Intelligent Waterflood Optimization Advisory System – A Step Change Towards Digital Transformation

Abstract

In this paper, we discuss the development of an end-to-end waterflood optimization solution that provides monitoring and surveillance dashboards with artificial intelligence (AI) and machine learning (ML) components to generate and assess insights into waterflood operational efficiency in an automated manner. The solution allows for fast screening of waterflood performance at diverse levels (reservoir, sector, pattern, well) enabling prompt identification of opportunities for immediate uptake into an opportunity management process and for evaluation in AI-driven production forecast solution and/or a reservoir simulator. The process starts with the integration of a wide range of production and reservoir engineering data types from multiple sources. Following this, a series of monitoring and surveillance dashboards of key units and elements of the entire waterflood operations are created. The workflows in these dashboards are framed with key waterflood reservoir and production engineering concepts in mind. The optimization opportunity insights are then extracted using automated traditional and AI/ML algorithms. The identified opportunities are consolidated in an optimization action list. This list is passed to an AI-driven production forecast solution and/or a reservoir simulator to assess the impact of each scenario. The system is designed to improve the business-time decision-making cycle, resulting in increased operational performance and lower waterflood operating costs by consolidating end-to-end optimization workflows in one platform. It incorporates both surface and subsurface aspects of the waterflood and provides a comprehensive understanding of waterflood operations from top-down field, reservoir, sector, pattern and well levels. Its AI/ML components facilitate understanding of producer-injector relationships, injector dynamic performance, underperformance of patterns in the sector as well as evaluating the impact of different optimization scenarios on incremental oil production. The data-driven production forecast component consists of several ML models and is tailored to assess their impact on oil production of different scenarios such as changes in voidage replacement ratio (VRR) in reservoir, sector, pattern and well levels. Opportunities are also converted into reservoir simulator compatible format in an automated manner to assess the impact of different scenarios using more rigorous numerical methods. The scenarios that yield the highest impact are passed to the field operations team for execution. The solution is expected to serve as a benchmark, upon successful implementation, for optimizing injection schemas in any field or reservoir. The novelty of the system lies in automating the insights generation process, in addition to integrating with an AI/ML production forecasting solution and/or a reservoir simulator to assess different optimization scenarios. It is an end-to-end solution for waterflood optimization because of the integration of various components that allow for the identification and assessment of opportunities all in one environment.