Unconventional oil and gas production has made a significant contribution in the past decade, yet many of these wells are not managed to their fullest potential. There is a significant opportunity to optimize well performance through continuous estimation and tracking of well performance for large-scale operations.
However, understanding and predicting well performance in unconventional reservoirs poses a significant challenge due to the complexity of capturing the relevant physics of multi-stage fractured horizontal wells (MFHWs). Traditional mechanistic or numerical models are not suitable for field-scale applications, as they may require information that is not easily available, are interpretive, need arduous manual efforts, have long runtimes, or produce results with high uncertainty.
In recent years, hybrid models have gained popularity as a solution to these challenges. These models combine physics-informed data-driven methods to accurately model transient well performance with low input requirements, fast convergence, and high accuracy. They enable fast decision making compared to pure numerical simulation, while reducing overfitting compared to pure data driven solution.
In this talk, we discuss the application of hybrid models in addressing major challenges in unconventional reservoirs, including well performance evaluation, artificial lift life cycle management, performance insights, production optimization, well interference, and forecasting.
