Prediction of vapor–liquid equilibrium data for ternary systems using artificial neural networks

Most solvents used in the semiconductor industry are toxic and costly. Thus, the solvents should be recovered for re-use in these processes by distillation methods, and vapor–liquid equilibrium data are necessary for the design and operation of distillation columns. These data can be estimated using activity coefficient models. In this work, artificial neural networks were applied to predict and estimate vapor–liquid equilibrium data for ternary systems saturated with salt. The databases taken from literature were split into training, validating and testing data and the best architecture was an 8–6–7–4 network. The absolute mean errors for the whole database were 0.0166, 0.0177, 0.0151 for the vapor mole fraction of components (y1, y2, y3) and 0.74 °C for the bubble point temperature. The artificial neural network predictions showed better agreement with experimental data than the thermodynamic model predictions.