Torque and drag analysis of a drill string using Sequential Monte Carlos methods

摘要

In this work we propose a methodology based on sequential Monte Carlos techniques for the estimation of dynamic systems states arising from the torque and drag phenomena of a directional drill string. This methodology arises from the need to consider the various mechanical factors and the geological parameters that influence the drift deviation, which conduce to no vertical drilling and phenomena of torque and drag. The anticipated model will acquire importance for the evaluation with the different efforts for which it will be in the next stage of the drill string. The methodology is exposed through the reconstruction of states of the synthetic model states proposed by Johancsik, Friesen, and Dawson, using the Sequential Monte Carlo algorithms, the unscented Kalman filter (UKF) and the ensemble Kalman filter (EnKF), observing good adjustment of torsion and drag states estimated by the filters with respect to the real values. Filters performance was evaluated in terms of the mean square error, showing variability of the estimated errors 0.05 for EnKF and 0.01 for UKF and a fast execution of the algorithms 3.07 (seg) for EnKF and 0.48 (seg) for UKF.