Practical application of machine learning on fast phase equilibrium calculations in compositional reservoir simulations

•Multi-layer ANN models are developed for phase equilibrium calculations to improve training efficiency and accuracy.•Compared with the SVM and RVM method, the proposed ANN model for stability tests require much fewer training data.•The ANN model for phase splitting calculation is able to provide more reliable initial guesses.•The ANN models are successfully applied on the practical compositional simulations.