Multidisciplinary Data Integration for Artificial-Intelligence-Assisted Well Placement and Trajectory Design Optimization Under Uncertainty

Abstract

Abstract Infill well placement performed as part of field-development planning is traditionally performed by identifying areas of high remaining mobile hydrocarbons and good reservoir rock quality to be targeted. The identification of hotspots was also largely performed on single-model realizations and, therefore, not robust considering the reservoir characterization uncertainties. Increasing efforts were put into incorporating the uncertainties as a key element of the infill well placement workflow by computing probability maps to identify the hotspots with higher chances of success for infill production. The maps were still generated solely based on dynamic reservoir-simulation model results. In this paper we present an intelligent workflow that integrates the opportunity index probability maps concept derived exclusively from dynamic reservoir simulation models, with drilling risk maps derived from drilling data analysis and completions quality maps derived from geomechanical studies, and artificial-intelligence-driven reservoir target classification. The integration provides more depth in the hotspot selection by identifying the most productive and feasible locations for infill drilling. The locations are then used for well placement and trajectory design optimization. The well trajectories optimize factors in the hotspot locations, locations of existing drilling centers, surface topology for new drilling centers to be designed, numbers of available slots on each drilling center, and capital costs such as drilling economics and drilling center cost. Infill injection wells are placed in conjunction with the infill production wells either following a pattern-type of design or peripheral injection. The designed wells are evaluated via an automated pipeline using reservoir simulation where the set of wells will be tested against the ensemble of realizations under uncertainty. A probabilistic approach is taken for the infill well performance and the economics evaluation for candidate screening and selection for the field-development plan optimization. This approach provides higher confidence in the decision making through the early integration of drilling risks and geomechanics data, and provides a more robust assessment of the technical and economic performance of the proposed infill wells under uncertainty. The solution combines various concepts including opportunity index, advanced ML methods for target identification, as well as multidisciplinary integration for well target identification. Well trajectory design evaluation considering both production and injection wells and the evaluation of the performance of the proposed candidates under uncertainty in this context provides more robust results under uncertainty compared to widely used industry practices that lack integration and uncertainty considerations.