Managed Pressure Drilling System State Estimation Using The Multiple Model Adaptive Estimation Approach

This paper studies state estimation in the presence of parametric uncertainties, with flow estimation in Managed Pressure Drilling as a case study. Downhole measurements in most MPD systems have low frequency due to communication with mud-pulse telemetry. Also, the drilling process has parametric uncertainties due to unknown friction and fluid density and unmodeled actuator dynamics and noise add to the system complexity. This paper proposes a bank of estimators for simultaneous estimation of the model states. A Multiple-Model Adaptive Estimation (MMAE) algorithm is presented that encompasses a bank of Kalman Filters (KFs). Each KF is designed to handle a specific segment of the parametric uncertainty. In this algorithm, a probabilistically weighted combination of the local state estimations is used. Simulation results reveal that the proposed method can satisfactory estimate the unmeasured states in the presence of a wide range of uncertainties. Finally, the MMAE method is compared with a supervisory estimation algorithm to further clarify technical issues in the MPD system state estimation problem.