Directional drilling capabilities continue pushing for a more challenging 3D wellbore path with higher dogleg severity. The ability of the planned BHA to drill the wellbore as per the planned trajectory determines the well's success. Drilling program critical decisions require an accurate prediction of BHA tendency. Inaccurate modeling results lead to failures in execution, such as the inability to deliver the planned trajectory, the failure to reach the target reservoir or the compromise of the well production result.

Physical modeling has been extensively used in well planning to simulate the drilling process, optimize the design, and identify potential risks. The drilling process, including BHA steering tendency, is complex, with multiple parameters involved. Physical modeling cannot accurately model all parameters and minimize various uncertainties. This limitation leads to a reduction in modeling result accuracy.

In recent years, artificial intelligence or machine learning has been the most popular technology to unlock the full potential of data across industries. We need novel methodologies to integrate traditional physics-based modeling with state-of-the-art machine learning techniques to solve complex science and engineering problems such as drilling simulations.

The physics-based approach of BHA tendency modeling is based on static and dynamic finite element analysis. This approach is often a good solution if the physics model is accurately calibrated. Yet, it requires deep domain knowledge and accurate formation, bit, and steering tools data and may incur a high computational cost.

The machine learning (ML) model is based on learning algorithms that find relationships between various parameters and actual tendency results in a training dataset. The limitation of the ML model is that it requires big historical data to get accurate modeling. Merging the two principles of physics and machine learning combines the best of the two worlds, resulting in higher accuracy, better scalability, and computational cost efficiency.