CMG RESEARCH: Numerical and Experimental Validation of Stochastic Upscaling for Subsurface Contamination Problems Involving Multiphase Volatile Chlorinated Solvents

University Of Colorado At Denver-Downtown Campus, Denver CO

Investigators

Abstract

Nonlinear problems with multiple scales, heterogeneity, and uncertainty are pervasive in science and engineering. In this context, analytical theories are limited by unrealistic modeling assumptions, and extensive large-scale physical experiments are impractical. Computational modeling and experimentation, the "third methodology" of science, has a critical role to play in conjunction with theory and physical experiments. This project combines the three methodologies to study multiphase flow in porous media, specifically the remediation of groundwater contaminated by dense nonaqueous-phase liquids (DNAPLs). Field scales are much too large for small-scale effects to be directly representable in practical computations. Multiphase flow in heterogeneous subsurface formations involves local phenomena at material interfaces whose valid effective representation at larger scales is not understood. Uncertainty is central, because of the lack of complete or precise data, and because of the randomness of small-scale properties and flows when viewed from a larger scale. The project develops a nonlinear upscaling methodology based on a macro-scale stochastic theory of multiphase flow and transport that accounts for local phenomena at micro-scale material interfaces. The theory is applied numerically, so that limiting theoretical assumptions are local in space and time. The model is validated by multiple-scale heterogeneous laboratory experiments, using realistic chemicals and soil properties, that creates a "physical Monte Carlo" database for comparison. Environmental restoration is important to society. Remediation of contaminated groundwater often requires the study of the simultaneous flow of multiple non-mixing fluids through porous subsurface soils and rocks. These aquifers consist of many different materials and exhibit features such as layers and faults on a multitude of scales, all of which affect the fluid flow. In small-scale detail, these effects are uncertain in practical situations, because available data are insufficient to describe the subsurface precisely, and because these details are quite random from the standpoint of the larger field scale at which a hydrologist wants to analyze the appropriate strategy for a remediation project. This project develops a methodology that enables a scientist to describe the aquifer at the field scale, so that strategies can be formulated and tested in the setting where they will be applied. This "upscaling" methodology based on probabilistic concepts is an advance with wide implications beyond environmental restoration itself, important as it is, because similar issues arise throughout science and engineering. The project's combination of theory, computation, and physical experiments requires the close interdisciplinary collaboration of a mathematician and a geoscientist.