Basic Research Needs for Geosciences - Energetics Meetings and ...

GRAND CHALLENGE: SIMULATION OF MULTISCALE GEOLOGIC SYSTEMS FOR ULTRA-LONG TIMESproperties is perhaps the most important feature for causing different physical/chemicalphenomena to be important in different parts of the geologic system at different times. Aseamless multiscale framework would allow the scientific community to understand theemergence of patterns, trends and behaviors observable at larger scales that may arise fromstructures and processes at smaller scales.Many processes associated with multiphase flow and reactive transport occur over characteristiclength and time scales, while heterogeneity inherent in geologic systems occurs at all lengthscales. The fundamental challenge is to identify and map the scales at which connectionsbetween process and heterogeneity are found. Such a map would inform efforts to describemultiphase flow and chemically reactive transport in a seamless way and allow an organizeddescription of emergent behavior useful for developing the next generation of models applicableto the full range of relevant length scales.The scientific challenges are:• Accurate accounting of small-scale effects on the larger scales. The problem is complicatedby the fact that in many cases there may be no obvious scale separations. Challenges includethe appropriate mathematical and numerical representations at each scale and across scalesfor dynamic processes.• Development of stochastic mathematical and numerical frameworks that capture thenonlinear dynamic physical and chemical processes in geologic formations. The probabilisticdescription must account for uncertainty due to (1) incomplete knowledge of inaccessiblescales, (2) imperfectly understood physical phenomena, and (3) measurement errors.• Accurate representation of physics, including nonlinear interactions spanning a multiplicityof scales which are not amenable to scale-averaging. The complexity of the representation isamplified in systems pushed far from equilibrium.• Limited characterization of geologic structure on the field scale due to difficult, spatiallysparse and temporally infrequent sampling, and continually changing conditions.• Developing iterative feedback loops between model formulation and data acquisition.Significant computational challenges exist because the computational resources required to solvethese problems using Direct Numerical Simulation (DNS) at the required scale are not attainableeven with predicted increases in computing power. The computational challenges are:• The upscaled model typically includes fewer degrees of freedom at the expense of morecomplex (difficult to model) constitutive equations.• Upscaled models must be calibrated to ensure that fine scale influences are properlycaptured.• Accuracy estimates for the upscaled models need to be developed (Wallstrom et al. 1999).This will require running ensembles of statistically equivalent realizations of the model.• Scaleable linear and nonlinear solver algorithms for very large-scale computations must bedeveloped.Basic Research Needs for Geosciences: Facilitating 21 st Century Energy Systems 99