Proactive and Reactive Strategies for Optimal Operational Design: An Application in Smart Wells

Abstract

The primary objective of this study is to quantify the benefit of utilizing smart-well technology. The approach presented here maximizes the expected net present value (ENPV) of a field development strategy while accounting for uncertainty in reservoir parameters. In particular the dominant objective is to quantify the relative merits of adopting proactive and/or reactive strategies when designing optimal control policies from their impact on asset value. Proactive or reactive control strategies are the methods typically deployed for controlling smart wells in the presence of reservoir uncertainty. In the case of an entirely proactive control strategy, the future operational strategy is optimized based on one's current understanding of the reservoir; this approach does not account for new measurements, as and when they become available, and the improved understanding of the reservoir that they bring. In the case of an entirely reactive control, operational strategy follows rules that are tied to future measurements without recourse to a model-based prediction, e.g., the valve may be closed if significant water ingress is observed. In summary, the proactive strategy seeks to prevent an undesired future result, while the reactive strategy actuates the valves when the undesired event occurs. The approach presented in this article combines these two strategies. This hybrid approach was evaluated against a reservoir model based on the Namorado field, Campos basin, Brazil. The results are compared with scenarios involving a conventional well, but without smart completions, and an entirely proactive control strategy, that does not consider future information. We demonstrate that the approach yields the highest ENPV. The sensitivity of the approach to changes in the number of time steps used for both proactive and reactive control was also considered. For proactive control, the sensitivity analysis examined frequency of valve adjustment in the absence of new information and for reactive control, how fast one needs to react to new and revealing measurement data. This examination revealed that while being operationally proactive is important, there is no value in planning for frequent valve adjustments when there is an absence of new information.