New predictive methods for estimating the vaporization enthalpies of hydrocarbons and petroleum fractions

Various models and correlations are available that can predict the vaporization enthalpy of hydrocarbons. The available methods normally have lower accuracy for predicting the vaporization enthalpy of heavy hydrocarbons and require further verification, because, during the development of the original predictive methods, experimental data describing the vaporization enthalpy of heavy hydrocarbons and petroleum fractions were not available. In this communication, after a quick review of the existing correlations reported in the literature, an empirical correlation is first proposed, which is capable of predicting the vaporization enthalpy of hydrocarbons, especially heavy hydrocarbons and petroleum fractions, from the specific gravity values and the normal boiling point temperatures. The capability of artificial neural networks (ANNs), as an alternative method, to predict the vaporization enthalpies of hydrocarbons and petroleum fractions is then demonstrated. Among the various neural networks reported in the literature, the feed-forward neural network method with a modified Levenberg-Marquardt algorithm is used. The method is trained using recent experimental data, especially for heavy hydrocarbons and petroleum fractions. Independent experimental data are used to validate and examine the reliability of this method. The results are also compared with the predictions of other predictive techniques. The predictions of this method are shown to be in better agreement with the experimental data reported in the literature, demonstrating the reliability of the ANN used in this work.