LABORATORY FLAMMABILITY TESTING

10 15

0 5 10 15 20 25 30 35 40

% volume of expanded plastic

% weight of unexpanded plastic

Material factor 1

Material factor 2

Material factor 3

Material factor 4

Figure 5.9. CEN prEN 12845 commodity material factor categories (from CEN, 2001)

plastics may fall into any of the three groups depending on density, the presence of plasticizing or flame retardant additives, and physical form. This illustrates the futility and inappropriateness of any rigid generic commodity classification scheme. Both Factory Mutual and NFPA realize that their generic classification schemes are more valuable for providing a preliminary indication of relative flammability than a firm irrefutable determination.

The European Committee for Standardization has a generic commodity classification scheme in its draft standard for automatic sprinkler systems (CEN prEN 12845, 2001). CEN has four commodity categories, which depend on the commodity material and storage configuration. The material factor depends on the expanded plastic and unexpanded plastic content of the commodity according to the regions delineated in Figure 5.9. If there is less than 5% plastic (both by weight and by volume) in the product and packaging, the commodity is designated as a Material Factor 1.

As the percent plastic is increased, the Material Factor increases to 2, 3, or 4. Cartoned commodi-ties are designated as Category I, II, III, or IV according to their Material Factor. Exposed plastic commodities are categorized according to their contents as explained in Annex B of the draft CEN standard. The physical nature of the material is also a factor if it is a solid block, a powder, or an open (low volume fraction) material. Annex C of the CEN standard is a listing of the categories of numerous specific commodities. Although the same factors are utilized in the NFPA commodity classification scheme, there is no direct relationship between a CEN Category IV commodity and a NFPA/FM Class IV commodity, and likewise for the other three categories.

5.4.2 LABORATORY FLAMMABILITY TESTING

Warehouse storage fires involve a combination of flame spread over the storage array and burning into the storage array. Since flame spread is the propagation of an ignition boundary, laboratory flammability tests measuring flame spread and heat release rates per unit area usually also measure sample ignition characteristics. Several test methods, described, by Tewarson (1995) and Drysdale (1998), have been developed for this purpose. The ignition tests involve measuring ignition delay times, t_ig, during exposure of material samples to a series of radiant heat fluxes.

Hamins and McGrattan (1999) have measured ignition times of 4 mm thick corrugated paper samples exposed to radiant heat fluxes in the LIFT apparatus (designated as ASTM E 1321 and ISO 5658) and the Cone Calorimeter (designated as ASTM E 1354 and ISO 5660). Their data

0.00

Thermally thin w H & M rho*c*delta

Figure 5.10. Inverse time to ignition versus heat flux for corrugated paper

are plotted in Figure 5.10 in the form of t_ig^−1/2 versus radiant heat flux. A linear relationship on this plot implies that the material is behaving as a thermally thick material, for which the ignition time should satisfy the following theoretical equation (Drysdale, 1998, p. 218)

tig = π

4kρc(Tig− T0)²

q_e^2 [5.4.1]

where k, ρ, and c are the thermal conductivity, density, and specific heat of the sample material, Tig is its ignition temperature, T0 is the ambient temperature, and q_e^ is the radiant heat flux impinging on the sample (and assumed to be absorbed by the sample). A best linear fit has been obtained over the heat flux range 28 – 75 kW/m²for the data in Figure 5.10.

If the material responds to the imposed heat flux as if it is thermally thin (uniform temperature), and if heat losses at the exposed surface are neglected, the ignition time should vary as:

tig= ρcδ(Tig− T0)

q_e^ [5.4.2]

where δ is the sample thickness. If convective heat losses at the exposed and unexposed surface are included, the equivalent thermally thin equation is (Drysdale, 1998, p. 214)

tig= ρcδ

where h is the convective heat transfer coefficient. Introducing the critical heat flux, q_cr^ (minimum heat flux for ignition at infinite exposure time) is introduced, equation [5.4.3] can be rewritten as

tig = ρcδ

Hamins and McGrattan (1999) fit the high heat flux portion of their corrugated paper ignition data to equations [5.4.3] and [5.4.4], and found q_cr^ = 14.5 kW/m2, Tig = 370^◦C, h= 0.042 kW/m²,

and ρcδ= 0.98 kJ/m^2oK. In a later paper (McGrattan et al. 2000), they report ρcδ= 1.5 ± 0.4 kJ/m^2oK. Both of their curve fits are shown in Figure 5.10. It is clear that from Figure 5.10 that the thermally thick approximation fits the high heat flux range of the data much better than equation [5.4.4]. This is because the ignition times at these high heat fluxes are small, and the corresponding heat conduction thickness, (ktig/ρc)^1/2, is smaller than the corrugated paper thickness.

Equation [5.4.4] has also been fit to the portion of the data in Figure 5.10 corresponding to heat fluxes less than 25 kW/m². Since the curve does seem to fit the data well in this range, the thermally thin approximation appears to be valid for these small heat fluxes and corresponding long ignition times. Thus, the corrugated paper can be treated as either thermally thin or thermally thick, depending on the heat flux range of interest. This is consistent with the observation of Silcock and Shields (1995) that many sample materials are intermediate between thermally thin and thermally thick in terms of best-fit correlations to inverse ignition time data.

One popular flame spread test is the ASTM E 162 (1983) test in which the flame spread is downward on an inclined sample opposite a radiant heat source. The result of the ASTM E 162 test is a flame spread index which is a product of a flame spread factor (proportional to the flame spread rate) and a heat generation factor. Underwriters Laboratories recently measured flame spread indices for a variety of warehouse container materials used by the US Air Force (‘Flammability Test Method/Requirements for Packaging Materials’, 1988). Downward flame spread rates were greatest for fiberboard materials (10 – 15 in/min or 25.4 – 38.1 cm/min) and smallest for medium density polyethylene (about 5 in/min, or 12.7 cm/min). Since results in this test configuration are influenced by the melting and dripping of thermoplastics such as polyethylene, other test configurations may be more relevant to warehouse commodities.

Time to ignition data for treated and untreated cardboard carton materials have been obtained by Khan (1987) and Tewarson (1995) using the FMRC Flammability Apparatus. Tewarson (1995) correlated all his data with the following version of equation [5.4.1].

t_ig^−1/2=

√4/π

T RP(q_e^− qcr^) [5.4.5]

in which T RP = (kρc)^1/2(Tig− T0) is called the thermal response parameter. Tewarson and Khan reported values of q_cr^ for ordinary corrugated paper sheet and for flame retardant corrugated sheet of 10 kW/m²(0.88 Btu/sec-ft²) and 15 kW/m²(1.32 Btu/sec-ft²), respectively. The untreated paper critical flux of 10 kW/m², which is 4.5 kW/m²smaller than the value measured by Hamins and McGrattan. The difference may be due to the black paint applied to the surface of the samples tested by Tewarson and Khan in order to increase the surface absorptivity. The value of TRP reported by Tewarson for lightweight corrugated paper is 152 kW-s^1/2/m².

Khan correlated the time to ignition data for ordinary corrugated sheets by

t_ig⁻¹= 10⁻⁴q_e^{2 + 0.45qe^} for qe^>20 kW/m² [5.4.6]

and the data for fire retardant corrugated sheets by

t_ig⁻¹= 0.001qe^ for q_e^>30 kW/m² [5.4.7]

Equation [5.4.7] is an attempt to account for both the thermally thin and thermally thick behavior of the corrugated sheets at small and large heat fluxes, respectively. Equation [5.4.7] is more appropriate for a thermally thin solid, which the fire retardant sheets apparently approximate because of their relatively long ignition times.

Figure 5.11 is a plot of t_ig^−1/2versus q_e^for the various types of corrugated paper tested by Khan and Tewarson. The corrugated sheets treated with fire retardant coatings have substantially longer

0.00

Correlation of H & M data Tewarson light corrugated Khan ordinary corrugated

"Khan fire retardant corrugated"

Tewarson heavy corrugated Tewarson: heavy 10% coating

Tewarson heavy corrugated 10%

Khan fire retardant corrugated Tewarson heavy corrugated

Figure 5.11. Correlations for inverse time-to-ignition for different grades of corrugated paper

ignition times (smaller t_ig^−1/2 values) than the ordinary corrugated sheets, based on Tewarson’s reported values of TRP. This implies that flame spread rates over the fire retardant corrugated cartons will be significantly slower than those over ordinary cartons.

Laboratory measurements of fire heat release rates are being made by a variety of methods incorporating externally applied radiant heat fluxes. The most commonly used apparatus for this purpose is the cone calorimeter (ASTM E1354). The standard test sample for cone calorimeter testing is a 10× 10 cm horizontal surface exposed to the heat flux. The FMRC flammability tests (Tewarson, 1982) using a similar apparatus have generated heat release rate data for 80 cm² (12.4 in²) by 2 – 5 cm (0.8 – 2 in) thick horizontal samples of numerous polymers and packaging materials exposed to external heat fluxes and sometimes to enhanced oxygen atmospheres. Data for the asymptotic (with increasing heat flux) heat release rate for various polymers span the range 120 – 1200 kW/m² (10.6 – 106 Btu/sec-ft²).

Hamins and McGrattan (1999) have conducted cone calorimeter tests with a 10× 10 × 10 cm miniature corrugated paper cell containing a polystyrene cup used in fabricating the prototype Group A plastic commodity. The heat release rate per unit area at an imposed radiant heat flux of 50 kW/m²was 400 – 500 kW/m². They later report (McGrattan et al. 2000) that their computer model provides a better match to large-scale test data when they use a specific heat release rate of 600 kW/m²for the prototype Group A plastic. One reason for the higher heat release per unit area in large-scale tests is that the heat flux impinging on the vertical surface of the cardboard boxes from the flames below is typically 90 – 100 kW/m².

Another laboratory flammability test is the ASTM E 906 apparatus (ASTM E-906, 1983) in which a 10× 15 cm (3.9 × 6 in) horizontally oriented sample or a 15 × 15 cm (6 × 6 in) vertically oriented sample is introduced into an environmental chamber with a radiant heat source. Under-writers Laboratories (‘Flammability Test Method/Requirements for Packaging Materials’, 1988) used the ASTM E 906 apparatus to measure heat release rates for various Air Force warehouse container and cushioning materials exposed to a radiant heat flux of 35 kW/m² (3.1 Btu/sec-ft²) on the vertical exposed surface of the sample. Fiberboard and wood container materials had the lowest heat release rates (less than 220 kW/m² or 19.4 Btu/sec-ft²), while polyethylene had the

highest heat release rates (it melted and burned in a pool). Among the cushioning materials tested, flexible polyethylene foam had the highest heat release rates while fire retarded rubberized hair had the lowest heat release rates.

UL also measured heat release rates for 2 by 2 by 2 ft (61 by 61 by 61 cm) Air Force warehouse containers by testing under the UL product calorimeter (‘Flammability Test Method/Requirements for Packaging Materials’, 1988). The 30 sec average peak heat release rates were correlated to the corresponding ASTM E 906 bench scale heat release rates. A linear correlation seemed to represent most of the data except for the bins with polyethylene foam cushioning material which dripped and generated relatively large heat release rates in the ASTM E 906 tests, but was confined to the bins in the product-scale tests. This inconsistency notwithstanding, Underwriters Laboratories recommended using the ASTM E 906 heat release rates and smoke release rates as a basis for classifying warehouse commodities for use in un-sprinklered Air Force warehouses.

However, there have not been any large scale tests to confirm the UL recommended classifica-tion scheme. Moreover, there is much more interest in classifying commodities for sprinklered warehouses than un-sprinklered warehouses.

Factory Mutual Research Corporation has embarked on a long-range research program to develop suitable bench-scale flammability tests for classifying commodities for sprinklered ware-houses. Toward this end, Tewarson (1995) and his co-workers have recently extended the con-ventional horizontal sample measurements to vertically oriented samples and vertically stacked miniature packaging boxes with upward flame propagation. They have also started conducting tests with water sprays and films applied to the commodities. Data from both the horizontal samples and the vertically oriented configurations are currently used as a screening tool to sug-gest appropriate materials for testing in the small unit load storage array tests described in Section 5.4.3. Laboratory cone calorimeter tests with water application are also being conducted now by several other investigators (for example, Hietananiemi et al. 1999). Results to date indi-cate that the relative effect of the water application depends both on the material and the method of application (nozzle versus perforated pipe), as well as the applied water density.