Cone Calorimeter Analysis of FRT Intumescent and Untreated Foam Core Particleboards

The effectiveness of treatments of the surface layer of novel foam core particleboards were evaluated by means of Cone calorimeter tests. Foam core particleboards with variations of surface layer treatment, adhesives and surface layer thicknesses under similar processing conditions were used to produce the test specimen for the Cone calorimeter tests. Ignitability, heat release rate profile, peak of heat release rate, total heat released, effective heat of combustion, mass loss rate, gaseous emissions and specific extinction area were measured using the cone irradiance of 50 kW/m2.Additional analysis of this data provided fuel composition information that could reveal the pyrolysis events of the composite boards. Thermocouples at various depths were used to provide further verification of pyrolysis events. The unprotected foam core panels generally had much higher heat release rates, somewhat higher heat of combustion and much higher smoke production due to the EPS-foam component of tested panels, whereas time to ignition and total heat release were not pronounced from the veneer treated boards. Adding the commercial FRT veneer to the face particleboard provided a dramatic improvement to the measured flammability properties. It worked sufficiently well with a 3 mm thick surface layer to improve the predicted flame spread rating of the foam core particleboards. INTRODUCTION A novel technology to produce sandwich-type composites with wood based facing with a foam core in one single production step has been published [1]. This type of 19 mm thick lightweight foam core panels can be produced on standard particleboard production lines which can be adapted to the new technology some modifications of the machines. The presence of the Expandable Polystyrene (EPS) for in-situ foaming of the core material implies some restrictions in the production process. But also the fire safety of this type of innovative panels might become a crucial aspect when introducing these novel panels into the market. The cone calorimeter for evaluating flammability has gained very wide acceptance worldwide and has been considered to be especially useful for the development of new products [2, 3]. This ASTM E 1354-11a test apparatus measures the relevant reaction-to-fire parameters that have good correlations to full-scale fire behavior. The ignition time, heat release rate, total heat released, heat of combustion, mass loss rate, combustion products and specific extinction area are the main parameters measured and analyzed in this study. The need for a comprehensive investigation of fire performance of foam core sandwich panels which is indicated by the limited studies available on similar thin foam core sandwich panels. The first study in this project involved the cone calorimeter tests of samples exposed in the horizontal orientation with the conical radiant electric heater set at the irradiance 35 kW/m2. By testing 19 mm-thick panels with variations in surface layer thicknesses, core foam densities, and processing temperatures, it was found that the surface layers have an important impact on the fire behavior of sandwich structures [4]. In that study, the heat release rates (HRR) for the sandwich panels were much higher than for the conventional particleboard panel. Their flammability properties improved as the surface thicknesses increased from 3 to 5 mm. However, the levels of HRR were similar to some existing wood-based panels, and thus should have at least some market use on that basis. It is interesting that the EPS foam has thermal properties that suggest a fire retardant solution. It is stated that the polystyrene foams start to soften and shrink from 100 ̊C and melt at even higher temperatures (around 250 ̊C). Upon further heating, ignitable decomposition gases are created at about 350 ̊C. Without a flame source, temperatures above 450 to 500 ̊C lead to the ignition of the decomposition products. When exposed to a small flame, the flame retarded XPS melts away from the ignition source without itself igniting and ignition might only be observed after longer flame exposures. If the contact with the external flame stops, further burning or smoldering might not be observed. In conjunction with other combustible substances, even flame retarded polystyrene foam can burn (www.exiba.org/Properties_of_XPS.asp). Thus to avoid this burning condition the polystyrene can be kept below its decomposition temperatures via the insulation effects of either a thicker surface layer or the use of surface intumescent veneer or coating. The testing of the commercial intumescent surface layer with a high fire rating required the use of the more severe cone irradiance of 50 kW/m2, which is associated with large fires and severe reaction to fire tests. This paper reports on the in-depth study to verify this added fire retardancy mechanism. In addition to the standard flammability measures discussed in ASTM E1354, this study also utilized imbedded thermocouples at various depths in the sandwiched panels and advanced evolved gas analysis to reveal the decomposition behavior of sandwich panels with and without intumescent veneer coating. The construction of three sandwich panels with varying surface layers and the enhancement to the cone calorimeter gas analysis are described in the material and methods section. In the results and analysis section each relevant flammability feature is explained for the three sandwich panels for the exposure to irradiance at 50 kW/m2 and piloted ignition. Also from this data set, the flame spread index classifications (ASTM E84) were estimated. MATERIAL AND METHODS Three Variations for Surface Layers of Foam Core Particle Boards Basically, the foam core particleboards with a nominal thickness of 19 mm were manufactured from a three layered mat without additional gluing between the face and core layers. The resinated wood particles and urea formaldehyde resin (Kaurit 350, BASF, Germany) was used for the face layers. The expandable polystyrene (EPS, Terrapor 4, Sunpor, Austria) with a granule size of 0.3 to 0.8 mm were used as the core materials. According to the data sheet of Terrapor 4, it contains a small amount of flame retardant. Babrauskas and Parker [5] mentioned that fire retardant in foams work for very low ignition flux (<25 kW/m2) but fire performance is essentially unchanged when larger ignition sources are used. This material also contains 5.7 % pentane (by weight) as the blowing agent. Our unpublished study showed that between 2 and 3 % of the initial pentane remains in the foam cells after expansion, depending on process parameters (press temperature etc.). The three-layered mat was then pressed in a lab-scale single opening (Siempelkamp, Germany) hot-press. The press cycle consist of three consecutive stages: pressing phase, foaming phase, and stabilization phase by the internal cooling of the press plates. The temperature of the press plates was set at 130 ̊C. The target overall density was 320 kg/m3 with a face density of 750 kg/m3 and a core density of 124 kg/m3. Nominal surface thickness was 3 mm which corresponds to the foam core thickness of 13 mm and overall thickness of 19 mm. Shalbafan et al. 2012a has described in details the pressing schedules and foaming conditions. The two improvements utilized for this study were the use of conventional beech veneer without and with intumescent paper underneath of the veneer. The fire resistive adhesive used for veneering the samples was Firobond Ultra Adhesive (FUA) supplied from ENVIROGRAF, UK. The sandwich panels without any veneer were utilized as reference samples in this series of tests. At least two panels of each series were produced as replicates and one sample was cut out from each panel to do the fire performance test. All the samples were conditioned at 23 °C and 50 % relative humidity for at least two weeks prior to testing to meet equilibrium moisture content (EMC). Cone calorimeter upgrades and test procedure The tests were carried out according to the ASTM E1354 test method with a cone calorimeter apparatus (Atlas Electrical Devices, Chicago, IL) at the Forest Product Laboratory in Madison, USA. Samples were exposed in the horizontal orientation to the irradiance 50 kW/m2 upon opening the water-cooled thermal shutter and using an electric spark for piloted ignition. Prior to placing the specimen in the sample holder, four thermocouples were attached in the following manner. The exposed surface thermocouple (36 gauge Type K wire) was inserted into a slanted surface crevice formed with a razor blade. Two thermocouples (30 gauge Type K wire) were inserted in tiny long holes at the interface of the foam and particle board, with the bead situated at the sample’s middle. The fourth thermocouple was taped to the backside surface at the sample’s middle. These thermocouple measurements provided data to verify the insulating enhancements of the veneers. The Figure 1 shows the position of the inserted thermocouples in the cross section of the samples. Fig 1 The position of the thermocouples inserted in different places of the samples The specimens were tested in the optional retainer frame with a wire grid over the test specimen. As explained earlier, some of the pentane remained in the specimen. After ignition of the surface layer, the elevated temperature eventually reaches the foam core layer. This temperature stimulates the remaining pentane in the foam to cause the slight expansion of the foam during the test. To overcome excessive spalling and foam expansion that results in direct contact with the cone heater, a surface wire grid was used in all the cone tests to restrain the heated surface. Ignitability was determined by observing the time for sustained ignition of the specimen with a 4 seconds criteria for sustained ignition. Exhaust gas composition was determined using three gas analyzers from Sable Systems (www.sablesys.com) and a relative humidity sensor from U.P.S.I. (www.upsi.fr). Oxygen was measured using the PA-10, a paramagnetic analyzer capable of resolution to 0.0001 %O2 and modified to provide even faster response by reducing internal volume of the filters. Exhaust gas to the sensor was dried using the Sable ND-2, a permeable-membrane dryer. Carbon dioxide was measured using the CA-10, a dual wavelength infra-red sensor capable of resolution to 1 ppm. The same technology was used in the CM-10A for Carbon monoxide detection. Gas was delivered to the analyzers using two pumps. The first larger pump pulls exhaust quickly to the location of the Sable equipment through a pre-filter and water-bath controlled (50 °C) water-to-air heat exchanger to provide consistent incoming air conditions. Then a sub-sample pumps pulls exhaust smoothly through the dryer and analyzers. The relative humidity was measured using the F-TUTA.34R, a quick responding sensor placed very early in the gas sample path inside the cone calorimeter. The lines and sample location were heated with heat tape to near 50 C to avoid condensation on the lines after the ring sampler. The F-TUTA.34R provides analog signals corresponding to relative humidity and temperature. Similarly the Sable components provide analog signals, including the barometric pressure. These signals along with the type K thermocouple readings at various locations in the specimen were captured by the data acquisition system (Measurement Computing USB-1616HS) at 4 Hz. Raw signals were then time-shifted based on time-offlight to the sensor to have all changes correspond to the mass loss signal from the cone calorimeter. Exhaust flow rate calculations were based on Bernoulli’s formula using pressure drop across the orifice, temperature of the exhaust, and various gas concentrations. Further fine tuning of the exhaust flow rate is based on matching the computed mass flow rates of depleted oxygen, carbon dioxide, and water with that determined from nearly complete combustion of pure ethylene glycol, whose fuel mass flow is measured with the weigh scale. As a basis for comparison, we have that for any incomplete hot combustion, the dynamic mass flow rate (g/s) of a fuel mixture with empirical formula CXHYOZNUSV has six equivalent calculations as derived from simple mass balances as [6], 6 5 & 4 70 38 14 16 12 14 38 14 16 12 3 12 9 9 38 14 16 12 2 12 44