Mixed Integer Programming Optimizes Well Placement

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 220621, “Well Placement Optimization Using Mixed Integer Programming,” by Fouad F. Abouheit, SPE, Menhal A. Al-Ismael, SPE, and Raheel R. Baig, SPE, Saudi Aramco. The paper has not been peer reviewed. _ Placement and design of wells in high-potential regions within hydrocarbon reservoirs is a critical step in field development planning. However, identification of these regions does not address the challenges of maximizing ultimate hydrocarbon recovery. This identification must be complemented by optimizing wellbore placement and design. The complete paper presents an efficient mathematical optimization method for well placement that maximizes contact with the productive zones for the best locations in the reservoir. Problem Statement Accommodating multiple field development objectives while managing complexities poses a significant hurdle. Multiobjective functions have been designed to balance competing goals, such as maximizing hydrocarbon production, minimizing water injection, and optimizing net present value (NPV). Unfortunately, simulating numerous development scenarios involving large reservoir models, high well counts, and complex designs incurs substantial computational costs. As a result, relying exclusively on traditional optimization algorithms to place hundreds of wells in vast reservoir models becomes computationally prohibitive. Stochastic optimization techniques have been widely used in well-placement optimization. Hybrid approaches have been explored combining two or more optimization techniques to overcome limitations inherent to individual algorithms. These optimization approaches typically involve maximizing NPV as the objective function, subject to various constraints. The search space often spans the entire 3D grid of the simulation model, necessitating extensive computational resources and prolonged turnaround times. To mitigate these issues, some researchers reduced the search space by using sweet-spot maps. It is crucial to identify the sweet spots in an oil field for determining the most-promising locations for drilling new wells. In practice, sweet spots can be pinpointed by evaluating various reservoir attributes, including connectivity, sweep efficiency, productivity, and other key performance indicators. Within the context of well placement and design optimization, accurate sweet-spot identification plays a vital role, using inputs from 3D reservoir models. Specifically, each grid cell in the simulation model is assigned a value between 0 and 1 denoting its relative reservoir quality, with values closer to 1 indicating highly favorable conditions for well placement. A recent study demonstrated the effectiveness of using a reservoir opportunity index (ROI) to filter the 3D search space. Many other sweet-spot identifiers have significant value in optimizing well placement. The approach in this work streamlines the optimization process by applying ROI. This approach aims to alleviate the computational burden associated with traditional methods. In response to the computational challenges inherent in large-scale well-placement optimization, this work introduces a novel, efficient approach capable of handling full-field scale applications. By leveraging advanced computational mathematical modeling techniques, this methodology seeks to maximize the contact between wells and highly productive hydrocarbon zones, ultimately boosting overall field performance. The methodology uses mixed-integer programming (MIP) to formulate and solve the problem of complex optimization, facilitated by the power of high-performance optimization solvers. This synergy enables the rapid identification of optimal well placement and design, paving the way for enhanced hydrocarbon recovery and more-informed decision-making in field development.