Development of an artificial neural network model for the prediction of hydrocarbon density at high-pressure, high-temperature conditions

In this study, a new approach for the prediction of density of pure hydrocarbons such as n-pentane, n-octane, n-decane, and toluene has been suggested. The available experimental data in the literature have been selected at high pressure (∼500 MPa) and high temperature (∼400 °C) conditions. The data are analyzed accurately using artificial neural networks and have been compared with different results obtained by various EOSs such as, PC-SAFT, SAFT, Peng–Robinson and SRK equations. The values of “Average Absolute Deviation Percent” for the densities of each material are calculated using artificial neural networks. These are 0.2 for n-pentane, 0.11 for n-octane, 0.66 for n-decane and 0.51 for toluene, which are substantially more accurate than those obtained with various EOSs. Finally, it has been shown that artificial neural network as an applicable and feasible instrument can be proposed to predict the density data for such materials with high accuracy.

► Densities of four pure hydrocarbons were gathered at high pressure and temperature.
► An artificial neural network model was designed to predict hydrocarbon densities.
► The model results were compared with different results obtained by various EOSs.
► The accuracy of the developed model was better than applied EOSs.