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

PANEL REPORT: MODELING AND SIMULATION OF GEOLOGIC SYSTEMSMODELING AND SIMULATION OF GEOLOGIC SYSTEMSEverything should be made as simple as possible, but not simpler. – Albert EinsteinComputer modeling of subsurface systems has been practiced for over 50 years and in manyrespects is a mature technology. However, modeling capabilities fall short of application needs inareas such as nuclear waste disposal or carbon storage, where the long time scales needed forsafe containment and the subtlety of interacting processes place far greater demands than can besatisfied with present capabilities. There are opportunities to develop multiscale, multiphysics,nonlinear modeling capabilities that will take advantage of evolving petascale computingplatforms. These will provide qualitative improvement in the ability to simulate and understandsubsurface systems, resulting in large technical and economic benefits for design and operationof waste disposal facilities, improved environmental protection, and higher recovery rates ofenergy resources. To accomplish this requires an additional research objective to develop firstprinciples-basedapproaches to chemical speciation, thermodynamics, and reactivity in aqueousand nonaqueous systems.CURRENT STATUSComputer modeling of subsurface flow systems, such as groundwater aquifers and oil and gasreservoirs, has been performed at increasing levels of sophistication since the 1950s (Peacemanand Rachford 1955). Much is known about the physical and chemical processes in such systems.Equations have been formulated that relate process dynamics and equilibria to fundamentaldriving forces, such as fluid pressure and chemical potential (Aziz and Settari 1979; Firoozabadi1999). Algorithms have been developed for solving these equations and have been incorporatedinto sophisticated computer programs routinely used by engineering professionals in applicationsfor recovery of subsurface resources and environmental protection.In groundwater modeling, methods integrating modeling of complex systems with large anddiverse data sets can build on a substantial legacy, which started with the work of, for example,Cooley (1979) and Yeh and Yoon (1981). Hill and Tiedeman (2007) review more recent workthat addresses understanding and accounting for heterogeneity (e.g., Hyndman et al. 1994;Hyndman and Gorelick 1996; Kolterman and Gorelick 1996; Carle et al. 1998; Chen and Rubin2003; de Marsily et al. 2005), integrating many kinds of data (e.g., Rubin et al. 1992; McKennaand Poeter 1995; McLaughlin 2002; Binley and Beven 2003; Sanford et al. 2004; Hunt et al.2006), sensitivity analysis (e.g., Yager 1998; Saltelli et al. 2004; Barth and Hill 2005; Tiedemanet al. 2003, 2004), quantifying uncertainty (e.g., Moore and Doherty 2005; Meyer et al. 2004),and model discrimination (e.g., Poeter and Anderson 2005).Software currently available to address some of these issues includes, for example, PEST(Doherty 2005), UCODE_2005 (Poeter et al. 2005), iTOUGH2 (Finsterle 2004, 2006; Kowalskyet al. 2005), MMA (MultiModel Analysis; Poeter et al. 2007), OPR-PPR (Observation-PRediction and Parameter-PRediction statistics; Tonkin et al. 2006), and DAKOTA (at SandiaNational Laboratory). In addition, relevant environments for constructing such software arebecoming available. For example, the USGS and EPA recently produced JUPITER (Banta et al.2006), while the EPA is producing the COSU API.Basic Research Needs for Geosciences: Facilitating 21 st Century Energy Systems 49