Flammability characteristics at applied heat flux levels up to 200 kW/m2

This paper documents the first of the two interrelated studies that were conducted to more fundamentally understand the scalability of flame heat flux, the motivation being that it has been reported that flame heat flux back to the burning surface in bench‐scale experiments is not the same as for large‐scale fires. The key aspect was the use of real scale applied heat flux up to 200kW/m² which is well beyond that typically considered in contemporary testing. The main conclusions are that decomposition kinetics needs to be included in the study of ignition and the energy balance for steady burning is too simplistic to represent the physics occurring. An unexpected non‐linear trend is observed in the typical plotting methods currently used in fire protection engineering for ignition and mass loss flux data for several materials tested and this non‐linearity is a true material response. Using measured temperature profiles in the condensed phase shows that viewing ignition as an inert material process is inaccurate at predicting the surface temperature at higher heat fluxes. The steady burning temperature profiles appear to be invariant with applied heat flux. This possible inaccuracy was investigated by obtaining the heat of gasification via the ‘typical technique’ using the mass loss flux data and comparing it to the commonly considered ‘fundamental’ value obtained from differential scanning calorimetry measurements. This comparison suggests that the ‘typical technique’ energy balance is too simplified to represent the physics occurring for any range of applied heat flux. Observed bubbling and melting phenomena provide a possible direction of study. Copyright © 2007 John Wiley & Sons, Ltd.     

Effects of variations in incident heat flux when using cone calorimeter test data for prediction of full‐scale heat release rates of polyurethane foam

The development of methods to predict full‐scale fire behaviour using small‐scale test data is of great interest to the fire community. This study evaluated the ability of one model, originally developed during the European Combustion Behaviour of Upholstered Furniture (CBUF) project, to predict heat release rates. Polyurethane foam specimens were tested in the furniture calorimeter using both centre and edge ignition locations. Input data were obtained using cone calorimeter tests and infrared video‐based flame area measurements. Two particular issues were investigated: how variations in incident heat flux in cone calorimeter tests impact heat release rate predictions, and the ability of the model to predict results for different foam thicknesses. Heat release rate predictions showed good agreement with experimental results, particularly during the growth phase of the fire. The model was more successful in predicting results for edge ignition tests than for centre ignition tests and in predicting results for thinner foams. Results indicated that because of sensitivity of the burning behaviour to foam specimen geometry and ignition location, a single incident heat flux could not be specified for generating input for the CBUF model. Potential methods to determine appropriate cone calorimeter input for various geometries and ignition locations are discussed. Copyright © 2014 John Wiley & Sons, Ltd.     

Radiative heat flux measurement uncertainty

As part of an effort to characterize the uncertainties associated with heat flux measurements in a fire environment, an uncertainty analysis example was performed using measurement data from a room corner surface products test that followed the guidelines of ISO 9705. Equations to model the heat transfer at the surface of a Schmidt‐Boelter (thermopile) type total heat flux gauge were selected for use to calculate the incident radiative flux from a total heat flux measurement. The effects of the heat flux measurement uncertainty sources were evaluated by conducting an uncertainty propagation on the resulting equation for incident radiation. For the model equations and the example conditions selected, the free‐stream temperature estimate and the heat flux gauge calibration constant were determined to be major uncertainty contributors. The study demonstrates how to systematically identify major sources of uncertainty for the purpose of reducing total uncertainty and thereby enhancing experiment design. Published in 2003 John Wiley & Sons, Ltd.     

A theoretical basis for flammability properties

Theoretical formulations are presented for the fire growth processes under external radiant heating. They included ignition, burning and energy release rate, and flame spread. The behaviour of these processes with external heating is described along with the critical conditions that limit them. These include the critical heat fluxes for ignition, flame spread and burning rate. It is shown how these processes and their critical conditions depend on a limited number of properties measurable by a number of standard test methods. The properties include heat of combustion, the heat of gasification, ignition temperature and the thermal properties of the material. Alternatively, the properties could be related to parameters easily found from data; namely: (1) the critical heat flux (CHF) for ignition; (2) the slope of the energy release rate with externally imposed flux, defined as heat release parameter (HRP); and (3) the ignition parameter, defined as thermal response parameter (TRP). It is further shown that the flame heat flux differences between small laminar flame ignition sources and larger turbulent flames can affect flame spread due to heat flux and ignition length factors. Finally, it is found that the critical energy release rates theoretically needed for ignition, sustained burning, and turbulent upward flame spread are roughly 13, 52, and 100 kW/m², respectively, and independent of material properties. Copyright © 2005 John Wiley & Sons, Ltd.     

Modelling of heat release rate in upholstered furniture fire

Many fatal residential fires started from burning upholstered furniture, and so upholstered furniture fire has been studied rather extensively in developed countries. As many upholstered furniture were made in China, the hidden fire risk should be studied more. In this paper, full‐scale experiments on the burning of upholstered furniture manufactured in China were conducted and analyzed. The oxygen consumption method was used to measure the heat release rate in a room calorimeter. An ignition source of a 20‐kW gasoline pool fire of 0.2‐m diameter was used to test square foam cushions and 4‐seater sofas. A model of heat release rate predicting upholstered furniture fire in a room was developed on the basis of earlier Swedish works. Results were then used to justify the application of the Combustion Behaviour of Upholstered Furniture model to predict the heat release rate of furniture manufactured in China. The numerical values of key parameters in the model were determined. It is proposed to build up a database that can be used to model heat release rates upon burning furniture. Detailed procedures are illustrated in this paper.     

Ceiling heat transfer during fire plume and fire impingement

The impingement of turbulent fires and fire plumes on a horizontal ceiling was considered. Free flame heights, impinging flame lengths along the ceiling and ceiling heat fluxes were measured for both unconfined and confined ceilings. The study was limited to the initial stage of ceiling heating by fire under conditions where convection dominates the flow. Fire sources were simulated by burning liquid methanol, ethanol, 1-propanol or n-pentane from the top surface of a cylindrical wick. Test variables include fire heat release rate of 50–7890 W, ceiling diameters of 610 and 660 mm, ceiling heights of 58–940 mm, wick diameters of 10–107 mm and curtain wall lengths (for confined ceilings) as large as the ceiling height. Simplified models were employed to suggest expressions for data correlation. Flame lengths increased up to 40% when the ceiling was confined. Ceiling heat fluxes were relatively independent of position in the stagnation region (radius along ceiling <20% of the ceiling height). Heat fluxes in the stagnation region for plumes were 25–40% of those measured for impinging jets at comparable conditions. In the ceiling jet region, at larger distances from the point of impingement, the heat flux decreased with increasing radius, in agreement with other studies. Confinement tended to increase ceiling heat fluxes in both regions. Ceiling heat fluxes for impinging flames and plumes were approximately the same, for flame lengths along the ceiling up to 25% of the ceiling height; however, stagnation point heat fluxes decreased for longer flame lengths.     

The role of flame flux in opposed-flow flame spread

The flame spread process is driven by the net heat flux to the specimen surface, including the flux from the flame itself. This flame flux is important since it comprises a major part of the driving force causing flame fluxes were obtained. The values which are reported do not appear consistent and show more deviation among materials than would be anticipated. The most common fire test used for obtaining engineering data on flame spread (ASTM E 1321) also is not formulated in terms of flame flux as a driving force. This motivated an experimental programme, whereby six materials have been studied using the flame spread geometry of the ASTM E 1321 test, but with additional instrumentation for recording heat fluxes. The flame fluxes obtained experimentally in this study show much less variation among materials than the comparable data from the literature survey.     

Influences of specimen size and heating mode on the ignitability of polymeric materials in typical small‐scale fire test conditions

Small‐scale fire tests including the Underwriters Laboratories 94 (UL94) vertical burning test and the cone calorimeter test are widely used. In this paper, the ignition times of materials heated by the conical heater of a cone calorimeter and the UL94 flame were measured. It was found that for polymer bars heated by the UL94 flame, the ignition time is relatively short and increases with the specimen thickness. But the contribution of the specimen thickness to the delay of the ignition time is limited. The intrinsic properties of materials play a more important role in the ignition time than the specimen thickness. In addition, respectively corresponding to one‐dimensional, two‐dimensional, and three‐dimensional heat transfer, three heating modes of the UL94 flame were presented and compared with the conical heater. It was found that whether the heat source is the conical heater or the UL94 flame, the ignition time depends on the heat flux and the multidimensional heat transfer. The ignition time decreases with the increasing heat flux, and the magnitude order of the ignition time might drop when the heating mode changes from one‐dimensional to multidimensional heat transfer. Copyright © 2011 John Wiley & Sons, Ltd.     

The generation of CO in bench-scale fire tests and the prediction for real-scale fires

Carbon monoxide (CO) is the single most important factor associated with deaths in fires; thus, predictions of CO developed in fires is an essential aspect of fire quantification. It is considered crucial to have correct CO prediction methods for post-flashover fire stages, since, in the United States at least, the majority of fire deaths are associated with fires which have gone to flashover. In this paper it is shown that the yiels of CO observed in real-scale fires are generally not related to either the chemical nature of the material being burned nor to the yield observed for the same material in bench-scale testing. Instead, the generation of CO in real-scale fires is determined largely according to the oxygen available for combustion, with thermal conditions of the fire plume also playing a significant role. This behavior is in sharp contrast to many other material fire properties, including yields of gases such as CO₂ and HCI, which can be predicated for real-scale fires from bench-scale results. Finally, results from various studies completed thus far indicate how effective prediction of real-scale CO yields may be accomplished. While bench-scale measurements are not necessary to predict real-scale CO, bench-scale toxic potency measurements can be in error if the CO component in them does not reflect on the real-scale CO yield. Thus, a method is developed whereby the bench-scale toxic potency measurements can be computationally corected to better approximate the toxic potencies measured in real-scale, post-flashover room fires. These techniques will, undobtedly, be further refined as additional experimental results become available.     

Evaluating ignition data using the flux time product

A protocol based on the flux time product (FTP)¹ is used to analyze ignition data obtained from the Cone Calorimeter under an impressed flux in the range 20–70 KWm^?2 for different orientations and modes of ignition for conditioned cellulosic materials. The mean, maximum and minimum ignition times are depicted graphically by orientation and mode of ignition. Flux time products, FTP indices, critical irradiances and estimates of the convective heat loss associated with a change in specimen orientation are derived using the mean time-to-ignition data. It is demonstrated that consideration of the thermal thickness of a specimen may not be necessary when the proposed FTP methodology is utilized to determine valid correlations between the time-to-ignition and the incident radiant flux.     

Further variations of smoke density with heat flux

The heat flux in the NBS smoke chamber has been varied from 1.0 to 5.0 W cm^?2 in 0.5 W cm^?2 steps for some fifteen natural and synthetic materials. These experiments confirm the results given in previous papers for a smaller heat flux range, namely that smoke density varies ove4r the temperature/heat flux range. The extended range studied ensured that a much higher proportion of the materials reached a heat flux at which ignition occurred. As before at this and higher heat fluxes, the experiments were repeated in the flaming mode to ensure better reproducibility. With a few exceptions the general shape of the specific optical density/heat flux plot was similar, with a steep rise to a peak and then a corresponding drop to a low value at high fluxes. At 5 W cm^?2the drop had not been reached in some cases, but it is assumed to occur later by inference from earlier work on a smaller-scale apparatus at high temperatures. Complex materials can give unusual plots during the transition from pyrolysis to combustion. This is true of certain polymers containing fire retardant when the pyrolysis of the additive can have a separate effect.     

PEEK polymer flammability and the inadequacy of the UL‐94 classification

Fire safety is dependent on reliable information on material properties, particularly relating to burning behaviour. The Underwriters Laboratory UL‐94 test is a widely used simple Bunsen burner test for vertically upward flame spread. Aryl polyetheretherketones (PEEK) are polymers of exceptional thermal stability, typically decomposing at around 600°C and forming 50% carbonaceous char residue. Tests on seven PEEK polymers, and two related materials, in independent laboratories have revealed large inconsistencies in both the final broad classification and the scatter within each set of test results. In many cases, this variance is so large that if samples from the same batch of many of the materials were repeatedly submitted to test laboratories, this would ultimately result in one set remaining below the maximum burn time criteria, and so meeting the least flammable V‐0 rating. Initial data are presented indicating that a larger ignition source actually results in shorter burning times and more consistent burning behaviour. The reported behaviour of PEEK indicates that the inconsistencies reported here are not a function of inconsistencies in the material itself but rather a consequence of the low applied heat flux of the test method being very close to the critical heat flux for ignition of the PEEK polymer, which is rather high. With higher applied heat fluxes, this generates sufficient heat for a protective char to form, creating an effective barrier to further flame spread. Copyright © 2011 John Wiley & Sons, Ltd.     

Thermal-balanced integral model for pyrolysis and ignition of wood

The pyrolysis and ignition of wood is of great importance to understand the initial stage of combustion, helping control the occurrence and spread of unwanted building and forestry fires. The development of a thermal-balanced model is introduced for examining the analytical relationship between the ignition time and external heat flux. The critical heat flux, one of the important fire-retardant characteristics of combustible solid, is determined from a correlation study between the ignition time and external heat flux. One of the thermal-balanced integral models, considering the effect of surface heat losses, average absorptivity and moisture content, is employed to give the prediction of surface temperature rise, ignition time and ignition temperature of the Aspen. The results show that the model readily and satisfactorily predicts ignition temperature and ignition time of wood with different moisture contents.     

Flammability studies of thermally resistant polymers using cone calorimetry

The effectiveness of a set of thermally resistant polymers was evaluated for aircraft applications using the cone calorimeter (ASTM E1354/ISO 5660) under heat fluxes simulating real scale fires. This study included eight developmental and commercial thermally resistant polymers available in the literature and/or marketplace. The polymers included were aromatic polyester, polyetherimide, fluorine‐containing polyetherketone, phosphorus and fluorine‐containing co‐polyetherketone, fluorine and phosphorus‐containing polyether, fluorine‐containing polyester, poly(dimethylsiloxane)etherimide and polysulfone. The effects of fluorine, phosphorus, silicon and sulfone group in polymers were examined. This evaluation was based on time to ignition, peak, average and total heat release rates obtained at an external heat flux of 50 kW/m². Other parameters such as effective heat of combustion, mass loss and rate of smoke and toxic gas evolution were collected during the cone calorimeter test. Copyright © 2000 John Wiley & Sons Ltd.     

Effect of the Re number on heat and mass transport in a catalytic monolith

The numerical investigation of inter-phase heat transfer in a catalytic combustor, under laminar flow regime at different values of the Re number, is performed by means of a 2D axi-symmetric model of a single monolith channel. Numerical results highlight that axial diffusion comes to play an important role in the ignition region also at high convective fluxes (high Re) due to the strong flow perturbation accompanying light-off, with the consequences that: (a) the ignition position is not a linear function of contact time, as it would be expected at high Re; (b) the heat and mass transfer between surface and bulk gas phase are non-linearly affected by Re, especially in the entrance and in the ignition regions.A previously developed correlation for Nu and Sh is, hence, extended to include the effect of the Re number on heat and mass fluxes, enabling the prediction of the local value of Nu and Sh in the main features, and in particular the enhancement in the ignition region and the dependence on Re.     

Comparison of cone and OSU calorimetric techniques to assess the flammability behaviour of fabrics used for aircraft interiors

Two test methods for measuring the heat release rate, HRR have been compared on fabric composites used for aircraft interior materials as side‐wall panels. These methods are based on the principles of direct measurement of the convective and radiant heat by thermopiles using an Ohio State University (OSU) calorimeter, and oxygen consumption using a cone calorimeter. It has been observed when tested by standard procedures, cone results at 35 kW/m² incident heat flux do not correlate with OSU results at the same heat flux. This is because in the cone calorimeter, the sample is mounted horizontally whereas the OSU calorimetric method requires vertical sampling with exposure to a vertical radiant panel. A further difference between the two techniques is the ignition source—in the cone it is spark ignition, whereas in the OSU it is flame ignition; hence, samples in the OSU calorimeter ignite more easily compared to those in the cone under the same incident heat fluxes. However, in this paper we demonstrate that cone calorimetric exposure at 50 kW/m² heat flux gives similar peak heat release results as the 35 kW/m² heat flux of OSU calorimeter, but significantly different average and total heat release values over a 2 min period. The performance differences associated with these two techniques are also discussed. Moreover, the effects of structure, i.e. type of fibres used in warp/weft direction and design of fabric are also analysed with respect to heat release behaviour and their correlation discussed. Copyright © 2005 John Wiley & Sons, Ltd.     

Effectiveness of pre‐applied wetting agents in prevention of wildland urban interface fires

The purpose of the work described in this paper is to investigate the effectiveness of pre‐wetting structures, dead fuels, and landscaping plants in preventing fire spread from wildland fires to structures. Critical fluxes for fire growth were determined using intermediate‐scale testing for three wetting agents (water, type A foam, and gel) applied to 10 landscaping plants conditioned to 20% moisture, a mulch material, and four external structural materials (vinyl siding, plywood siding, asphalt shingle roofing, and cedar shake roofing). The critical flux for fire growth values was determined at 3‐min heat radiation exposure and simultaneous 300‐mm long flame exposure. Test specimens were exposed to various durations and intensities of drying prior to exposing them to heat radiation. Application of water or foam provided no noticeable protection. Gel was effective in providing protection even after 60 min of laboratory condition drying but was less successful when exposed to fire weather simulating accelerated drying. Some uncertainty is associated with the results of this work because of the variability of landscaping plants and gel wetting agent application uniformity. The intermediate‐scale test results were verified using full‐scale testing. Copyright © 2012 John Wiley & Sons, Ltd.     

Ignition performance of new and used motor vehicle upholstery fabrics

This paper examines the standards for fire safety in transport systems and in particular the test method for the flammability of materials within passenger compartments of motor vehicles. The paper compares data from ignition tests conducted in the cone calorimeter and the FIST apparatus with tests conducted using the FMVSS 302 horizontal flame spread apparatus. Ten materials were selected as representative of those used as seat coverings of private and commercial passenger vehicles. The time to ignition of new and used materials subject to exposure heat fluxes between 20 kW/m² and 40 kW/m² was measured. The results from the ignition tests were analysed using thermally thick and thermally thin theoretical models. The critical heat flux for sustained piloted ignition was determined from the time to ignition data using the thermally thin approach. Derived ignition temperatures from both the thermally thick and thermally thin methods were compared with measurements using a thermocouple attached to the back surface of materials in selected tests. The flame spread rates in the FMVSS 302 apparatus were determined and a comparison was made between the performance of the materials in the flame spread apparatus, the cone calorimeter and the FIST. The results suggests that a critical heat flux criterion could be used to provide an equivalent pass/fail performance requirement to that specified by the horizontal flame spread test although further testing is needed to support this. Copyright © 2004 John Wiley & Sons, Ltd.     

Bench-Scale Tests on Simultaneous Ignition of Two Different Plastics Through “Bridge-Mixing”

“Bridge-mixing” was proposed earlier to explain why different combustibles might be ignited simultaneously during flashover. Different fuel vapors gasified from those combustibles were ignited by thermal radiation first. Thermal feedback from flames above the combustibles would then ignite the solid. This point is further studied in this article by bench-scale tests with a cone calorimeter. Sample cubes of two materials with different ignition temperatures were tested at a high thermal radiative heat flux of 70 kWm^?2. Materials selected are polymethylmethacrylate (PMMA) and polyvinyl chloride (PVC). The two sample cubes were placed on the cone tray at different separation distances. In this way, different mixing of gasified fuel vapors released from the plastics can be achieved. Those two combustibles were found be ignited simultaneously at shorter separation distances. This might explain why combustible items of different ignition temperatures appear to be ignited at roughly the same time during flashover for some scenarios. On the basis of this study, the mixing of gasified fuel vapors from different combustibles by the fire-induced airflow is a key factor in igniting each item. Geometry of the rooms, combustibles, and ventilation provisions are important factors affecting simultaneous ignition.     

The effects of specimen edge conditions on heat release rate

When bench-scale specimens are tested for heat release rate, it is generally of interest that the behavior of the specimen simulate, as much as is possible, that of a real-scale product performing in a real fire. A number of issues have been raised recently by workers trying to understand the optimal conditions of specimen preparation and mounting. In the present study a large number of materials were explored in the Cone Calorimeter to determine the effect of edge conditions and edge frames. It was found that by the use of an insulated edge frame, heat release rate values can be obtained which are slightly closer to expected true values. The testing procedure, however, is significantly more complicated. This makes the insulated edge frame useful for collecting specialized data for fire modeling, but not for conducting routine reaction-to-fire tests. For routine testing use, it is recommended: (1) that no edge frame needs to be used unless the test specimen presents special difficulties, such as due to intumescence; (2) that in those cases where the use of the steel edge frame is found necessary, the results should be reported on the basis of an effective exposure area of 0.0081 m². When reported on such a basis, the heat release rate results do not show a systematic bias, compared to results with no edge frame.