Robust MPC for optimal oil production under the presence of uncertainties

Abstract

Industries, business and private people often want to maximize income. The petroleum industry is no exception. Higher production leads to higher income. Model predictive control can help maximizing the production at oil field.

The objective is to design a robust multi-stage linear MPC for oil production optimization such that it is robust to uncertainties present in the system. The conservativeness and robustness of the developed robust MPC must be studied through detailed simulations.

To design a linear MPC it is necessary to have a linear model, which is derived of a nonlinear model of a gas lifted oil field. The robust multi-stage linear MPC is designed stepwise from a deterministic MPC to a robust MPC.

It is designed two variants of the robust multi-stage linear MPC. One with only hard constraint at wsmax and one with a soft constraint “wsmax – soft” below the wsmax. The first MPC works, but under some circumstances it is infeasible. Bounds can be violated, and the inputs can oscillate. Therefore, the soft constraint is introduced. Then the simulation is feasible at all times, and constraints and bounds are satisfied. Introducing the soft constraint makes the MPC slightly more conservative. The robustness is taken care of since the hard constraint wsmax is still operative.

The robust multi-stage linear MPC with an extra soft constraint below wsmax using the IPOPT solver is a robust and efficient MPC.