Heat release rate measurements of thin samples in the OSU apparatus and the cone calorimeter

ARTICLE IN PRESS

Fire Safety Journal 40 (2005) 628–645 www.elsevier.com/locate/firesaf

Heat release rate measurements of thin samples in the OSU apparatus and the cone calorimeter Robert Filipczak, Sean Crowley, Richard E. Lyon Fire Safety Branch, Federal Aviation Administration, William J. Hughes Technical Center, Atlantic City International Airport, NJ 08405, USA Received 24 March 2003; received in revised form 20 May 2004; accepted 30 May 2005 Available online 1 September 2005

Abstract The Ohio State University (OSU) apparatus and the cone calorimeter are two devices commonly used to measure the heat release rate (HRR) of materials and products in forced flaming combustion. Each operates on a different principle but is calibrated in the same way. However, HRR results from these two test methods do not agree in most cases. For the present study, the OSU was modified to measure oxygen consumption and sensible enthalpy (temperature rise) of the apparatus in addition to the usual sensible enthalpy of the exhaust gases during the test. After calibration, total sensible heat (exhaust gases+apparatus) and oxygen consumption methods gave similar results for thin samples in the OSU. However, OSU results for thin samples did not agree with results from the cone calorimeter (ASTM 1354/ISO 1556) unless the HRR history in the cone calorimeter was corrected for smearing that results from dilution of the combustion gases with air in the sample chamber, exhaust duct, and scrubbers and the response time of the oxygen analyzer. r 2005 Published by Elsevier Ltd. Keywords: Fire calorimetry; Fire; Flammability; Cone calorimeter; OSU calorimeter; Heat release rate; Heat release

Corresponding author. Galaxy Scientific Corporation, 3120 Fire Road, Egg Harbor Township, NJ,

USA. 0379-7112/$ - see front matter r 2005 Published by Elsevier Ltd. doi:10.1016/j.firesaf.2005.05.009

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

629

Nomenclature q qf;l m r c T TN v V E

Heat release rate (W) heat losses from the flame (W) mass flow rate (kg/s) density (kg/m3) heat capacity (J/kg K) temperature (K) steady-state temperature (K) volumetric flow rate (m3/s) volume (m3) heat of combustion of oxygen with typical organic materials, 13.1 MJ/ kg O2 f fraction of the heat of combustion lost to surroundings (dimensionless) OSU Ohio State University rate of heat release apparatus ASTM American Society for Testing and Materials ISO International Standards Organization h average convective heat transfer coefficient (W/m2 K) Kc empirical calibration factor for OSU (W/K) Ka empirical calibration factor for OSU (J/K) Ke empirical calibration factor for OSU (W/K) Y O2 oxygen mass fraction (dimensionless) t response time of instrument to 63% of full-scale deflection HRR heat release rate in flaming combustion (W/m2) Subscripts a air e c

OSU apparatus Air Exhaust gases Convection

1. Background Many approaches have been used to measure the rate at which heat is released during the burning of materials and products [1,2]. The devices used to measure the heat release rate (HRR) are fire calorimeters and they operate on a variety of principles including sensible enthalpy (temperature rise) of the gas stream or enclosure, with or without a substitution/compensation burner, and analysis of the combustion gases for excess carbon dioxide or depleted oxygen [1,2]. The Ohio State University (OSU) rate of heat release apparatus [1–4] is one these devices. In the standard method [3,4], the OSU apparatus estimates the HRR of a material from the sensible enthalpy (temperature) rise of the air flowing past a 15 cm
15 cm burning

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

630

specimen. Prior to adopting the OSU as a regulatory requirement in 1988 [5], the Federal Aviation Administration (FAA) [6,7] and others [8–14] conducted testing that showed that oxygen depletion and sensible heat (thermal) methods of measuring HRR in the OSU apparatus produced different values for the same materials even after calibrating each method with the same methane burner. Although the two methods of measuring HRR in the OSU were generally proportional [6–14], large differences were observed for materials that burned quickly or with a smoky flame [11–14], in which case oxygen depletion usually produced significantly higher HRR. The transient response of the OSU apparatus to a HRR input was studied as a possible source of the discrepancy between the sensible heat and oxygen consumption methods [8,10,13–15]. However, the OSU dynamics alone could not account for all of the difference which varied from sample to sample [2,8–14]. The variability of results between HRR methods and between samples prompted some laboratories to adopt oxygen consumption calorimetry as the preferred method for measuring HRR in the OSU apparatus [10–14,16,17]. Later research [18,19] showed that adding the sensible enthalpy (temperature change) of the fire calorimeter enclosure to the sensible enthalpy of the air stream accounted for all of the heat of flaming combustion, and agreement with oxygen consumption measurements was thus demonstrated [19]. Most aircraft interior materials are thin or low density and burn quickly in a fire calorimeter, so a fast instrument response is necessary to capture the maxima in HRR. Moreover, HRR of materials is the main fire hazard in an aircraft cabin [5,7], so accurate methods of fire calorimetry are needed to test thin samples. The present work compares the total sensible enthalpy [18,19] and oxygen consumption methods of measuring HRR in the OSU apparatus and examines the conditions under which these results in the OSU are comparable to those obtained in a cone calorimeter for the thin (p1 mm) samples typically used in aircraft interiors.

2. Approach 2.1. Sensible enthalpy of the air stream The methods of measuring HRR in flaming combustion have been reviewed in detail [1,2] for fire calorimeters including the OSU [3,4] and cone calorimeter [20,21]. The sensible enthalpy method for determining the heat release rate q (W) in fire calorimeters derives from the relationship [1] 1 q  qf;l  me cðT e  T 1 e Þ ¼ re ce ve ðT e  T e Þ,

(1)

where qf,l is the heat loss from the flame to the apparatus due to radiation and convection, me is the mass flow rate of air and combustion products having density, heat capacity, and temperature re, ce, and Te, respectively, that are exhausted from the apparatus at volumetric flow rate ve (m3/s), and T 1 is the steady-state  temperature of the exhaust gases in the absence of sample burning. If the heat losses from the flame to the apparatus are independent of the type of fuel and are a

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

631

constant fraction f of the heat of combustion, qf;l  fq.

(2)

Eqs. (1) and (2) can be combined q

m e ce ðrvcÞe 1 ðT e  T 1 ðT e  T 1 e Þ ¼ e Þ  K e ðT e  T e Þ. 1f 1f

(3)

The sensible enthalpy method used to measure HRR in the OSU apparatus [1–4] is based on Eq. (3). In practice, the individual thermophysical properties are not measured directly but are lumped into an empirical factor Ke that is assumed to be constant and is obtained experimentally from a methane burner calibration. Tewarson [22] has shown that f ¼ 0:14 for a methane diffusion flame (e.g., the calibration burners in the OSU and cone calorimeter), while f  0:2 for oxygenated polymers such as polyoxymethylene and polymethylmethacrylate that burn cleanly with little soot formation and, hence, a non-luminous flame. However, the vast majority of plastics, including those used in aircraft interiors, are aliphatic and aromatic hydrocarbons that burn with a luminous (sooty) flame so that the radiant fraction of the heat of combustion for most of these materials is in the range f ¼ 0:420:6. The magnitude and range of f for common materials shows that Ke in Eq. (3) as determined from a methane calibration will underpredict the HRR of common plastics and their composites by as much as 50% for a given temperature rise, the exact amount depending on the type of fuel. The difference in HRR measured in the OSU apparatus by the sensible enthalpy method (i.e., Eq. (3)) and the oxygen consumption method [6–14] is a result of the fact that only the sensible heat of the exhaust gases is measured in the former. During flaming combustion, a fraction of heat f is temporarily absorbed by the apparatus walls so the proportionality between q and T e  T 1  in Eq. (3) is material specific [6–14] and time dependent [8,10,15,18]. The following methods have been proposed to eliminate these shortcomings of the standardized method [3,4] for measuring HRR in the OSU apparatus using sensible enthalpy of the exhaust gases. 2.2. Total sensible enthalpy If the rate of energy accumulation (sensible enthalpy rise) of the OSU apparatus is maca(dTa/dt) and this is equal to the heat loss from the flame to the walls qf,l minus heat loss of the walls to the air stream in the apparatus hðT a  T 1 a Þ, then the lumped energy balance for the OSU apparatus during flaming combustion is ra V a ca

dT a ¼ qf;l  ha Sa ðT a  T 1 a Þ, dt

(4)

where ra, Va, ca, Ta, T 1 e , ha, and Sa are the density, volume, heat capacity, instantaneous temperature, steady-state temperature, convective heat transfer coefficient, and surface area of the apparatus, respectively. Defining the temperature change of the apparatus, DT a ¼ T a  T 1 a , the heat losses from the flame in the OSU

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

632

chamber are d d DT a ðT a  T 1 , (5) a Þ ¼ K c DT a þ K a dt dt where KcDTa is the heat transferred from the apparatus test chamber to the external environment and air stream by convection and Ka dDTa/dt is the rate of sensible enthalpy change of the apparatus where Kc and Ka are effective parameters determined by calibration. Combining Eqs. (1) and (5) gives the energy balance for flaming combustion in the OSU in terms of the total sensible enthalpy change of the combustion gas stream and the apparatus qf;l ¼ ðhSÞa ðT a  T 1 a Þ þ ðrVcÞa

d DT a . (6) dt Dr. Edwin E. Smith of OSU [18] proposed Eq. (6) as a means to account for transient heat exchange between the air stream and OSU apparatus during testing and used an electrical resistance heating panel (f ¼ 0:8020:85) and a methane burner (f ¼ 0:14) to determine the constants Ke, Kc, and Ka in Eq. (6) experimentally. Using this total energy balance approach, Smith obtained good agreement between HRR histories calculated from sensible enthalpies (Eq. (6)) and those computed from the methane flow rate to a calibration burner for a squarewave HRR history [18]. Smith did not measure HRR by oxygen consumption in the OSU to validate the total sensible enthalpy method for more complex HRR histories as did Moussan et al. [19] in an apparatus similar to the OSU. q ¼ K e DT e þ K c DT a þ K a

2.3. Chemical heat of combustion To complete the set of fire calorimetry equations used in this study, the heat released (HR) by the chemical reaction of oxygen with fuel species as measured by oxygen consumption calorimetry is   q ¼ E mair Y aO2  me Y eO2  Eme DY O2 , (7) where E ¼ 13:1 kJ=gO2 is the heat of combustion of oxygen with typical organic fuels, mair, me are the mass flow rates, and CaO2 , CeO2 , are the oxygen mass fractions of the incoming air and exhaust gases, respectively, and DCO2 is their difference. The oxygen consumption technique for determining HRR is the basis for standard methods of fire calorimetry that use a conical radiant heater [20,21], but the OSU apparatus has also been modified and used for oxygen consumption calorimetry [16,17,23]. In the present study, a cone calorimeter was used to test thin samples at reduced air flow rate to improve sensitivity. A first-order transfer function [24,25] was used to deconvolute (correct) the HRR history in the cone calorimeter for instrument response:   d DY O2 ðtÞ q ¼ Eme DY O2 ðtÞ þ t . (8) dt

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

633

The time constant t is the time required for the oxygen concentration to reach 63% of the equilibrium value for a step change in HRR. The time derivative of the oxygen concentration in Eq. (8), d DY O2 =dt, was evaluated using a 5-point central difference formula [25].

3. Materials and sample preparation All gases used for calibration and testing were high purity (499.5%) grades obtained from Matheson Gas Products. Single-ply fiberglass-reinforced epoxy specimens were prepared as follows. The diglycidylether of bisphenol-A (BPA) (DGEBA, DER 332, The Dow Chemical Company) was warmed to melting (60 1C) and two parts by weight of 2-ethyl-4methylimidazole (EMI-24, Air Products) per hundred parts resin (phr) was added to the DGEBA and mixed until homogeneous. The mixture was hand-impregnated into E-glass fabric (0.22-mm-thick, Style 6781 fiberglass fabric, 8HS weave, areal weight 304 g/m2, BGF Industries Inc.) and cured in a heated platen press (Carver) under 79 MPa pressure at 150 1C for 45 min to make a single layer, fiberglass fabricreinforced epoxy specimen. Specimens for OSU testing were cut to 152 mm
152 mm. Specimens for cone calorimeter testing were cut to 100 mm
100 mm. These sample dimensions were used to normalize all heating rates for surface area. Bisphenol-A polycarbonate (BPA PC) samples, 152 mm
152 mm, were cut from a commercial 1.6-mm-thick sheet (LEXAN, General Electric Plastics). Samples of a fire-resistant polycarbonate made from 1,1,-dichloro-2,2-bis(4-hydroxyphenyl)ethylene (bisphenol-C polycarbonate (BPC PC), The Dow Chemical Company) were compression-molded from granular resin in a heated platen press at 300 1C to a final size 152 mm
152 mm
1.6 mm for testing in the OSU apparatus (see Fig. 1).

4. Methods 4.1. Sensible heat of combustion from gas stream The OSU apparatus is constructed (see Fig. 1) and calibrated for sensible heat of the combustion gas stream according to the Federal Aviation Administration procedure [3] for measuring the HRR test of aircraft cabin materials as specified in chapter 14 of the code of federal regulations, part 25 (14 CFR Part 25). In this procedure, the baseline flow rate of methane to the diffusion burner is 1 L/min and the methane flow rate is increased to 4, 6, and 8 L/min in square-wave pulses of 2 min duration with 2 min allowed between pulses to re-establish baseline. Thermopile voltage is measured in the exhaust stack for the entire burner calibration sequence and the data averaged to obtain the HRR factor C (W/mV) for the apparatus, analogous to Ke in Eq. (3).

ARTICLE IN PRESS 634

R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

Thermopile Hot Junction Air Baffle

EXHAUST STACK

Oxygen Probe 7.6 cm Inner Cone Outer Cone

PYRAMIDAL SECTION AIR MANIFOLD

HOLDING CHAMBER

SILCON CARBIDE HEATING RODS

Sample Position Apparatus Thermocouple (Wall Surface Mounted) Thermopile Cold Junction

Air Distribution Plate

Air Inlet

Fig. 1. Side view of the OSU apparatus showing location of thermopile junctions, wall thermocouple, and oxygen probe used to measure HRR.

4.2. Sensible heat of combustion from apparatus Calibration of the sensible heat of the apparatus was conducted using an electrically heated panel (Watlow Type 5 ceramic fiber panel heater, VF506A06S) having the same dimensions as the test specimen (152 mm
152 mm) and adapted to a standard OSU specimen holder as per the Gardon gauge used to calibrate the heat flux from the silicon carbide heating rods. The sensible enthalpy of the OSU apparatus was measured using a 26-gauge Chromel/Alumel thermocouple (Type K) that was silver-soldered to the outside wall of the OSU, 150 mm below the air manifold and 50 mm in front of the plane of the sample face as shown in Fig. 1 and described in detail elsewhere [18,23]. The portion of the combustion heat that is absorbed by the apparatus as a temperature rise (sensible heat) during calibration and testing was measured by inserting the electrical resistance heating panel into the chamber at the sample location during testing for a range of power levels. Voltage was used to control the heating power of the radiant panel which was held outside

ARTICLE IN PRESS

Temperature (°C)

R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645 500 490 480 470 460 450 440 430 420

635

∆Ta(0) 0

20

40 60 Time (minutes)

80

100

Fig. 2. Temperature history of OSU apparatus wall for a 45 min square wave 1031 W power history generated by an electrically heated panel.

the OSU apparatus for 15 min to reach steady state before being inserted into the holding chamber of the OSU. After 1 min in the holding chamber, the electrically heated panel was inserted into the test chamber of the OSU and allowed to come to thermal equilibrium, which usually required more than 30 min. After equilibration of the heating panel in the OSU test chamber, the power to the panel was turned off and the OSU was allowed to return to baseline, i.e., steady state at zero power input. Fig. 2 shows a typical experiment using 1.031 kW power to the electrically heated panel. Note that DTa in Fig. 2 does not return to zero after the electric power is turned off when the panel assembly remained in the OSU test chamber. A wall temperature increase DTa(0) ¼ 10 1C persisted for nearly 2 h after the heating history was terminated and was accompanied by a small increase in thermopile voltage in the stack. An increase in wall and stack temperatures is a consequence of the increase in radiant energy transfer to the high-surface-area sample/panel assembly from the silicon carbide heating rods, and is normally observed in the standard 5 min test [3] as baseline drift. Fig. 3 shows the temperature history from the time of panel insertion into the sample chamber of the OSU for heater powers 0, 131, 491, and 1031 W. An instantaneous wall temperature increase was observed when the electric panel was inserted even though the thermopile in the exhaust stack showed the typical reduction in voltage normally associated with cool air entering the OSU from the holding chamber during sample insertion. At low and zero electric panel power, the wall temperature of the OSU also decreased due to convective heating of the sample holder/panel assembly by the relatively hot OSU walls. The initial slope of the temperature Ta versus time (t) curve for the four temperature histories in Fig. 3 was computed using a linear least-squares fit of the data from the first minute of the test. Fig. 4 shows the results of these tests plotted as the initial rate of wall temperature rise, i.e., dDTa/dt at t-0, versus heater power. A least-squares linear regression of the data in Fig. 4 gives the following values for the

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

Apparatus Temperature, T (°C)

636

480 1031 Watts

470 460 450

491 Watts

440 131 Watts 0 Watts

430 420 410 400

0

1

2 3 Time (minutes)

4

5

Wall Temperature Derivative (K/sec)

Fig. 3. Apparatus wall temperature history for step changes in electrical heater of power 0, 131, 491, and 1031 W.

0.3 0.2 0.1 0.0 -0.1 -200

0

200

400 600 800 Heater Power (W)

1000

1200

Fig. 4. Initial rate of wall temperature change for step changes in electrical heater of power 0, 131, 491, and 1031 W.

coefficients in Eq. (6): K c DT a ð0Þ ¼ 161 W; K a ¼ 2:92 kJ=K: By comparison, Smith [16] obtained Ka ¼ 3.58 kJ/K for his OSU apparatus. Fig. 5 is a plot of the HRR of the methane burner calculated from the net heat of complete combustion (50.03 MJ/kg), the combustion efficiency under well-ventilated conditions (99% [22]), and the density at standard temperature and pressure (r ¼ 0:647 kg=m3 ) of methane for the volumetric flow rates indicated in the plot, with 1 L/min methane flow rate defined as zero HRR (baseline). The calculated HRRs are 2.1, 3.2, and 4.3 kW for the 4, 6, and 8 L/min methane flow rates. Components of the HRR measured as the sensible heat of the combustion gas/air stream (Eq. (1)) and

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

Methane (Theoretical) Total Sensible Air (Sensible) Apparatus (Sensible)

Heat Release Rate (kW)

6 5

637

8 L/min

4 6 L/min

6 L/min

3 4 L/min

2

4 L/min

1 0 -1 0

5

10 Time (minutes)

15

20

Fig. 5. Sensible HRRs computed from air and apparatus wall temperatures compared to total (air+apparatus) for burner calibrations at 4, 6, and 8 L/min methane flow rates.

that of the apparatus/walls (Eq. (5)) are also shown in Fig. 5. The data in Fig. 5 show that at the beginning of the 2 min HRR pulse, the sensible enthalpy rise of the exhaust gases and apparatus are comparable in magnitude. By the end of the 2 min pulse, about 80% of the HRR of the methane flame is accounted for by the sensible enthalpy of the exhaust stack gases while about 20% remains in the apparatus as an enthalpy change. Fig. 5 shows that the rate of change of sensible enthalpy (temperature) of the exhaust gases and apparatus are opposite in sign during the HRR pulse and of opposite sense afterwards, characteristic of the dynamics of an inertial system. In this case, the temperature rise of the exhaust gases is damped by the thermal inertia of the apparatus. Accounting for these dynamics using Eq. (6) gives a HRR from total sensible enthalpy that closely follows the HRR of the burner over the entire calibration sequence as demonstrated in Fig. 5. 4.3. Chemical heat of combustion from oxygen consumption (OSU) The OSU apparatus was modified to measure oxygen depletion by inserting a thin walled, 6.35 mm outside diameter (OD) stainless steel tube into the inner cone of the pyramidal section as shown in Fig. 1. Combustion gases were drawn into the probe by a TEFLON pump (KNF Labport Model N86KT) at a nominal flow rate of 4 L/min through three 4 mm holes drilled into the tube and oriented downstream (upward in Fig. 1). After exiting the pump, the gas stream was split and a needle valve was used to meter 0.1 L/min of the combustion gas stream into a 6 mm OD TEFLON tubing packed with a small plug of glass wool to remove soot, anhydrous calcium sulfate (Drierite, 20-mesh) to remove water, and sodium hydroxide-coated silica (Ascarite II, Thomas Scientific, 20-mesh) to remove CO2. The scrubbed combustion gas stream then passed through a calibrated oxygen analyzer (Panametrics Series 350, Zirconia) and a flow meter (Rotameter, Matheson Gas

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

638

Heat Release Rate (kW)

6

Total Sensi ble Oxygen Consumption

5

8 L/min

4 6 L/min

6 L/min

3 4 L/min

2

4 L/min

1 0 -1 0

5

10 Time (minutes)

15

20

Fig. 6. Chemical (O2 consumption) and sensible HRRs for OSU burner calibration at 4, 6, and 8 L/min methane flow rates.

Products) and the oxygen concentration was recorded at 1 s intervals by the data acquisition computer. The oxygen consumption system was calibrated simultaneously with the thermopile using the standard 4, 6, 8, 6, 4 L/min methane flow rate sequence to the methane burner [3] as per Fig. 5. The relative standard deviation (RSD) of the HRR measurements by oxygen consumption was 5%. The 90% response time of the oxygen consumption system was 6 s as determined from the step changes in methane flow rate (see Fig. 5). Fig. 6 shows the results of these calibrations graphically compared to the total sensible enthalpy (air+apparatus) calculation. Reasonable agreement between the HRR histories obtained by the two methods is observed for the methane calibrations. 4.4. Chemical heat of combustion from oxygen consumption (cone calorimeter) A commercial fire calorimeter operating on the oxygen consumption principle (Atlas Fire Science Products CONE2 Combustion Analysis System) constructed in accordance with ASTM E 1354 [21] was used for cone calorimeter tests. The standard test procedure was modified by replacing the 57 mm orifice plate with a 29 mm diameter orifice plate to reduce the air flow through the test chamber from the nominal value of 24 L/s rate to 6 L/s. This change increased the transit time of the exhaust gases from the sample location to the analyzer to 19 s but provided sufficient oxygen depletion for accurate HRR measurements of thin and low-heat-release samples. The dynamic response of the cone calorimeter modified for reduced flow was measured by imposing square-wave, 600 W HRR pulses of duration 15, 30, and 60 s at 1 min intervals by quickly inserting and removing the standard ASTM calibration burner attached to a methane tank with a flow meter (rotameter) and high-accuracy needle valve. The transit time for the gases between the burner/sample and the oxygen analyzer in the cone calorimeter was 19 s and this value was subtracted from the elapsed time to synchronize the mass loss and HRR histories for

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

Original data

639

Deconvoluted Data

Oxygen Concentration (% v/v)

20.9 20.8 20.7 20.6 20.5 20.4 20.3 20.2 20.1 0

50

100

150 200 250 Time (seconds)

300

350

400

Fig. 7. Oxygen analyzer response in cone calorimeter to HRR pulses of 15, 30, and 60 s duration. Original data compared to time-deconvoluted data.

all experiments. The apparatus response time to the step change in HRR imposed by inserting the burner was t ¼ 9 s to reach 63% of the equilibrium value. Fig. 7 shows the results of these tests where it is seen that the equilibrium oxygen concentration is not achieved for HRR pulses less than 30 s in duration because the oxygen consumption system requires 21 s to reach 90% of equilibrium, which is significantly longer than the 6 s measured for the oxygen consumption system used with the OSU (see Section 4.3). The longer response time of the oxygen consumption system of the cone calorimeter compared to the OSU results from a combination of effects including mixing (dilution) of combustion gases with air in the sample chamber, duct, and scrubbers [24], and the response time of the oxygen analyzer itself, although the relative magnitude of these effects on the overall instrument response was not determined. Regardless of the cause of signal smearing in the cone calorimeter, time deconvolution of the oxygen consumption history using Eq. (8) with t ¼ 9 effectively compensates for the instrument response and produces the oxygen consumption history shown as the dotted line in Fig. 7 plotted alongside the raw data (solid line). Despite the slight overshoot/undershoot at the beginning/end of the square wave resulting from the simple first-order transfer function (Eq. (8)), deconvolution of the oxygen consumption history captures the burner experiment data better than the original oxygen signal, particularly with respect to the maximum oxygen consumption (HRR) measured during the short duration tests. Cone calorimeter tests of plastic and composite samples were conducted in the horizontal orientation with the spark igniter as per ASTM 1354 [21] but with time deconvolution to correct for instrument response. Cone calorimeter tests of plastic and composite samples were also conducted in the vertical orientation using a

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

640

methane pilot flame comparable in length to that used in the OSU to force ignition [3,4]. A pilot flame was produced at the end of a 3.4 mm OD stainless-steel tube inserted through the center of the conical heater by adjusting the methane flow rate to that of the OSU pilot flame. The pilot flame impinged on the vertical test specimen in the cone calorimeter at the center of the lower edge, 3 mm above the specimen holder. Oxygen consumed by the methane pilot flame in the cone calorimeter registered as a small shift in the baseline.

5. Results 5.1. HRR measurements in the OSU apparatus’

Heat Release Rate (kW/m2)

All results for HRRs of materials are normalized for sample surface areas of 0.0232 and 0.010 m2 for the OSU and cone calorimeter, respectively. Figs. 8 and 9 compare methods of measuring HRR in the OSU apparatus for 1.5-mm-thick samples of transparent BPA PC and BPC PC, respectively. It is clear that the total sensible enthalpy (Eq. (6)) and the oxygen consumption (Eq. (7)) methods are in substantial agreement for these two samples, while the sensible enthalpy of the air stream (Eq. (3)) significantly underpredicts both the peak HRR and total heat release because of the high luminosity (radiant fraction) of the polycarbonate flame relative to the methane flame used for calibration. Dozens of research samples of fiber-reinforced thermoset resins (phenolic, epoxy, cyanate ester, mineral, etc.), honeycomb sandwich panels, and filled/unfilled thermoplastic sheet having a wide range of chemical and physical characteristics have been tested in the OSU over the past several years in our laboratory. Results from 50 individual tests of these materials are plotted in Figs. 10 and 11. The peak sensible HRR during the first 5 min of the test deduced from the temperature history of the exhaust gases as per 14 CFR Part 25 [3] (i.e., Eq. (3)) is plotted versus the peak

Oxygen Consumption

350

Total Sensible Air Sensible (FAA method)

300 250 200 150 100 50 0 0

1

2 3 Time (minutes)

4

5

Fig. 8. BPA PC test in OSU apparatus at 35 kW/m2 external heat flux. HRRs calculated from oxygen consumption, sensible heat of exhaust gases, and total sensible heat (exhaust+apparatus).

ARTICLE IN PRESS

Heat Release Rate (kW/m2)

R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

641

Oxygen Consumption

120

Total Sensible 100

Air Sensible (FAA method)

80 60 40 20 0 0

1

2 3 Time (minutes)

4

5

Fig. 9. BPC PC test in OSU apparatus at 35 kW/m2 external heat flux. HRRs calculated from oxygen consumption, sensible heat of exhaust gases, and total sensible heat (exhaust+apparatus).

500 Exhaust + Apparatus (Total) Exhaust

Sensible HRR (kW/m2)

400

300

200

100

0

0

100

200

300

400

500

Chemical HRR (kW/m2)

Fig. 10. Comparison of peak sensible HRR in OSU apparatus from exhaust gases and total enthalpy (exhaust+apparatus) versus peak chemical HRR measured by oxygen consumption for 50 different materials.

chemical HRR deduced from the oxygen consumption history (Eq. (7)) measured simultaneously during the test (open circles). Also plotted in Fig. 10 versus chemical HRR in the OSU is the sensible heat of the exhaust gases and apparatus as per Eq. (6) (solid circles). The solid line is direct proportionality. It is seen that the sensible heat of the exhaust gases routinely underestimates chemical HRR, but by an amount that depends on the sample being tested [6–17].

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

642

500 Exhaust + Apparatus (Total) Exhaust

Sensible HR (kW-min/m2)

400

300

200

100

0

-100 0

100

200 300 400 Chemical HR (kW-min/m2)

500

Fig. 11. Comparison of HR in OSU apparatus during the first 2 min of test from exhaust gases and total enthalpy (exhaust+apparatus) versus chemical HR by oxygen consumption for 50 different materials.

Fig. 11 compares the sensible HR in the OSU apparatus during the first 2 min of the test to the chemical HR by oxygen consumption over the same period for the 50 samples in Fig. 10. Open circles are data from the sensible enthalpy of the exhaust gases as per 14 CFR Part 25 method (Eq. (3)). Solid circles are data for total sensible heat (Eq. (6)). The dotted line is direct proportionality. Fig. 11 shows that the sum of the total sensible heat of the exhaust gases and the apparatus agrees closely with the chemical HRR from oxygen consumption. 5.2. Comparison of HRR measurements in OSU apparatus and cone calorimeter Samples of fiberglass cloth impregnated with BPA epoxy were tested in the OSU apparatus and cone calorimeter. The hand layup used to fabricate samples resulted in a variation in resin mass, so the fiberglass cloth was weighed before and after impregnation/curing to determine initial resin mass. All samples were wrapped on the rear face and sides with aluminum foil and inserted into the OSU/cone sample holder before recording the total weight of the sample assembly. The specimen holder was weighed again after the tests to determine mass loss and residual mass. The rock wool backing material used in the cone calorimeter tests was dried in an oven at 100 1C overnight to remove adsorbed moisture which would otherwise be a significant fraction of the mass loss in HRR tests of thin samples. Fig. 12 compares HRR histories (normalized for sample area) for BPA epoxy fiberglass lamina in the OSU from oxygen depletion (Eq. (7)), and total sensible

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

OSU-O2 Consumption OSU-Total Sensible Cone-O2 (Raw Data) Cone-O2 (Deconvoluted)

350 Heat Release Rate (kW/m2)

643

300 250 200 150 100 50 0 -50 0

10

20 30 40 Time (seconds)

50

60

Fig. 12. OSU and cone calorimeter data for HRR of BPA epoxy/fiberglass lamina. OSU data for oxygen consumption (O2) and total sensible heat. Cone calorimeter data for by oxygen consumption with and without (raw data) time deconvolution.

enthalpy (Eq. (6)). Also plotted in Fig. 12 are HRR histories from the cone calorimeter measured at an external heat flux of 35 kW/m2 in the horizontal orientation using the spark igniter without time deconvolution (raw data) and with time deconvolution (Eq. (8)). Fig. 12 demonstrates that samples tested in the cone calorimeter exhibit delayed ignition and peak HRR relative to the OSU because of the spark igniter and time deconvolution of the oxygen consumption data to obtain accurate HRR histories for thin samples that burn quickly (o30 s) in the cone calorimeter. The peak HRR of BPA epoxy/fiberglass lamina during the first 5 min of the test measured by oxygen consumption in the OSU apparatus and in the cone calorimeter is plotted versus sample mass loss in Fig. 13. Cone calorimeter measurements were made in the horizontal orientation with spark ignition as per ASTM 1354 [21], and in the vertical orientation with an impinging pilot flame to force ignition. All cone calorimeter data is time deconvoluted using Eq. (8) to correct for instrument response. Fig. 14 is a plot of the time integral of the HRR up to 2 min into the test measured by oxygen consumption in the OSU apparatus and cone calorimeter versus sample mass loss. Cone calorimeter measurements of the 2 min heat release were made in the horizontal orientation with spark ignition as per ASTM 1354, and in the vertical orientation with an impinging pilot flame to force ignition. All cone calorimeter data is time deconvoluted using Eq. (8) to correct for instrument response. Figs. 13 and 14 demonstrate that the peak HRR during the 5 min test and the total HR at 2 min into the test are proportional to the sample mass consumed by burning (above about 1 gm) in both the OSU and cone calorimeter for the thin (0.13 mm) BPA epoxy fiberglass lamina. Sample orientation and mode of ignition (spark, pilot flame) have no significant effect on the results of these tests if the instrument characteristics are properly accounted for.

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

Peak Chemical HRR (kW/m2)

644

600 500 400 300 200 100 0

Cone (H) Spark Ignition Cone (V) Flame Ignition OSU-O2 (V) Flame Ignition

0

1

3 2 Sample Mass Loss (grams)

4

5

Chemical Heat Release at 2-Min (kW-min/m2)

Fig. 13. Peak chemical HRR (O2 consumption) versus sample mass loss for BPA epoxy/glass lamina in OSU and cone calorimeter. Cone calorimeter tests conducted in horizontal orientation (h) with spark ignition and in vertical orientation (v) with pilot flame ignition. Cone calorimeter data is time deconvoluted.

Cone (H) Spark Ignition Cone (V) Flame Ignition OSU-O2 (V) Flame Ignition

80 60 40 20 0

0

1

2 3 Sample Mass Loss (grams)

4

5

Fig. 14. Chemical heat release (O2 consumption) at 2 min into test versus sample weight loss for BPA epoxy/glass lamina in OSU and in cone calorimeter. Cone calorimeter tests conducted in horizontal orientation with spark ignition and in vertical orientation with pilot flame ignition. Cone calorimeter data is time deconvoluted.

6. Conclusions Accurate HRR histories for thin (0.13–1.5 mm) specimens of plastics and composites or those with low HRR in flaming combustion can be measured in the OSU apparatus or the cone calorimeter with a proper accounting of heat transfer, ignition conditions, test dynamics, and instrument response. The measurement of the total sensible heat in the OSU apparatus is shown to be a convenient thermal method for obtaining accurate HRR histories of materials and products using a standard 14 CFR Part 25.853 HRR apparatus and procedure with a modified analysis. References [1] Janssens M. Calorimetry. In: SFPE handbook of fire protection engineering, section 3, 3rd ed. Quincy, MA:National Fire Protection Association; 2002. p. 38–62.

ARTICLE IN PRESS R. Filipczak et al. / Fire Safety Journal 40 (2005) 628–645

645

[2] Babrauskas V. From bunsen burner to heat release rate calorimeter. In: Babrauskas V, Grayson SJ, editors. Heat release in fires. New York: Elsevier; 1992. p. 7–29 [chapter 2]. [3] Heat release rate test for cabin materials. In: Aircraft material fire test handbook, section 5. Final report DOT/FAA/AR-00/12. Federal Aviation Administration; April 2000 [chapter 5]. [4] ASTM E 906-98. Standard test method for heat and visible smoke release rates for materials and products (Configuration A). West Conshohocken, PA: American Society for Testing of Materials; 1998. [5] Improved flammability standards for materials used in the interiors of transport category airplane cabins. Federal Register, 53(165), 32,564; 25 August 1988. [6] Hill RG, Fitzgerald LT. Interlaboratory comparison of heat release data from aircraft panels. DOT/ FAA/CT-TN86/3; March 1986. [7] Sarkos CP et al. Implementation of heat release measurements as a regulatory requirement for commercial aircraft materials. In: 50th calorimetry conference, National Institute of Standards and Technology, 23–28 July 1995, Gaithersburg, MD. [8] Lyon RE, Abramowitz A. Effect of instrument response time on heat release measurements. Fire Mater 1995;19:11–7. [9] Tsuchiya Y. Heat release rate measurement for evaluating the flammability of aircraft materials. In: Proceedings of the symposium on aircraft fire safety, Advisory Group for Aerospace Research and Development, North Atlantic Treaty Organization, 21–26 May 1989. p. 32. [10] Krause RF, Gann RG. Rate of heat release measurements using oxygen consumption. J Fire Flammability 1980;11:117–30. [11] Babrauskas V. Comparative rates of heat release from five different types of test apparatuses. J Fire Sci 1986;4(2):148–59. [12] Babrauskas V. Comparative heat release rates for aircraft materials measured in different apparatuses. In: Babrauskas V, Grayson SJ, editors. Heat release in fires. New York: Elsevier; 1992. p. 583–90 [chapter 17(c)]. [13] Tran HC. Modifications to an Ohio State University apparatus and comparison with cone calorimeter results. In: AIAA/ASME proceedings, Seattle, WA; 1990. [14] Babrauskas V. Performance of the OSU rate of heat release apparatus using PMMA and gaseous fuels. Fire Safety J 1982;5:9. [15] Evans DD, Breden LH. Time delay correction for heat release rate data. Fire Technol 1978;14:85–96. [16] Tsuchiya Y, Mathieu JF. Measuring degrees of combustibility using an OSU apparatus and oxygendepletion principle. Fire Safety J 1991;17:291–9. [17] Ostman AL, Svensson IG. Comparison of three test methods for measuring rate of heat release. Fire Mater 1985;9(4):176–84. [18] Smith EE. Heat release rate calorimetry. Fire Technol 1996;32(4):333–47. [19] Moussan F, Delfau JL, Vovelle C, Cang CPV, Bosseboeuf G. Heat release rate and effective heat of combustion measurements: a comparative study of thermal and oxygen consumption techniques. J Fire Sci 1995;13:482–99. [20] Babrauskas V. The cone calorimeter. In: SFPE handbook of fire protection engineering, Section 3. 3rd ed. Quincy, MA: National Fire Protection Association; 2002 p. 63–81. [21] ASTM E 1354-97. Standard test method for heat and visible smoke release rates for materials and products using an oxygen consumption calorimeter. West Conshohocken, PA: American Society for Testing of Materials; 1997. [22] Tewarson A. Generation of heat and chemical compounds in fires. In: SFPE handbook of fire protection engineering, Section 3. 3rd ed. Quincy, MA: National Fire Protection Association; 2002 p. 82–161. [23] Filipczak R, Lyon RE. The correlation of heat release calorimetry measurements. Technical note DOT/FAA/AR-TN02/104, Federal Aviation Administration, November 2002. [24] Lyon RE, Blake D. Heat release rate measurements of objects burning in compartments. In: Proceedings of the 10th international fire science & engineering conference (INTERFLAM 2004). London: Interscience Communications; 2004. p. 509–14. [25] Salvadori MG. Numerical methods in engineering. Englewood Cliffs, NJ: Prentice-Hall; 1961 [chapter 2, p. 85–7].