A generalized predictive control for management of an intelligent well's downhole, interval control valves - Design and practical implementation

This paper presents an advanced control method for online regulation of downhole Interval Control Valves (ICVs) to achieve optimal production via choke performance management. A Generalized Predictive Controller (GPC) has been shown to be capable of automatically controlling the area open to flow of multiple ICVs to achieve a specified production rate. A black box model was established using real-time, downhole, instrument data as a predictive model for the controller. The model parameters, were updated in real-time using the Decay Recursive Least Squares (DRLS) method. A case study in which a multi-zone horizontal intelligent well was located in complex reservoir showed that the GPC operation is highly effective. The robustness of the technique was illustrated by its ability to operate effectively in the complex reservoir environment when the signal is perturbed by outliers or by random noise levels up to the control error limits. The value of these control error limits must be increased as the step size of each valve operation becomes larger. Copyright 2009, Society of Petroleum Engineers.