Ensemble-Based Proactive Optimization using a Reactive Strategy for ICV Control

Interval Control Valves (ICVs) are widely used to control fluid flow in producers and injectors. These devices help increase recovery throughout a field's lifespan by enabling effective drainage of reservoir zones. Suitable control methodologies are essential to ensure effective application of this technology in field cases. This paper introduces a workflow for assisting ICV operation based on a rule-based control criteria applied over an ensemble of models and how the optimal control targets can be found by putting these targets as control variables in ensemble optimization. The proposed methodology is developed to assist engineers in planning ICV operations using ensemble-based reservoir simulation. The inclusion of multiple realizations in the procedure ensures that a proper representation of uncertainty is considered in the development of the ICV operational plan. The methodology implements a rule-based control concept performing the following steps: (1) check whether the control criteria are satisfied, (2) modify the ICV setting, (3) advance the simulation a given small time step, followed by returning to (1) to check whether the criteria are satisfied. On top of this, a gradient-based optimization is performed that uses an ensemble-based stochastic approximation of the derivative. The controls of the optimization are the individual values of the target parameters used in step (1), and the objective of the optimization is to maximize hydrocarbon the ensemble mean oil recovery. The methodology is applied on a synthetic reservoir model with ten realizations. The case studied involves five producers and two water injectors. One ICV each is installed on four of the producers, while the fifth producer consists of two laterals with one ICV in each branch. The rule-based control ensures that each individual producer’s water-cut limit is fulfilled for every realization, while the optimization ensures the optimal values of the limits are found. This approach provides a way to efficiently operate ICVs given reservoir uncertainty. Two definitions for the water-cut limit are tested. Results for this simple case study yield 5% increase in mean oil recovery across the ensemble.