Future of High Performance Computing in Integrated Petroleum Systems Modeling
Chevron ETC, Houston, TX.
Petroleum systems modeling (PSM) is an integration of different geological disciplines to analyze the formation and evolution of sedimentary basins and to study processes such as generation, migration, entrapment and preservation of hydrocarbons. PSM estimates mechanical and chemical compaction of sediments and the resulting porosity/permeability, computes pressure, estimates source rock maturity and the degree of kerogen transformation, computes multi-component hydrocarbon generation, expulsion and migration, provides likely locations where hydrocarbons are trapped, and estimates of accumulated hydrocarbon composition and volumes. In addition to its primary function which is to help reduce exploration risk related to hydrocarbon charge, PSM has become very useful in pressure and effective stress prediction which are utilized in reservoir and seal quality analysis and serve to improve velocities.
Computational complexity of PSM depends on the quality and resolution of seismic and well data, project maturity (exploration, development or production), availability of tectonic/structural/mechanical earth models, and availability of geochemical data. Typical models at present are not too large (several millions grid cells) and subsurface is represented by relatively simple structured meshes. The utilization of structured meshes often results in inadequately represented faults and other internal boundaries and may lead to incorrect hydrocarbon migration scenarios. The availability of high resolution seismic and well data allows for higher resolution and more complex models spanning from seismic to nano, hence allowing for more accurate representation of complex features and processes. This requires incorporation of unstructured/adaptive meshes and also the utilization of algorithms that couple poromechanics, basin modeling, seismic inversion and other disciplines, which requires high performance computing platforms, e.g., GPU-based clusters and optimized libraries for solving large, ill-conditioned, sparse matrices.
This paper presents the state-of-the-art in basin modeling and discusses recommended directions for next generation exploration for conventional and unconventional resources.
AAPG Search and Discovery Article #90155©2012 AAPG International Conference & Exhibition, Singapore, 16-19 September 2012