Vapour–liquid equilibrium prediction for synthesis gas conversion using artificial neural networks

Vapour–liquid equilibrium (VLE) modeling is of paramount importance since it affects the efficiency of downstream processing during product recovery in the Fischer–Tropsch synthesis. multi-layer perceptron neural network (MLPNN) was used to simultaneously model VLE of 1533 gas-liquid solubilities divided over sixty binary systems at pressures up to 5.5 MPa and temperatures from 293 to 553 K using literature data. The network was trained using the Levenberg–Marquardt algorithm in MATLAB® for developing and optimizing the model while Bayesian regularization was used to improve the performance of the network. Results obtained from the network suggest that the MLPNN has a better capability in estimating VLE when compared to conventional thermodynamic models.