Enhancing the Accuracy and Predictability of the Oxy Field Optimizer for Dynamic Steam Allocation in the Mukhaizna Steamflood Field

Abstract

The main challenge for the Mukhaizna steamflood field is to allocate steam dynamically throughout the entire field, which consists of more than 3,200 wells, to obtain the most attractive reservoir performance forecast. To address this challenge, Occidental has developed a state-of-the-art closed-loop optimization solution called the Oxy Field Optimizer (OFO). The aim of this study is to enhance the accuracy, robustness, and predictability of the OFO. Recent advances include connection design, simulation stability, history-matching workflow, model predictability (blind test), and the optimizer. To improve the proxy simulator, 2D connections between wells were introduced and various strategies to handle convergence issues were implemented. The history-matching workflow has been enhanced by automating the temperature match, multistep saturation tuning, and relative permeability tuning. The results show that the implementation of gridblock material balance check, well equation check, and Not a Number (NaN) value check after line search solved multiple convergence problems. The automated temperature match process is five times faster compared with the manual process, and the automated relative permeability tuning decreased average oil mismatch by 55%. The optimizer now utilizes a parallel implementation of a novel ensemble-based optimization scheme (EnOpt) algorithm, which is twice as fast as the original implementation. These proven advances make OFO an essential tool for obtaining optimal steam allocations.