New models for predicting the flash point of mixtures containing different alcohols

Abstract

In the present study, new empirical models have been proposed to predict the flash point of mixtures containing different alcohols. The models predict the flash point of mixtures as a function of molar composition and the flash point of pure compounds composing the mixture. The models have been developed by the generalization of the available correlations for the prediction of the flash point of petroleum blends using the Levenberg–Marquardt algorithm. 633 experimental data on the flash points of 61 different binary mixtures containing alcohols along with different compounds such as alkanes, alcohols, ketones and esters was used for the model development. The average absolute relative deviation (AARD) of the optimized models over all binary experimental data is 1.29% and 1.22% for the first and second developed models, respectively. The capability of the developed models has been examined by predicting the flash point of 4 different ternary mixtures containing 221 experimental data points.