Multiscale Heterogeneity and Solute Transport Model Parameter Uncertainty Study for a Fractured Low-Level Nuclear Waste Disposal Site in the Eastern United States

The objective of this research is to determine multiscale fluid and solute transport parameters in support of a site characterization effort at the Waste Area Grouping 5 (WAG 5) at ORNL (Oak Ridge National Laboratory) in eastern Tennessee, USA. The study site is located within the top 10 m of the subsurface, in which groundwater flow dynamics is influenced by both infiltration and recharge events. The soil and rock formations are macroporous and/or highly fractured at WAG 5. A natural gradient, multiple tracer injection experiment (bromide, helium and neon), was conducted to quantify the solute transport and mass transfer processes in the highly fractured shale bedrock. The field site is intensively instrumented with arrays of drive point and multi-level sampling wells. Field observations of hydraulic head and bromide solute movement dynamics are used in this study to calibrate a two-pore-domain, fracture-matrix flow and non-reactive solute transport model. We use a nested Latin hypercube (NLH) sampling technique to determine the near-optimal combinations of model parameters. As a result of the sampling technique, empirical probability distributions of model parameters are derived. Heterogeneity in two scales, field and matrix block, are quantified in terms of field scale distribution of hydraulic and solute transport properties and fracture spacing between matrix blocks. Uncertainties arising from tracer source density effect are also addressed through model prediction uncertainty analysis. It is also concluded that NLH is a relatively effective optimization technique, often capable of locating the near-optimal combination of model parameters in a few iterations.