An Intelligent Approach to Estimate Pressure-Volume-Temperature Properties in the System of Methane- Tetrafluoromethane: Densities and Compressibility Factors

In this study, the ability of Artificial Neural Network or ANN based on back-propagation approach for predicting the densities and compressibility factor of gaseous binary mixtures of CH4-CF4 has been investigated. Some experimental data (1507 data points) of gas densities for pure CH4, pure CF4, and three mixtures (0.25, 0.50, and 0.75 mole fraction of methane) are used to find optimal network, for which a density range from 0.75 to 12.5 mole/lit were covered. Finally, a network included 10-5-1 neurons in its layer is selected. By using this number of neurons, admissible absolute average deviations (about 0.112593% and 0.121046% for training and testing steps, respectively) are provided. Then, a comparison of compressibility factors for a mixture containing 50% CH4 shows an acceptable deviation, about 0.023604%. These results show that there is an excellent agreement between experimental data and ANN predictions.

Keywords:

Artificial Neural Network , Gas Mixture Density , Compressibility factor , CH4 , CF4