(53w) Prediction of Fire and Explosion Properties from Molecular Structure Using Quantitative Structure?Property Relationship (QSPR) Models

Experimental measurements involved in determining fire and explosion parameters are known to be risky. In order to provide another alternative, the use of a quantitative structure-property relationship model where physicochemical properties of substances can be mathematically expressed, affords an accurate modeled response of chemical structures. Recently, significant research has been done by OSU research group to study fundamental properties of fire and explosion, such as the minimum ignition energy and flammability limits, both for pure chemicals and mixtures. For example, it was reported in the journal, Industrial & Engineering Chemistry Research the development of two quantitative structure-property relationship models involving multiple linear regression analysis and support vector machine to determine minimum ignition energy values of chemical fuels. With the involvement of big data, more predictive models can be developed through statistical analysis or machine learning process. Depending on the availability of experimental data, these predictive analytics have shown strong external predictive ability to provide fire and explosion fundamental data in the process safety field.