Large-scale upward flame spread tests on wood products

This publication is a documentation of the results of experiments carried out to investigate upward flame spread on walls when ignited by a propane diffusion burner at the base of the wall. Experiments were made using 1.2 m x 2.4 m or 1.2 m x 7.5 m specimens. The product to be studied was mounted on a vertical timber-framed sample holder. The backing of the board was either a 50 mm thick mineral wool or a 10 mm thick calcium silicate board. The assembly was located under a 3 m x 3 m gas collection hood. The heat release rate of the ignition source was in the range 70 kW - 100 kW. The measurements comprised temperatures in the gas phase close to the surface, surface temperatures on the board, and the surface temperatures behind the board. Heat flux gauges were mounted flush to the surface at various heights along the centre line of the specimen. Heat flux emitted out of the wall was also measured with a similar gauge. Close to the top of the specimen a rake of bi-directional probes was used to measure the flow velocity distribution in the wall plume. At the inlet of the exhaust duct, the temperature was measured to enable the evaluation of the connective heat output. Farther in the duct the concentration of CO was measured for the purpose of estimating the CO yield in the fire. Standard instrumentation was also used to measure the total heat release rate by oxygen consumption. The original test results are given as tables in the appendix to this publication. They also available in electronic form. The test have also been recorded on video. Photo-graphs of the tests are shown as an appendix to this publication. In the tests with a non-combustible wall, the flame height of the gas burner against the wall was represented slightly better with by a linear function of the hear release rate than by a function with a 2/3-power dependence on heat release rate. The tests showed that the upward spread on a vertical combustible surface depends not only on the product but also on the thermal properties of the substrate. For particle board and wood panels on an insulating substrate the flame spreads upwards, retreats and then spreads upward again. In a high space, where there is no thermal feedback from the upper smoke layer, the flame spread becomes very slow high above the initial ignition source. The fire will probably die out on its own, if the initial ignition source is removed or burns out. It is therefore quite possible that wood products in high rooms do not create as large a hazard as they do in smaller rooms with the assumption that the contents of the building do not provide too big an ignition source. The results of the tests can be used to develop or verify upward flame spread models. One example of a numerical model flame spread models has been presented in a recent paper by the authors (Kokkala, M., Baroudi, D., and Parker, W.J., Upward flame spread on wooden surface: products: experiments and numerical modelling, 5th International Symposium on Fire Safety Science, 3 - 7 March 1997, Melbourne, Australia).