Numerical Simulation and Holdup Prediction of Oil-Water Two-Phase in Low Yield Shale Oil Near Horizontal Wells

The production of shale oil wells is typically low, and logging instruments such as the flow scanning imager (FSI) may exhibit poor turbine response in low-production wells, leading to measurement deviations and reduced interpretation accuracy. To investigate the oil-water two-phase flow behavior in low-production near-horizontal wells, 11 downhole pipeline models with varying well deviations (inclination angles), an inner diameter of 0.124 m, and a length of 6 m were established using ANSYS Fluent. The built-in volume of fluid (VOF) model is employed to simulate oil-water two-phase flow in wells with different inclinations, yielding flow patterns, water holdup data, and holdup imaging under varying well deviations, flow rates, and water-cut conditions. A water holdup database is constructed based on simulation results, and water holdup variation charts were generated. Numerical simulations are analyzed to derive the relationship between water holdup and well deviation under different parameter conditions, resulting in a functional expression for water holdup versus inclination. A multilayer perceptron (MLP) neural network is trained using the simulation data to predict the nonlinear relationships among well deviation, flow rate, water-cut, and water holdup. This approach achieved accurate water holdup predictions for near-horizontal well oil-water two-phase flow under diverse conditions. Comparative analysis with experimental data demonstrated high prediction accuracy, with a relative error of 3.222%, meeting the validation requirements for downhole instrument water holdup measurements.