Repeat fire tests of upholstered furniture: Influence of experimental conditions on test-to-test variability

Abstract

Experiments were conducted to illustrate the influence of experimental conditions on the statistical variation observed for furniture-scale calorimetry. Commercially available upholstered chairs were independently assessed using different ignition techniques, both with and without the presence of a wall corner. For each experiment, heat release rate (HRR), carbon monoxide/carbon dioxide yields, and heat fluxes from the fuel package were compared. Measurements made were used to link the burning behaviour to physical occurrences and to contextualize the variability between trials. The time resolved HRR and emission yields were found to be largely unaffected by the presence of a wall corner, while the radiant heat flux from the item was increased. Both the HRR and CO yield in time showed a dependence on ignition location. These differences were, however, contextualized through the use of key events that drive the burning process and global burning regimes. Further statistical analysis is conducted on specific metrics (e.g., peak HRR, total heat release, average species yields) to illustrate the similarity between these global metrics across each configuration. Statistical uncertainty is then quantified as a function of trials, providing a means to determine the gain in statistical confidence with increasing trials.