to BS 5852 no. 5 cribs [10] were used in this research providing approximately 16 to 17 g of conditioned total mass even though use of the wood cribs deviates from the standard ignition source used in ASTM E 1537-99 [34] , in which a controlled gas burner flame impinges upon the sample for 80 s. Furniture cushions were constructed to the nominal dimen- sions of (509 457 102) mm 3 for the seat cushion and

(635 457 51) mm 3 for the

having a tolerance of 7 3 mm. These dimensions were taken from FAR 25.853(c) [17] and formed a convenient and standardised method of seat construction. The cushions were placed on a metal seat frame to FAR 25.853(c), so that the combustible portion of

back cushion with all dimensions

the combustible portion of back cushion with all dimensions Fig. 8. Separate sample cumulative energy curves

Fig. 8. Separate sample cumulative energy curves for each inter-liner material type tested in the furniture calorimeter.

material type tested in the furniture calorimeter. Fig. 9. Average heat release rate of from furniture

Fig. 9. Average heat release rate of from furniture calorimeter for each inter-liner material type.

ARTICLE IN PRESS

246

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

the test seat comprised of only the soft materials (i.e. foam, fabrics, liners and stitching). Each specimen was conditioned to (20 7 3) 1 C and (50 7 5)% RH prior to testing. Lint on the wooden crib was wetted with 1.4 7 0.1 ml of alcohol and ignited with a match with the cribs placed at the join between the vertical and horizontal cushions, mid-way across the seat. The location for the wood crib was typical of most full-scale test requirements and results from Mitler and Tu [35] suggest it was likely to give the earliest time to peak HRR and little effect on the peak HRR value. Combustion of the seat assembly was allowed to follow its own natural progression without any external interference until complete self-extinction was achieved. HRR was determined from oxygen, carbon dioxide and carbon monoxide gas data. All tests were recorded on video with still digital photographs taken intermittently over the test duration.

5.2. Results

5.2.1. Observations

It was observed that all unprotected samples collapsed and broke away from the frame during the burning process and continued to burn as pool fires. The break-up of the unprotected samples was subject to randomness, and hence an observed variability in the HRR history. Conversely, protected samples remained in place whilst releasing the melted polymer at a proportionately slower rate onto the tray underneath the seat assembly. There was no substantial pool fire contribution from the protected samples, although some of this could be the result of using a metal tray that may have acted as a heat sink, thereby cooling the melted residue. Figs. 6 and 7 show photographic time series of a control sample and a Zirpro/Panox (Z/P) inter-liner sample that illustrate the difference in the fire development between unprotected and protected samples. An indication of the manner in which the samples were consumed, encompassing the break-up and the consequent progression of combustion, can be observed in Fig. 8 where the obvious variability of the control samples cumulative energy curves is contrasted with the more consistent curves produced by those samples incorporating an inter-liner. The 100% Zirpro wool samples did not perform any better than the controls in terms of combustion severity, but did not break-up or separate from the frame as early or to the same extent.

5.2.2. Heat release rate

Fig. 9 shows the averaged HRR curves for each composite type, excluding the Control A sample data, as it was too far removed from the other two control sample output curves as illustrated in Fig. 8 . The HRR for the unprotected and 100% Zirpro treated wool inter-liner composites began to increase at between 2 and 3 min,

whereas those samples with effective barrier protection did not show an increase in HRR until approximately 20 min following ignition of the cribs. The time when HRR increases and also the magnitude of the peak HRR is substantially different between the two performing groups. The best performers did not peak much above 50 kW, while the remainder peaked at values between 210 and 240 kW.

5.2.3. Combustion products

The CO, CO 2 and smoke production in the furniture calorimeter tests are shown in Fig. 10 and support an earlier indication from the cone calorimeter tests that the carbon oxide gas emissions would reflect a similar trend to the HRR. The Z/P inter-liner matched the level of performance of commercial materials for at least 20 min. Carbon dioxide peak levels from the Z/P blend are similar to those from the commercial products and carbon

are similar to those from the commercial products and carbon Fig. 10. Average furniture calorimeter test

Fig. 10. Average furniture calorimeter test results for each inter-liner material type: (a) carbon dioxide; (b) carbon monoxide; (c) smoke production.

monoxide levels are at least twice those of the commercial products, but are significantly less than the control and Zirpro test results. Smoke production follows a similar pattern, where the Z/P results are almost indistinguishable from the commercial product protected samples for approximately 1400 s. The smoke response of Z/P from the furniture calorimeter matches the results from the cone calorimeter tests in that Z/P behaves similarly to Kevlar in terms of smoke contribution for this type of application and does so for a significant duration of the test.

5.2.4. Results after 1200 s and at test completion

Table 5 gives a total value for each measured response during the furniture calorimeter tests. After 1200 s (20 min), the control and Zirpro samples had essentially finished burning, whereas the Z/P results had a very similar response to the Kevlar and Panotex samples with only a small proportion of their final yields having

ARTICLE IN PRESS

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

Table 5 Summary of furniture calorimeter results after 1200 s and at test completion.

247

 

After 1200 s

 

At test completion

Sample

Mass

Total

Total smoke (m 2 )

CO

CO 2

Mass

Total

Total smoke (m 2 )

CO

CO 2

lost (kg)

energy (MJ)

yield (g)

yield (g)

lost (kg)

energy (MJ)

yield (g)

yield (g)

Control A

0.903

40.5

172

3.19

2279

1.115

46.8

212

4.13

3015

Control B

1.630

69.2

457

5.40

4463

1.637

69.2

457

5.60

4506

Control C

1.422

55.1

477

4.48

3493

1.428

55.1

482

4.55

3532

Mean

1.318

54.9

369

4.36

3412

1.393

57.0

384

4.76

3684

S.D.

0.374

14.4

171

1.11

1094

0.263

11.4

149

0.76

757

Zirpro A

1.813

65.5

499

3.93

4360

1.813

66.0

506

3.94

4543

Zirpro B

1.914

66.1

420

5.57

4165

1.914

66.5

420

5.86

4260

Zirpro C

2.040

70.4

568

5.84

4576

2.040

70.4

577

5.95

4625

Mean

1.922

67.4

496

5.11

4367

1.922

67.6

501

5.25

4476

S.D.

0.114

2.7

74

1.03

206

0.114

2.4

79

1.14

192

Z/P A

0.202

4.7

0

0.63

322

1.712

50.9

273

11.06

3313

Z/P B

0.320

5.4

102

0.90

343

1.712

41.7

373

12.80

2677

Z/P C

0.310

6.2

44

1.40

472

1.637

41.0

307

11.58

2781

Mean

0.277

5.4

49

0.94

379

1.687

44.5

318

11.81

2924

S.D.

0.065

0.8

51

0.41

81

0.043

5.5

51

0.89

341

Panotex A

1.197

36.0

228

1.10

139

1.197

36.0

228

8.07

2426

Panotex B

0.201

4.3

13

0.75

170

0.201

4.3

13

2.82

472

Mean

0.699

20.2

121

0.93

155

0.699

20.2

121

5.45

1449

S.D.

0.704

22.5

152

0.25

22

0.704

22.5

152

3.71

1382

Kevlar

0.271

3.6

47

0.96

185

0.486

5.7

109

4.63

382

been recorded. Beyond 1200 s however, the Z/P barrier no longer maintained its protective quality. Table 5 also shows the same parameters after the samples have self-extinguished or after the protection of the barrier fabric was finally overcome. The period for self-extinction was around 50–60 min for the protected, slow burning samples. Fig. 8 shows that beyond 1500 s the Z/P samples begin to produce more CO, but this occurs at a late stage in contrast to the more combustible samples which reach peak CO emissions earlier and with higher magnitude. The increased CO 2 and CO production in latter stages is most likely the result of decomposition of the FR wool component to an extent where protection of combustible material is no longer possible. Flaming was then allowed to increase and react with the high carbon content of the technical fibre portion of the inter-liner in addition to the more exposed flammable components, thereby producing larger carbon oxide quantities.

6. Conclusions

A blend ratio of (75/25), Zirpro wool/Panox fibres forming a

protective barrier material has been identified which offers a comparable level of performance to established inter-liner products. The blend identified has higher wool and lower technical fibre proportion than inter-liners previously developed

for aircraft seat fire protection. Although there is scope for further research, the results demonstrate that the fire safety of poly- urethane foam-filled furniture can be improved by incorporating a low-cost inter-liner thereby providing a solution which could be adopted by the New Zealand furniture manufacturers.

In order to develop the inter-liner further it is recommended

that fabric blends are manufactured using fully commercial processes to ensure the uniformity and density of blended fibre in the material. Only soft components of idealised furniture mock-

ups were tested and only one type of foam was used. Any future work should consider actual furniture models lined with the fire barrier. This future work should examine how connections and stitching might affect the performance of the inter-liner and this

process will likely affect the cost of manufacture. Due to the environmental impact of the Zirpro treatment [19] , it would be advantageous to identify alternative fire retardant treatments for wool. Promoting the introduction of inter-liner materials does not imply a complete solution to solving the combustibility problems associated with foam-filled furnishings. A more thorough approach would be to incorporate an integrated fire resistance into every combustible sub-component of the furniture item i.e. upholstery, foam and frame materials, however this would need a cost/benefit study along the lines of the study carried out by Wade et al. [36] .

Acknowledgements

The authors would like to thank Grant Dunlop for his assistance during the furniture calorimeter tests. The Fire Engineering programme at the University of Canterbury is supported by the New Zealand Fire Service Commission.

References

[1] I., Miller Human behaviour contributing to unintentional residential fire deaths 1997–2003. Fire Research Report no. 47, H.C. Ltd, New Zealand Fire Service Commission, 2005. [2] C.R. Wong, in: Contribution of upholstered furniture to residential fire fatalities in New Zealand. Masters of Engineering in Fire Engineering project report, University of Canterbury, Christchurch, New Zealand, 2001. [3] P.E. Ingham, in: The flammability of polyurethane upholstered furniture, Part III House fire tests. Communication no. 73, Wool Research Organisation of New Zealand Inc., Lincoln, New Zealand, 1981. [4] P.E. Ingham, in: The flammability of polyurethane upholstered furniture, Part II Tests with larger ignition sources and improved foams. Communication no. 72, Wool Research Organisation of New Zealand Inc., Lincoln, New Zealand, 1981. [5] P.E. Ingham, M.J. Goddard, A.L. Grueber, in: The influence of fabric coverings on the flammability of polyurethane-upholstered furniture, Communication no. 66, Wool Research Organisation of New Zealand Inc., Lincoln, New Zealand, 1979.

ARTICLE IN PRESS

248

P. Martini et al. / Fire Safety Journal 45 (2010) 238–248

[6] H.R. Denize, in: The combustion behaviour of upholstered furniture materials in New Zealand. Masters of Engineering in Fire Engineering project report,

University of Canterbury, Christchurch, New Zealand, 2000. [7] A.R., Coles Flammability of upholstered furniture using the cone calorimeter. Masters of Engineering in Fire Engineering project report, University of Canterbury, Christchurch, New Zealand, 2001. [8] C.M., Fleischmann, F. Chen Radiant ignition of upholstered furniture, in:

Proceedings of the International Conference on Engineered Fire Protection Design, 243–252, 2001. [9] Private communications between New Zealand Ministry of Consumer Affairs, New Zealand Furniture Manufacturers Federation Inc., New Zealand Fire Service and Wool Research Organisation of New Zealand, 1988–

1992.

[10] British Standard BS 5852, Methods of test for assessment of the ignitability of

upholstered seating by smouldering and flaming ignition. British Standards Institution, 1990. [11] Technical Bulletin TB 117, Requirements, test procedures and apparatus for testing the flame retardance of resilient filling materials, State of California Department of Consumer Affairs, Bureau of Home Furnishings and Thermal Insulation, 2000. [12] Technical Bulletin TB 116, Requirements, test procedures and apparatus for

testing the flame retardance of upholstered furniture. State of California Department of Consumer Affairs, Bureau of Home Furnishings and Thermal Insulation, 1980. [13] The Bureau of National Affairs Daily report for executives, Furniture firms back flammability standard, but demand requirements must be practical, 119, p. A10–A11, 2002. [14] M.M. Hirschler, S. Shakir, Comparison of the fire performance of various upholstered furniture composite combinations (fabric/foam) in two rate of heat release calorimeters: Cone and Ohio State University instruments, Journal of Fire Sciences 9 (3) (1991) 223–248, doi:10.1177/073490419100900303. [15] J.A. Gallagher, Inter-liner effect on the fire performance of upholstered materials, Journal of Fire Sciences 11 (1) (1993) 87–105, doi:10.1177/

[16] G.H. Damant, Recent United States developments in tests and materials for the flammability of furnishings, Journal of the Textile Institute 85 (4) (1994)

505–525.

[17] Advisory Circular 25. 853-1, Federal Aviation Administration, Washington, D.C., 1986. [18] L. Benisek, Zirpro wool textiles, Fire and Materials 8 (4) (1984) 183–195,

[19] A.R. Horrocks Textiles, in: A.R. Horrocks, D. Price (Eds.), Fire Retardant Materials, Woodhead Publishing Ltd. and CRC Press LLC, 2001. [20] A.R. Horrocks, S. Anand, in: Handbook of Technical Textiles, Woodhead Publishing, 2000 p. 559.

[21] The flammability performance of wool apparel. WoolNews.net, http://www. woolnews.net, 2008.

[22] S. Bourbigot, X. Flambard, Heat resistance and flammability of high performance fibres: a review, Fire and Materials 26 (2002) 155–168, doi:10.1002/fam.799. [23] Kermel Tech high performance fibre, http://www.kermel.com, (accessed May

2008).

[24] Carbonaceous fiber reinforced polymeric composites. World Intellectual Property Organization, WO/1989/003766, 1989. [25] BS 4735. Laboratory method of test for assessment of the horizontal burning

larger than 150 mm 50 mm 13 mm

(nominal) of cellular plastics and cellular rubber materials when subjected to a small flame, 1974. [26] V. Babrauskas, in: Materials & Products, Tables and Charts, Section 6/Chapter 17. In Fire Protection Handbook, twentieth ed., National Fire Protection Association, Quincy, MA, 2008. [27] British Standard BS EN 367, Protective clothing-protection against heat and fire method of determining heat transmission on exposure to flame, British Standards Institution. [28] Grafil Inc. Sacramento CA 95828, USA, Material Safety Data Sheet. Grafil, MSDS, 2002. [29] ISO 5660-1, Fire tests Reaction to fire, Part 1: Heat release rate from building products (cone calorimeter method). International Organization for Standardization, Geneva, 1993. [30] ISO 5657, in: Method of measuring the ignitability of products subjected to thermal irradiance., International Organization for Standardization, Geneva, Switzerland, 1986. [31] British Standard BS 476 Part 13, in: Method of measuring the ignitability of products subjected to thermal irradiance, fire test on building materials and structures., British Standards Institution, 1987. [32] V. Babrauskas, W.H. Twilley, W.J. Parker, The effects of specimen edge conditions on heat release rate, Fire and Materials 17 (2) (1993) 51–63,

characteristics of specimens no

[33] T.G., Cleary, J.G. Quintiere A framework for utilizing fire property tests. In Fire Safety Science, in: Proceedings of the 3rd International Symposium, University of Edinburgh, Scotland, 647–656, 1991. [34] ASTM E1537-99, in: Standard test method for fire testing of upholstered furniture., ASTM, American Society for Testing and Materials, West Conshohocken, USA. [35] H.E., Mitler, K.-M., Tu Effect of ignition location on heat release rate of burning upholstered furniture. in: Proceedings of the Annual Conference on Fire Research, NIST, Gaithersburg, MD, 1994. [36] C.A., Wade, M., Duncanson, D., O’Dea, C.R., Duncan Costs and benefits of regulating fire safety performance of upholstered furniture in New Zealand, New Zealand Fire Service Commission Research Report no. 35, 0-908920-94- 6, Wellington, 2003.