Dynamic Self-Optimizing Control for Oil Reservoir Waterflooding

Waterflooding is a common oil recovery method where water is injected into the reservoir for increased productivity. Optimal operational strategy of waterflooding processes has to consider proceeds realized from produced oil and cost of productions including both injected and produced water. This is a dynamic optimization problem. The problem could be solved through numerical algorithms based on traditional optimal control theory which can provide only open-loop control solutions and rely on an accurate process model. However, reservoir properties are extremely uncertain, and hence open-loop solutions based on a nominal model are not suitable for applications with real reservoirs. Introduction of feedback into the optimization structure to counteract the effect of uncertainties has been proposed recently. In this work, a novel feedback optimization method for optimal waterflooding operation is presented. In the approach, appropriate controlled variables as combinations of measurement histories and manipulated variables are first derived through regression based on simulation data obtained from a nominal model. Then a feedback control law was represented as a linear function of measurement histories from the controlled variables obtained. Through a case study, it was shown that the feedback control solution proposed in this work was able to achieve a near-optimal operational profit with only 0.45% worse than that achieved through the true optimal control (with system's properties assumed to be known a priori), but 95.05% better than that obtained with the open-loop solution under uncertainties.