Anti-slug control design: Combining first principle modeling with a data-driven approach to obtain an easy-to-fit model-based control

Abstract

The limit cycle is an unexceptional problem in the oil industry that may cause significant losses in production. Also called slug flow or slugging, the unsteady flow can be handled by feedback control, although nonlinear issues must be considered. As an oil well production valve is opened, its transfer function gain tends to decrease until it reaches zero, meaning that the valve actions lose effect against the system backpressure. Notwithstanding, this sensitivity loss can be compensated by adapting a suitable tuning according to the well operating point. In this work, a methodology to generate this control policy is proposed based on combining first principle modeling with a data-driven approach. The method aims at improving closed-loop performance through a gain scheduling curve resulting from an easy-to-fit model to plant data. A systematic procedure is defined and validated through an actual deployment in a Petrobras ultra-deepwater oil rig. As a result, it was possible to suppress unsteady flow and increase oil production by more than 9%. Although the method has been validated in a satellite offshore well, one expects that feedback control can be used in different scenarios successfully, regardless of the slugging mechanism.