Prediction of smoke production and heat release by convolution model

Using information obtained from bench-scale tests, a simple convolution integral model is applied for predicting heat release rate and smoke production rate of large-scale test fires of surface materials. The model is based on a convolution product of time derivative of effective area with the corresponding Cone Calorimeter data per unit area. In the convolution model of Wickström and Göransson the model is applied only for the prediction of the heat release rate. The analytical expressions for the effective areas were developed from the visual observations of the burning area of the material in the large-scale test. In this report a trial was made to extend the convolution model also to the prediction of the smoke production rate. Because determination of effective areas cannot in this case be based on the observations, they have to be solved by numerical inverse methods. For the inverse solution of the apriori unknown effective areas, several techniques were developed. Mathematical method and a computer program that uses regularized Singular Value Decomposition was written. Fast Fourier transform technique and simple average solution method were successfully applied as alternative methods. The smoke production rate and heat release rate in the Full Scale Room Corner Test ISO 9705 and in the Cone Calorimeter Test (ISO 9705) were analysed for 9 materials of the EUREFIC set of building products. Effective areas both for the prediction of heat release rate and smoke production rate were calculated. A correlation was found between the areas of heat release and smoke production but no general area functions for the smoke production rate could be found. Therefore, a smoke prediction computer program could not be written based on the model. Determination of general area functions would need more comprehensive studies and a larger set of test data to be analysed. For future projects, the tools developed in this study give a good starting point.