Fire development in rooms partially lined with timber

Abstract

Combustible internal linings have been identified as a major contributor in several high-profile fire events. Legislators have sought to address the risk posed by combustible linings by classifying and restricting their use in some spaces. This study compares the standard tests and classification methods for internal timber linings in the U.S.A, Canada, U.K., Europe, Australia and New Zealand. Building legislation such as the New Zealand Building Code generally assumes that all wall and ceiling surfaces in a space are lined with the products of similar reaction to fire performance and this reduces the choices available to the designer and often excludes partial combustible linings, apart from some minor concessions. To examine the justification for this assumption a series of seven room experiments are conducted in the conditions required by ISO 9705 with different configurations of 7 mm thick plywood linings on the walls and ceiling and the remainder of the enclosure lined with non-combustible calcium silicate board. The heat release rate, gas temperatures and flame spread rates were measured. The relative performance of the lining configurations is assessed by either ranking the time-to- flashover and peak rate of heat release or using the FIGRARC methodology. Flashover was achieved in 4 tests. FIGRARC was found to be a better means of comparing the performance of the seven configurations as it more accurately and intuitively represented the rate of fire development than time to flashover or peak heat release rate measurements. The feasibility of using a modified version of the B-RISK zone model to simulate fire growth in the same enclosures is examined. The rate of heat release and time to flashover from the experiments are compared with simulations using a modified version of B-RISK. The B-RISK flame spread algorithms have been altered to allow for modelling of ceilings and walls which are only partially lined with combustible linings. Model input data for the plywood material, including the ignition temperature and flux-time-product, as well as the heat release rates of the plywood at various incident heat fluxes are characterised using data measured in fifteen ISO 5660-1 cone calorimeter experiments. The modelled fires show good agreement with early fire growth in the experiments, however, the B-RISK models tended to overestimate the upper layer temperature and depth, leading to early flashover compared to the experiments. The model is shown to be largely insensitive to the minimum temperature for flame spread and the flame spread parameter. Predicted lateral flame spread was strongly influenced by the wall surface temperature and entrainment into the upper layer which was found to be underestimated. Further work is recommended to refine the entrainment rates for burning surfaces to improve agreement between the simulated and actual fire development, and investigate the performance of the model in simulating fire growth in larger spaces with partial combustible linings.