This work presents a new multivariable controller for management of topside and bottomhole objectives during underbalanced drilling (UBD). A model predictive control (MPC) solution is used to control pressures, rate of penetration (ROP), and flow downhole while also ensuring that the topside processing constraints are respected.

With automated control, it is possible to reduce nonproductive time (NPT), improve safety, and operate closer to the process constraints. MPC is a good fit for UBD because of its easy inherent handling of multiple objectives and constraints. With good pressure control, it is in some cases also possible to reduce the number of casing strings.

The control solution is evaluated through simulations in a high-fidelity multiphase-flow oil and gas simulator (OLGA). It is shown that we can meet multiple objectives both at the surface and at different locations in the well. The optimization problem is solved with good results well within the given time constraints.

The used linear prediction models are relatively easy to understand and maintain. They are also fast and well-suited for optimization and predictions. However, the process is nonlinear, and the linear models will be less accurate as the process conditions change. Retuning or model adaptation might be required to obtain the desired performance. It is possible to include nonlinear models in the control framework, referred to as nonlinear MPC (NMPC), but this will add complexity and require more computational power.