A Multisolution Optimization Framework for Well Placement and Control

Abstract

Summary Field development and control optimization aim to maximize the economic profit of oil and gas production. Mathematically, this results in a complex optimization problem with a large number of correlated decision (also known as control) variables of various types at different levels (e.g., at the level of well location variables or at the level of well production/injection control also know as well control or control settings variables) and a computationally expensive objective function (i.e., a reservoir simulation model). Current multilevel optimization frameworks provide only a single optimal scenario for a field development and control problem. However, unexpected problems that commonly arise during field development and operations can impose extra constraints, resulting in operators having to eventually select an adjusted, and most-likely, nonoptimal scenario. This work proposes a novel multisolution optimization framework, based on sequential optimization of control variables at multiple levels, providing the flexibility for operators to make optimal decisions while considering operational constraints. An ensemble of close-to-optimum solutions is selected from each level (e.g., from the well location optimization level) and transferred to the next level of optimization (e.g., to the control settings optimization), and this loop continues until no significant improvement is observed in the objective value. Fit-for-purpose clustering techniques are also developed to systematically select an ensemble of solutions, with maximum differences in well locations and control settings but small variation in the objective values, at each level of the optimization. The developed framework has been tested on two benchmark case studies. The results demonstrate high economic and operational efficiency of the developed multisolution framework as compared to the traditional approaches that rely on single-solution optimization. It is shown that suboptimal solutions from an early optimization level could approach the optimal solution at the next level(s), highlighting the value of the developed multisolution framework to deliver operational flexibility by a more efficient exploration of the search space.