Computationally Intensive Research Project
Enabling Quantitatively Predictive Understanding of Multifluid Flow and Multicomponent Biogeochemical Reactive Transport in Complex, Subsurface Systems
MSC News
Additional Information
Steve B. Yabusaki1, Mark D. White1, Diana H. Bacon1, Vicky L. Freedman1, David R. Rector1, Joel M. Malard1, Mathew Rosing2, Peter C. Lichtner3, John C. Parker4, Eungyu Park4, Jin Ping Gwo5, Gregory P. Flach6, Karsten Pruess7, Carl I. Steefel8, Albert J. Valocchi9, Roelof J. Versteeg10, Alexandre Tartakovsky1, Mark L. Rockhold1, Yilin Fang1, Fan Zhang4, Mart Oostrom1
1Pacific Northwest National Laboratory, 2Independent Researcher, 3Los Alamos National Laboratory, 4Oak Ridge National Laboratory, 5University of Maryland, Baltimore County, 6Savannah River Technology Center, 7Lawrence Berkeley National Laboratory, 8Lawrence Livermore National Laboratory, 9University of Illinois, 10Idaho National Laboratory
FY07 Allocation - 1,000
Abstract
The activities described in this proposal are directed at the goal of a quantitatively mechanistic predictive understanding of multifluid flow and multicomponent biogeochemical reactive transport in complex subsurface systems. This capability is critical to the development of scientifically defensible decisions that serve DOE's long-term missions of energy security and the protection of human health and the environment. Massively parallel computing is used in conjunction with the problem-driven development of comprehensively detailed coupled process simulators to enable and accelerate new lines of research by increasing process and property detail, coupling more of these detailed processes, increasing spatial and temporal resolution and extent, linking more spatial and temporal scales, and testing larger ranges of modeling scenarios. Common themes from the large range of subsurface research applications addressed in this proposal are
- mechanistic process upscaling that provides a theoretical framework for linking information from a variety of length scales to predictions at the scale of interest,
- characterization of field-scale model parameters that accurately and robustly represent the effects of multiscale variations in material properties, and
- the development of improved simulation tools for studying subsurface processes in both the lab and the field.
This project uses several different subsurface modeling approaches developed for massively parallel computers to make progress on important subsurface science and engineering issues:
- pore-scale modeling of fluid flow and reactive transport in discrete pore spaces of vadose zone sediments;
- multiple interacting continua modeling to account for flow and transport behavior in distinct subsurface regions (e.g., fractures/matrix, mobile/immobile water, connected/disconnected pores);
- multiscale modeling linking ab initio molecular dynamics with pore-scale models to develop continuum scale representations of contaminant incorporation into secondary precipitates;
- multiphase modeling of nonaqueous-phase liquid (NAPL) migration in heterogeneous multidimensional subsurface materials where the chemical composition of the waste mixture can change the wetting conditions at the fluid-mineral interface;
- multifluid flow and multicomponent reactive transport with complex reaction networks involving aqueous complexation, mineral, sorption, and biologically-mediated reactions; and
- inverse modeling for the characterization of subsurface properties, flow field and geochemical status based on coupling of variably saturated hydrologic and geophysical models.
Underlying the mathematical modeling of the subsurface processes in this project is an equally strong commitment to continuously improving the algorithms, computational methods, and computer science that are the foundation for robust, accurate, efficient, portable, and scalable subsurface simulation software.
A research team that includes DOE national laboratories, universities, and private industry will perform the proposed activities in concert with support from projects that (will) provide data from the laboratory and the field, as well as site-specific characterization and monitoring information. Funding for these projects comes from EM programs at DOE waste sites; SC support of the ERSP and SciDAC programs; FE programs in oil shale, gas hydrates, and carbon sequestration; and LDRD. By combining to form a single proposal, the team will use MSCF's resources efficiently, share ideas, and collectively benefit from the development and incorporation of new process models, robust parallel solvers, and high performance parallel libraries. An important aspect of supporting these activities will be the testing and evaluation of parallel programming tools, debugging environments, and visualization software.
