Techniques for calibrating borehole propagation model for direction drilling in real time

Abstract

Current disclosure proposes calibration methodology that uses the knowledge learned from historical data and real-time data to update a simplified physics-based borehole propagation model for real time steering control of a directional drilling system. The proposed methodology updates model parameters of the simplified physics-based model in real time through a process of setting up and optimizing an objective function, which indicates the difference between the estimated trajectory and the feedback from the real-time data. Penalizing terms, which prevent the parameters from deviating drastically from the prior knowledge of the parameter, and various components of the objective function are learned from the historical data. As a result, the updated model can accurately estimate and predict the system behavior allowing an input sequence of the system to be accurately mapped to the output response.