Thermodynamics-Informed Neural Network (TINN) for Phase Equilibrium Calculations Considering Capillary Pressure

The thermodynamic properties of fluid mixtures play a crucial role in designing physically meaningful models and robust algorithms for simulating multi-component multi-phase flow subsurface, which is needed many subsurface applications. In this context, the equation-of-state-based flash calculation used to predict equilibrium each phase given mixture going through splitting component, often bottleneck, simulations. paper, capillarity-wise Thermodynamics-Informed Neural Network developed first time propose fast, accurate approach calculating unconventional reservoirs. trained model performs well both stability tests calculations large range reservoir conditions, enables further simulations with strong basis.