XGBoost and Physical-Informed Neural Networks as Surrogate Models for VLE and LLE in PC-SAFT

Phase equilibrium calculations are crucial in chemical engineering design and optimization processes. The PC-SAFT equation of state (EoS) can precisely calculate phase equilibrium, but is relatively complex and computationally intensive. Surrogate models are mathematically simple models that map or regress the input-output relationships of more complex, computationally demanding models. This work employs XGBoost and a hybrid XGBoost and physical-informed neural networks (XGBoost-PINN) as surrogate models to replace PC-SAFT EoS calculations for the vapor-liquid equilibrium (VLE) and liquid-liquid equilibrium (LLE) of binary associating systems. This work investigates the VLE and LLE of water+ethanol and water+1-butanol systems using data generated by the PC-SAFT EoS. The surrogate models take temperature, pressure, liquid phase mole fractions, and the equilibrium type (VLE or LLE) as inputs, and predict the mole fraction in the other phase. Both surrogate models exhibit good accuracy. However, the surrogate model based on XGBoost-PINN demonstrates superior performance for the LLE data compared to one that employs XGBoost.