Simple Model Representation of Underbalanced Drilling Hydraulics and Control

Abstract

Oil reservoirs can be sensitive to unfamiliar fluids such as drilling fluids. In order to prevent fluids to enter the formations surrounding the well, the well pressure can be lowered below the reservoir pressure, called underbalanced drilling. This technique calls for injecting of gas into the mud to lower the pressure, with additional inflow of oil and gas to the well as a result. This thesis evaluates current modelling of two-phase flow related to underbalanced drilling, and possible simplifications. Through investigation of simulation data, the current modelling based on first principles, is considered too complex compared to data of the key dynamics. To simplify the modelling, a black box system identification approach is used. By identifying solely from simulation data, low order models with good fit to validation data are obtained. Through testing, the models' validity are tested, and it is concluded that a number of simple models will be needed to represent the whole system given by the simulator. Models with specific orders are tested at different set points, and one model order produces good accuracy at all the operating points tested. This model is likely to be able to accurately represent the whole system with updated parameters.The accuracy of the simple models identified reveals simple system dynamics, and a PI controller is considered to be sufficient. The models are being used to tune the controller, and the controller perform great on set point changes. In addition, the controller is tested with an emulated connection, with satisfactory performance.