Energy-efficient strategies are highly demanded in oil and gas fields to reduce production costs and carbon dioxide emissions. One way to increase efficiency is to improve the effectiveness of the process control. This project will develop and test advanced models for process control and optimization, based on model predictive control (MPC) and nonlinear model predictive control (NMPC), using machine learning and the so-called Bayesian Neural Network. 
    
We will study a complete production system including gas-lift, three-phase gravity separator, hydrocyclone, injection pump, and a booster pump. These sub-systems mutually interact in complex ways, and an optimal point for a single sub-system alone might result in high costs for another.

Model predictive control (MPC) and nonlinear model predictive control (NMPC) with different optimization strategies are the themes of this study. MPC uses linear system models, whereas NMPC uses nonlinear system models. We want to understand how nonlinear dynamics affect subsea production and how NMPC can improve the control system.