Nonlinear Model Predictive Control of Gravity Separators
MetadataShow full item record
For economical reasons, it is necessary to operate separators such that the the purity specification is maximized while having the levels and the system pressure controlled. In this thesis, a nonlinear model predictive control is applied to a three phase subsea gravity separator system. The main modeling principle used to describe the oil droplets rising and settling of the droplets is given by Stokes law. Nonlinear dynamic equations were derived for the water level, total liquid level and for the pressure of the gas in the system based on the inflow and outflow dynamics, and were implemented in MATLAB as script files. The equations were nonlinear due to the geometry of the separator. These nonlinear models were formulated as a semi-explicit DAE system. The NMPC optimization problem used in this report is solved using direct collocation method using CasADi software within the MATLAB programming environment. Simulation studies have been carried out for four cases, namely setpoint tracking, disturbance rejection, sensitivity to measurement noise and the optimal water level investigation. The performance of the closed loop Nonlinear Model Predictive Control (NMPC) has been studied based on the simulation results. Simulation results were carried out to analyze the effect of measurement noise on the performance of the controller. Based on this analysis, the performance of the system for remaining cases was studied. The controller was analyzed for the setpoint tracking scenario where step change in water level was introduced to the system. Additionally, the performance of the controller to reject the disturbances was studied. Finally, the optimal level of water in the separator was investigated based on maximization of the total volume of oil entering its native phase.