Abstract
Drilling fluids have a profound influence on drilling performance and well cost, both directly and indirectly through mud-related non-productive time. Despite the evident importance of optimal fluid performance, fluid treatment decisions are often based on superficial calculations or estimates using past experience by the mud engineer. Manual mixing of drilling fluid additives also poses significant environmental, health, and safety (EHS) risks. To help improve upon these issues, this paper introduces a model predictive control (MPC) algorithm for automated drilling fluid maintenance.
MPC is an advanced process control method based on predictive modeling of the system. Using such modeling, an optimization problem is solved to find the optimal input to move the system to the desired reference state. A methodology is proposed here for developing and applying hybrid composition-based models for MPC using machine learning techniques and analytical equations. Advantages of the proposed method are discussed, such as the ability to take advantage of estimated future formation types and rates of penetration for improved prediction of future fluid properties and the ability to implement lag time correction for various chemical additives.
The feasibility of this method was tested experimentally. A compositional model was developed for a simple two-polymer fluid using experimentally collected data. The lag time for chemical addition was also incorporated. Extensive laboratory experiments were conducted in a flow loop with a fully automated mud measurement and maintenance system, including a helical pipe viscometer, Coriolis meter, chemical screw feeders, and peristaltic pumps. A MPC algorithm was developed and tested to control and adjust viscosity at multiple shear rates by controlling the rates of addition of the two polymers. Tests were performed at various conditions; it was shown that the algorithm was able to achieve the desired control objectives.
Automated mud maintenance based on MPC is a promising method that will lead to more consistent and reliable drilling fluid properties. It takes advantage of formation knowledge and prediction to resolve issues inherent with simple PID control. Automated mud maintenance is aimed at eliminating the human error inherent with treatment decisions and manual mixing, leading to reduced fluid over-treatment/under- treatment events and reduced fluid-related non-productive time. Stable and consistent fluid properties will also enable better hydraulics control with improved ECD management in the field.
