Nonlinear model predictive control of a Hammerstein Weiner model based experimental managed pressure drilling setup. (May 2019)

• Record Type: Journal Article
• Authors: Amin, Al; Imtiaz, Syed; Rahman, Aziz; Khan, Faisal
• Is Part Of: ISA transactions. Volume 88(2019)
• Page Start: 225
• Page End: 232
• Publication Date: 2019-05
• Subjects: Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2018.12.008
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: Design and implementation of a nonlinear model predictive controller (NMPC) on a pilot scale managed pressure drilling (MPD) setup is demonstrated. The goal of the controller is to maintain constant bottomhole pressure (BHP) and mitigate kick during reservoir influx scenario. Under normal condition, the controller tracks bottomhole pressure to a predefined setpoint. A Hammerstein Weiner nonlinear model has been used for pressure prediction, and genetic optimization algorithm for calculating optimal control input. During reservoir influx the controller switches to flow control mode to balance the pump flow and choke flow. After kick mitigation the controller switches back to pressure regulation mode by revising the setpoint pressure to estimated reservoir pressure. The NMPC controller delivered good performance over PI controller during normal operation, pump failure, and gas kick cases where flow demand changes frequently. Highlights: A Hammerstein Weiner nonlinear model based nonlinear model predictive controller (NMPC) has been developed for managed pressure drilling (MPD) system. A fault tolerant mode has been implemented to deal with the reservoir. The controller switches to flow control model to mitigate reservoir kick. Controller performance has been demonstrated in real-time on a lab-scale MPD system. Controller showed superior performance compared to a PI controller under various operational scenarios: normal operation, pump failure, and gas kick cases.