See underneath. High Performance Computing, Geophysics and Electromagnetic Methods

2017 edition

Octavio Castillo Reyes

Electromagnetic methods (EM) are an established tool in geophysics to “see underneath”, finding application in many areas such as hydrocarbon and mineral exploration, reservoir monitoring, CO2 storage characterization, geothermal reservoir imaging and many others. Particularly, the three dimensional marine controlled-source electromagnetic modelling (3D CSEM) is well established and widely used in industry and academy to define and characterize bodies by its electric resistivity, which help us to conduct exploration campaigns with a significant reduction of costs and risks.

On the other hand, the simulation and modelling tools help us to formalize and simplify the complexity we observe in nature. This simplification together to High Performance Computing (HPC) advances allow us to render natural phenomena treatable and testable. Although some contributions have been made to the development of algorithms and tools for 3D CSEM modelling, such as parallel codes developed by [1, 2, 3], the knowledge on this subject is still limited, with plenty of room for improvement. For instance, most codes that meet the requirements at real-scale modelling are commercial and often inaccessible to the wider scientific community, aspects that can all hamper advancements in the field.

To help reverse this trend, the main target of my PhD thesis is to implement an edge-based finite element Python code on unstructured meshes for 3D CSEM forward modelling in geophysics, namely, PETGEM: Parallel Edge-based Tool for Geophysical Electromagnetic Modelling (http://petgem.bsc.es). Because real scenarios cannot be handled in modest computational architectures, PETGEM was designed using an architecture-aware approach for HPC platforms. I have developed this new tool as open-source so that it can be used, modified and redistributed freely with the aims of fostering reproducibility and encouraging investigations about these fields.

PETGEM is based on the integration of Nédélec Edge Finite Element Method (EFEM), 3D CSEM and pure Python programming language, which is to our knowledge the first time this type of approach has been systematically applied for running simulations of this kind on HPC platforms. The efficacy of this tool has been demonstrated on different application scenarios of 3D CSEM models. In all cases the numerical solutions obtained with PETGEM were found in good agreement with reference models

[1] Alumbaugh, D. L., Newman, G. A., Prevost, L., and Shadid, J. N.: Three-dimensional wideband electromagnetic modeling on massively parallel computers, Radio Science, 31, 1–23, 1996.

[2] Key, K. and Weiss, C.: Adaptive finite-element modeling using unstructured grids: The 2D magnetotelluric example, Geophysics, 71, G291–G299, 2006.

[3] Koldan, J.: Numerical solution of 3-D electromagnetic problems in exploration geophysics and its implementation on massively parallel computers, Ph.D. thesis, Polytechnic University of Catalonia, 2013.