Fuel reformulation has been seeded by the consciousness of the potential damages their combustions are causing. Removing one or several components or adding molecules, to reduce pollutants emissions for instance and therefore improve air quality, can modify many properties including safety parameters such as flash points. The flash point of a liquid is the lowest temperature at which a vapor phase at equilibrium with the liquid, is flammable, once mixed with an oxidizer (generally air at atmospheric pressure). For lots of fuel properties, the surrogate fuel blends approach is very often preferred because fuels are generally very complex mixtures containing up to hundreds of components. Predictive methods of flash points for such multicomponent systems are rather scarce in the literature.
In this work, we propose a general approach for accurately predicting the flash points of any kind of fuels or mixtures (pure compounds, solvent mixtures, multicomponent fossile and bio-fuels, chemicals, ...). The predictive method is based on a combination of Catoire-Naudet model [1] with the fully predictive COSMO-SAC dsp activity coefficient model [2]. The main advantage of this method is that experimental flash points of the pure substances are not required, contrary to most other predictive methods for flash point mixtures. Excellent predictions of the flash points of mixtures are obtained for a broad variety of systems. The new model predicts that a small amount of alcohol (methanol, ethanol) in biodiesels can dramatically decrease the flash point of the fuel, in agreement with experimental observations.
