Seismic-scale finite element stress modeling of the subsurface. (December 2021)

• Record Type: Journal Article
• Authors: Fischer, Karsten; Pearce, Andrew; Garcia-Teijeiro, Xavier; Mallinson, Andrew; Lloyd, Ian; Anderson, Stephen; Gomez, Francisco; Kisra, Saad; Rodriguez-Herrera, Adrian
• Is Part Of: Geomechanics for energy and the environment. Volume 28(2021)
• Publication Date: 2021-12
• Subjects: Large-scale stress simulation -- Advances in numerical simulation -- 3D seismic to numerical​ simulation -- Billion element model -- High-performance computing; Engineering geology -- Periodicals; Power resources -- Periodicals; Energy development -- Technological innovations -- Periodicals; Engineering geology -- Environmental aspects -- Periodicals; Energy development -- Technological innovations; Engineering geology; Engineering geology -- Environmental aspects; Power resources; Geology -- Periodicals; Energy-Generating Resources -- Periodicals; Periodicals; Electronic journals; 621.042
• DOI: 10.1016/j.gete.2021.100245
• Languages: English
• ISSNs: 2352-3808

Abstract

Abstract: The stress field in rock formations of the subsurface is a key element to understand for the planning of underground operations in a safe and sustainable way. Such operations include the drilling of wells, hydrocarbon production, fluid injection for enhanced oil recovery and CO2 storage, hydraulic stimulation for unconventional resources, geothermal reservoirs, and many more. Modeling the subsurface stress field using 3D finite element (FE) simulations is common practice in the oil and gas and other industries. It allows to understand the prevailing in-situ stress and changes in terms of stress, deformation and rock failure that could result from planned operations. Such modeling is thus key to guide safe subsurface operations of many types and it is paramount that all available data and known complexity is captured in the highest possible resolution in these models. This results in a continuous quest for larger numerical simulations, which in turn demand software and linear solver technology to keep pace and ensure those do not become a limiting factor for computing the in-situ stress field. In this study, a FE simulator for 3D geomechanical modeling is optimized in multiple aspects to efficiently solve high resolution numerical stress simulations on large scales consisting of hundreds of millions of elements. This capability is demonstrated using a real-world dataset from the Gulf of Mexico to generate multiple geomechanical models from 500 million to over 1 billion elements. The models include fully heterogeneous material distributions and are built directly from seismic inversion data following a new direct modeling workflow that significantly facilitated pre and postprocessing. Using state-of-the-art high-performance compute (HPC) infrastructure, such models can be solved in less than 4 h with more than 3000 processors by applying hybrid parallelization methods. This demonstrates feasible turnaround times for such large stress simulations that enable geomechanical investigations in sufficient detail to improve the safety and sustainability of subsurface operations. Highlights: Advances in numerical simulation and HPC hardware for large-scale stress modeling. New efficient workflow to generate finite element simulation from 3D seismic data. Demonstration on case study example modeled with up to 1 billion elements. Enabling numerical geomechanics to guide safe and sustainable subsurface. operations.