Fast and Robust Prediction of Multiphase Flow in Complex Fractured Reservoir Using a Fourier Neural Operator

Predicting multiphase flow in complex fractured reservoirs is essential for developing unconventional resources, such as shale gas and oil. Traditional numerical methods are computationally expensive, deep learning methods, an alternative approach, have become increasingly popular topic. Fourier neural operator (FNO) networks been shown to be a hundred times faster than convolutional (CNNs) predicting conventional reservoirs. However, there few relevant studies on applying FNO predict with fractures. In the present study, FNO-net U-net (CNN-based) were successfully applied pressure saturation fields 2D heterogeneous The tested results show that can accurately depict influence of fine fractures, while CNN-based method has relatively poor performance treatment fracture systems, both terms accuracy computational speed. addition, by adding initial conditions boundary loss function FNO, we prove necessity physical constraints data-driven model. This work contributes improving understanding applicability FNO-net, provides new insights into