Observer-Based Robust Model Predictive Control for a Multi-DOF Oilwell Drill String with Stick-Slip Oscillations

Abstract

In the petroleum industry, drilling plays an important role. One of the critical issues in the drilling process is human-machine interactions since a more automated method is desired to reduce process costs and improve the safety of operations. This research includes two main parts. The first central part analyzes sets of field data to show the effect of a Process Automation Control (PAC) on the drilling process, which in the results section indicates the usefulness of PAC systems. The second central part is designing a control system that suits the drill string dynamic system to increase process automation. The first part is straightforward, and it was published in 2018 (Farhangfar et al. 2018). The second part includes an observer design that can estimate downhole data, which is difficult and expensive to measure. A Robust Model Predictive Control (RMPC) system (in the finite horizon and infinite horizon forms) is presented, which can stabilize drilling performance by scheduling rotational speeds along the drill string. After designing the observer-based control system and proving that it is working on a simple dynamic system, it is applied to the drill string dynamic system. Its results are compared with the installed controller on the field.