Smoke production in fires

Characterization of smoke, factors influencing smoke production and experimental methods for measuring smoke production are discussed in this literature review. Recent test-based correlation models are also discussed. Despite the large number of laboratories using different fire testing methods, published smoke data have been scarce. Most technical literature on smoke production from building materials is about experimental results in small scale tests. Compilations from cone calorimeter tests have been published for a few materials, e.g. upholstered furniture materials and some building products. Mass optical density data and compilations of gravimetric soot data are available for various materials as well as a number of smoke obscuration values. For a given material often a wide range of values of smoke output can be found in the literature and care should be excercised in applying the appropriate value in each case. Little information is available in scientific literature concerning modelling or predictive correlations between small scale smoke data and real fires or large scale test results. The scale dependence of the rate of smoke production is generally not well known. In laboratory experiments, the production of smoke and its optical properties are often measured simultaneously with other fire properties as heat release and flame spread. The measurements are usually dynamic in full scale, i.e. they are performed in a flow-through system. In small scale they may be either dynamic, as in the cone calorimeter, or static, i.e. the smoke is accumulated in a closed box. Small-scale tests are necessary as practical tools. Full-scale tests are generally considered to be more reliable and are needed to validitate the small-scale tests. Quantitative prediction of the rate of smoke production using basic principles for commercially available construction products is not currently possible. Test-based correlation studies are therefore needed to enable the use of small scale test data in determining the fire hazard of products. There seems to be some evidence that the early stages of certain full scale scenarios can be predicted, whereas post-flashover smoke production cannot yet be resolved. Flow-through systems have been suggested as one means of obtaining correlations between small scale and large scale data